Tutorials

(It helps if you read the previous tutorial: http://theratutorial.tumblr.com/post/105461686188/digging-into-the-surface-materials)

So, I recently watched "The Mummy" and "The Mummy Returns" on an HD tv, and let's say that while these were certainly shameless B-movies to begin with, the special effects get even more so on a HD tv.

So, inspired, we're gonna do some ancient Egypt! Alright, so here's our fancy palace bedroom's initial render in Blender's Cycles. There's several issues with this render:

Badly lit(a common problem with environment+lamp lights).

Outside is a little boring

None of the objects do really look like part of a bed room.

That 'fire' looks terrible.

Doesn't look much like ancient Egypt.

Noisy(due to the way cycles works)

We're gonna solve the first two with a Matte Painting! Matte painting is officially a non-reflective painting used to create the illusion of distance. It has a long history starting with Theatre sets which had the sky and the cities painted on a backdrop behind all the set pieces. Later it became part of the early special effects in film: with nonreflective paint(matte), set builders and clver camera men managed to create the illusion of endless distant lands while the studio was hardly bigger than a sports field. As special effects progressed, Matte painting started to encompass other terms as well: think of scrolling backgrounds on a flying carpet. And it became digitized.

What we're going to do is replace the environment light with a plane with on that a picture. Unlike the traditional matte painting, we can use this picture to light up the scene as well!

We first place a plane in Blender, parallel to our camera's plane.

We then uv unwrap it, assign an image, and fill it with white. (I haven't done any blender tutorials, but there's plenty on the internet, so please look at those.)

Set the image with the UVs to an emission texture.

View via the camera and use texture painting to draw the horizon, and doodle some random compositionally interesting shapes.

Save the image out as an EXR.

Open it in Krita.

Now we're going to take the LUT docker, and enable OCIO. Download the ocio configs

here

while you read the following explanation. :)

OCIO, or Open Color IO, is a special colour management system that was designed for dealing with colours for VFX work. It's primary use is to quickly shift between different screen types(LUT), like linear RGB(mathematically correct, good for painting) to sRGB(like your monitor) or a raw look at the values.

In Krita, it's also required to paint in high dynamic range.

Unzip the archive you just downloaded, and select the nuke config file, set the input colorspace to linear, and the view to sRGB.

Select everything outside of the area that is painted on, and make it black:

Make a new layer, and fill the sky-portion with black.

We're now going to make another new layer, and fill the sky portion with an hdr yellow.

Here comes the confusing part: selecting that HDR yellow. To do so, you must lower the exposure slider on the LUT management. You can also use 'Y'+dragging your mouse on the canvas. Using this you have an infinite value range so the matte painting's light is more precise.

(Above, the lower the exposure, the higher the maximum color you can pick. All colours above 1.0 in a floating point space is a HDR color.)

Now use a transparency mask on the upper-layer, and draw a black and white gradient onto it, with the white portion on the lower half:

This will give 'proper' mixing.

Right-click the yellow layer, and 'flatten layer' to apply the transparency mask. Now merge it down.

Now use the colour picker to mix in colours:

We'll be using an analogues ramp for this, so the colours you want to mix in are from brightest to darkest:

Yellow

Orange

Red

Purple

Blue

I left out blue in the end:

I then cropped the layer, so I had a better idea of what I was doing.

Fill in the lower part, and refine the sky:

Draw in some dunes, I myself ended up with something like this before I stopped:

Assign the texture to the plane(make the uvs span the whole uv space if you cropped it), and do some renders.

Eventually, you should have given a strong feeling of depth in the painting, and that while we only had a few

When you're done, do a proper render, save that out as exr, and open that file in Krita.

Importing in Krita.

It's very dark now, but if we turn up exposure:

All the colour information is still there, including that of our matte-painting!

Now, we'll denoise it a bit(not necessary if you rendered it longer than I did)

Do this by first selecting the area you're going to denoise. Then duplicate the layer, rightclick the duplicated layer, and make a transparency mask.

Then apply a gaussian blur. (While G'MIC has many more smoothing and blurring filters, you shouldn't use it when working in floating point: G'MIC outputs 8bit colour information, which destroys our fidelity).

Now to polish the look, paint the last noise away and add in definition.

Use the crop tool to crop the layer afterwards, so it takes up less space.

Do this with all important elements.

For adjoining areas, use invert-selection and then intersect mode on the polygon select.

This allows you to have sharp adjoining areas:

Finally, you should have something like this:

(Note the gamma and exposure settings. We haven't touched the colour information with any brightness and contrast, but exposure and gamma allow us to quickly up these values so we may see the image more clearly. I also added it a brighter line where the walls meet, because light often gets trapped there.)

Now we can start drawing!

First, let's fix our fire:

Use the colourpicker(defaultly on CTRL, remember?) to select the HDR yellow.

Now, with very low opacity make some wiggles. Notice how they look more orangy? That's because we're working with hdr values in a linear space: The colour mixing maths is far more precise here.

The yellow will of course level out, so pick a lighter hdr yellow to paint the white bits below. Blur it with one of the smudge brushes to mix it in a bit and draw out the orangy colours.

Now we're going to do something about those walls.

Go into google image search and set the license to 'reuse'

With that, you'll be able to find images that can be reused legally in commercial work. Reuse with modification is also a good option, since we'll be modifying all images anyway. Select a few images you like.

Now we've pilfered Google Image Search we can import our images via layer->import/export->import layer.

Now transform one with perspective transform so that the straight lines match up with the straight lines in the drawing.

Then use the color to alpha filter(under filters->color->color to alpha) to remove the yellow standstone colour.

The lightsource of that image is however point in exactly the opposite direction.

Use invert (filters->adjustment->invert) to rectify this.

Make the transparency stronger by using the alpha curve in adjustment curves(I didn't take a screenshot of this, my apologies)

Then lock the layer-alpha(checkered squares next to the padlock icon), and colour pick some oranges to paint in.

Copy the layer and move it further to the next wall.

Fix the colours like the previous layer with colour picking and painting.

Set up drawing assistants to help drawing architectural lines:

Align the four helper dots of the vanishing point assistant to preexisting lines, so you can find the vanishing point quickly.

Turn on assistant drawing(in the freehand tool options) to draw nice straight architectural details.

Now for a special Krita trick. We're going to make the floor red. Every item in the picture is technically 'grey', so all we need to do is use 'multiply'. In the previous tutorial, we used 'hard light' to transfer all the tone values, but because we're working in a linear exr where we have a lot of fine lighting information, multiply is sufficient. Select the floor, set your blending mode to multiply.

Then use shift+backspace to fill the area with multiplied red.

Draw in some tiles with the vanishing point assistants to draw in tiles quickly.

Use multiply to determine yellow and blue. Then paint in yellow for the bed covers and blue for the sheets(We're painting the sheets on messify, because of story: the occupant of this room was in a hurry this morning and didn't tidy his bed):

This is what our image looks like right now:

Keep painting in architectural details as well as curtains(I couldn't find referenced of how they would look, but I am pretty certain they had them, so I drew the simplest curtains I could come up with).

I wanted to add a curious animal walking in, so I searched for a proper image of a Gray Heron(as that is the animal that looks the most like the extinct egyptian bennu heron)

There's again, a few big problems.

The grass needs to be removed.

The image is way to big.

It's in the wrong colours.

For the second, we first add a transform mask:

Then we select the mask, and click the picture on canvas with the transform tool, and we resize it.

Press enter to apply.

If you did this correct, making the transform mask invisible will allow you to undo the transform dynamically. So we can remove the background from the full resolution item, while also having the transform ready.

Let's remove that background now we're add it.

So, hide the layer, and press alt-click on the layer on which the picture is.

You have now isolated it.

If this were GIMP you'd use object extraction, but Krita doesn't have that: It's a painting program and not a photo manipulation program like the GIMP.

Instead, use the path-select tool to select a rough outline.

Invert the select and press 'delete'

Use the freehand brush in eraser mode('e') to remove the last details:

Select any other layer to get out of isolation mode. Unhide the transform mask to see that transform again.

Much better. Now we need to do something about those colours.

Put the layer into a group, and in the same group, make another layer, set it to 'multiply'. Also set the 'a' for alpha inheritance. Then you don't need to do the selection stuff I did below.

Fill it with a orange.

Paint with a darker colour to indicate shadows.

Color pick from behind and in front to define the shadows. Use ctrl+alt+click to select colours on the current layer only.

The bird looks pretty awkward being so sharp in this blurry image. You could paint it over, but this is outside the current scope.

We repeat this trick with a vase:

To solidify it, add a layer below the vase with some soft black for shading.

As a final touch, I paint in the hieroglyphs. The ancient Egyptians loved decorating their reliefs and hieroglyphs with paint.

This is largely done by hand, but we can use the contiguous select(the wand icon), to select the areas outside the glyphs.

Use 'shrink' selection to create a bit more room around it.

Do some hand-painting to polish it.

You can keep on adding things, but this is where I end it:

Now it's finished, we can continue on to reuse it in Blender as a matte painting again:

Or we try to convert it to sRGB...

Getting out of the Linear workflow and ready for web.

Our image has been, in all this, a 16bit float image.

There's a few disadvantages to that:

Not all image viewers can view these types of images.

It's VERY heavy compared to 8bit images.

There's often no need to have 16bit float images on the web, just like you rarely put 300dpi images on the internet.

Now, you can flatten(important) the image, and convert it to an 8bit sRGB profile(turn off optimisations, we're going from Linear to sRGB after all)

But then we don't have the changed contrast and brightness like the LUT docker allows us to.

Krita 2.9 doesn't have functionality to do this yet.

However Natron can.

Natron is a compositing program, which means it is used to do complex color correction and transformations for film, but also for single images. It does this using nodes:

Nodes are somewhat like layers, but instead we look at filter layers, transform masks, transparency masks and images like little boxes of processes we can connect to each other. You may be familiar with nodes if you used Blender before.

To the left are the possible nodes, in the middle, at top the preview and at the bottom the node editor. There's only one node right now: of the previewer above.

To the right, we have the configuration and properties of each node.

Use the upper left 'image' button to 'read' an image. Select our exr file.

The image should appear as below, going into the previewer. If you look at the configuration, the input and output are both set to 'linear' this is good.

Rightclick the image node, and go into color->math->gamma.

You should now have a blue block with Gamma on it. This is our filter.

In the Gamma window to the right, turn up the value slider till the image looks nice to you.

Go to Render->Render all writers, and save you image.

If it doesn't render immidiately, make sure you attach the yellow block below to the blue gamme-filter. (drag the black arrow going into the yellow block to attach it to the blue block)

Now press render.

Eventually, Natron will tell you this:

And the end result:

Conclusion and Afterthoughts:

It's very interesting to be able to paint in Linear space, and if your computer can handle it(16bit float images take up twice as much ram as 8bit images), I'd recommend playing a lot with it.

Working in a linear method allowed us to paint our own Matte Painting for our 3d render.

But working in Linear is also significantly simpler in how your paint fire or change the light(notice we only used 'multiply').

Basic Theory.

So, you may know that perspective brings depth to an illustration. You know that things seem to get smaller the further away they are, so an adult man can be the size of your hand as long as the distance is right.

That you see this happening in real life has actually to do with the lens of our eyes:

Due to the images coming in as rays and then projected onto our retina(here depicted with a flat place, but of course, in real life dome shaped), images closer by get more real estate on the retina.

As such, we can see large objects becoming smaller the further away they get. (Curvature of the eath also has a little to do with this, but not as significantly as our lens does)

(Fresco from the Villa of Publius Fannius Synistor, image from http://foundinantiquity.com)

(Priests of Anubis, the guide of the dead and the god of tombs and embalming, perform the opening of the mouth ritual, from wikipedia.org . It is not that there were such tiny people living in Egypt as much as that they just weren't as important as the scene in the middle)

While some early artists tried to grasp this illusion, others, who painted more in symbols decided that instead of bothering to figure out the illusion of depth, they instead would mark the most important things as large, and less important things as tiny. This type of perspective is more of a compositional choice and we'll get back on it in a composition tutorial.

Around the renaissance artist started to slowly make sense of this perspective thing. They had yet to understand the cause of perspective deformation, but they were getting consistent results, with figuring out the rules of linear perspective.

Linear perspective.

We call Linear perspective linear, because you can take a ruler to it and see everything deform based off straight lines.

This is not like the real world, where perspective is actually curved(due to the shape of the retina), and the deformation becomes worse the closer you are to the edge of your vision.

(Image taken with a fish-eye lens… this image is actually closer to our vision than regular photos. image from wikimedia.org )

We use linear perspective a lot in drawing because while not fully accurate, it does provide the illusion of depth sufficiently and is easy to draw compared to fish-eye lens.

Traditionally, many art tutorials would go through the concepts of one-point, two-point and three-point perspective:

While these are nice basic ways of handling perspective, they are not fundamentally true, and you may come across the problem of having slightly rotated objects in a one-point perspective for example.

We can make a square, and subdivide the square, to draw a rotated square based off that.

And there is where it breaks apart. The rotated square has it's own set of vanishing points. So, in that manner, there's not really such a thing as a one-point perspective, or two point perspective, or even three point perspective strictly. They are however, a good thing to start from.

However, always remember that any set of parallel lines meet at their personal vanishing point in a drawing using perspective.

Using this, you can combine perspectives and create dynamic drawings to your fancy.

Theory, part two

But, for this tutorial, we're also going to focus a little on making a good background. Surprisingly enough, to make a good background is not just a matter of perspective.

It's a matter of environmental storytelling.

Environmental storytelling is mostly heard in the context of interactive environments, like theme parks and video games. The idea is that no environment was created purposelessly: Houses were build for people to live in, natural reserves were shaped by nature, your room is filled with the things you like.

So each environment has a little story embedded in itself of how it became that way. And this is just as important for background artists as it is for any other environment artist.

For example, a mining town would be settled near a mine. The people living there are usually mine workers, so the buildings were made quickly, are made of local materials, and are fairly spartan in appearance. There will, or will not be much debris from the mine. There will be a good supply route towards the village, and there won't be many trees close by because these have been either cleared or cut-down for resources. There will be a small amount of livestock, like chickens and perhaps a cow to provide for some daily produce. A baker will be present because flour is much easier to transport than baked goods. A smithy will be present to restore damaged mining equipment.

There will be many transport items as well as mining equipment lying around.

Depending on natural circumstances, the village will be mostly a base for the miners to work, or where they raise their children. If it's too cold during winter, or the mine floods, it's more advantageous for the miners to go home to their families during that time period. If they can work all year round, the miners may settle there with their families.

A trading town on the other hand, is lying at a crossing of multiple trading routes. These can be roads, rivers or open sea. The houses will be made from local resources(like wood and clay), or traded resources. As one can become really wealthy with trade, some of the houses might be really opulent. There will be a market place. The main roads will be wide and paved, to accommodate traffic.

There will, aside from many shops, be basic necessities, like baker, grocer, butcher and smith. Next to that there might also be many food shops or cafes. There might be small industries, like textile, or pottery, taking advantage of the raw resources being transported into town, and their product being transported out of town by merchants.

The town will may get a wall or some defence, the wealth attracting criminals and burglars. Trading is a year round job, and there will be whole families living in the town, children playing in the streets when not doing chores for their parents.

Getting a good idea of how either of these places look? That makes it much easier to start drawing them, right? This requires a bit of knowledge on geography, ecology and architecture, but you'll find you know more than you think if you ask yourself questions about how people would live there.

Practical.

Sketching

This first part is software agnostic, meaning you can do it with any software you like. Or even traditionally, if you'd want.

But for this part I'll be using MyPaint, because of the infinite canvas it has.

The first thing is that you have a story in mind, and that you start planning the composition.

So, going to our mining town, what are the initial stories we could depict?

-People leaving the mine during sunset?

-Traders coming to the mine?

-Children playing outside during the day?

-Workers getting drunk at the bar?

I'll go for the first one: people leaving the mine during sunset.

This means we'll need to depict:

*The mine. This can be a hole in the mountain, or a lift.

*The miners.

**The miners might be in a good mood because the day has ended.

**The overseer counting or checking out the men.

**Equipment being checked.

***Lanterns

***Pickaxes

***Ropes

***Helmets

**Women and children greeting their husbands.

**Canaries.

*Minecarts.

*Debris.

*Boxes & barrels

*Houses.

**Dinner is being cooked.

*Nature

**Mountains or hills.

**If it's a coal mine, hills.

**Birds.

**Plantlife

***Grass

***Flowers

***Herbs

*Clouds

*sunsetting.(strong shadows)

Let's start with determining the location of the horizon.

To easily draw a straight line in MyPaint, use shift and drag the cursor over the canvas as you would with drawing a normal line. To make it snap to degrees of 15°, press in ctrl as well.

For Gimp, you can use shift as well.

For Krita, the default stickykey is v. Then to snap to an angle, press shift AFTER your start your line.

We decide where the mine is going to be.

We decide what way the rest of the village will lie.

In between both, the miners will walk.

We draw some hills and mountains.

We decide where the supervisor is standing, and where the supplies will go.

Now we try to figure out how balanced out composition is. As you can see there's quite a few white spots.

Let's fill up the lower white spots with some boxes and minecarts, while we fill up the upper spots with plants.

Now we have our rough composition. We'll now try to determine the vanishing points.

If you had used some perspective before, you can try and draw straight lines to the horizon. As you can see, this does not always result in proper vanishing points. But that's alright, that's why we're sketching.

If you had not worked in some perspective, it's time to determine the vanishing points.

In this case, we have a very central composition with the mine in the middle. Therefore, we put our central vanishing point at the middle of the mine.

Now, here comes the tricky bit that not everyone realises:

Only objects which are positioned at the same rotation(and thus whose length and width lines are parallel to each other), share the same vanishing point(s).

This means, that a single composition can, and likely will, contain multiple vanishing points. As long as all these vanishing points share the same horizon, these objects will look consistent with each other.

It becomes a little more complex with three-point perspective, where the third vanishing point refers to the height. But with that, you'll have to consider the canvas as a 3d coordinate system.

I'll get to the why of this once we start doing perspective projection.

So, let's return to our drawing.

We could have our drawing look like this, with all the lines using the same vanishing point, but it looks very boring, doesn't it? Even though we manage to tell a story of sorts, the story feels artificial. The mine-cart, for example, why was it put precisely parallel to the houses? Wouldn't it make more sense to put it aside, slightly angled?

That's why we're going to rotate some things slightly. So we'll put in a couple of two-point perspectives in there!

You determine the lines by drawing the width line and the depth line to the horizon. Where it crosses the horizon is your vanishing point for that line and all lines parallel to it. You should take the time to correct the angles of your sketch here with correct fitting to the horizon type of vanishing points.

Create vanishing points for all the important rectangular objects. We'll get to other objects in a bit.

Watch the angle of where the perspective lines meet in the object. There's specific rules for this in perspective deformation, but for now, remember that.

1. The further away from the horizon the more angular an object.

2. The closer by to the camera, the further away are the individual vanishing points.

Now we have our basic vanishing points, we're going to add an image border.

In MyPaint this is a function of the program to deal with infinite canvas. Set the canvas to the layer size, and the hover the mouse while in the canvas menu over the border-edges to drag them and make it smaller.

Drag it around till you find something interesting.

In Krita you just make a new layer and add a rectangle.

In Gimp, you'll have to make a new layer, make a rectangular selection, fill it with black, go to selection->shrink selection. Shrink for a few pixels, and then press delete.

Now you have your composition and vanishing points, and we're in the second phase of sketching.

Now, start by defining the foreground of your picture. Like, start with the minecart up front. Look up a minecarts if you have to, and pay attention to the logic of each element:

How do the wheels attach? How is it fortified? How do the brakes work?

Looking at wikimedia commons, you can see that all minecarts feel very stocky, and the wheels are close-together and look heavy. Furthermore, the walls of minecarts taper inwards near the bottom, likely to strengthen the design of a container for heavy rocks. We need to replicate this design to make sure it resembles a minecart.

First, we're making it look deeper. We'll use our vanishing points to find the angle of the lower parts.

Then, we subdivide the surface. You do this by drawing a line from each corner to the opposite. That way, you can find the middle point of any square or rectangle, and thus subdivide the surface. Do this twice on both sides.

You can use subdivision to make the minecart taper inwards. Using subdivision allows you to make sure that the diagonal line stays in perspective.

We continue detailing the minecart and hide the perspective lines layer.

For wheels, try to figure out where a square that would encompass the wheel would be, and use that as your basis.

Then, in mypaint 1.1 you can make use of the geometric tools. In Krita you can use the assistants. Gimp has no tool for drawing this easily.

(To be honest though, this is such a complex thing to do, that I usually just guess it. I have no patience ;) )

We add the last set of details, like how it was constructed, where the bolts go, a dent.

And then we clean away the vanishing lines.

Continue this with all foreground elements.

Around this part I realised that I was missing a water-barrel.

Of course, the story rears it's head again. This water barrel collects water from the roof as it rains, for cheap washing water… and of course, that rainpipe is connected to a house…

Next up…

the shed.

Using the subdivision method, we find the proper-top of the roof.

We define the skeleton…

Put in boxes, and remove unnecessary construction lines.

We then attempt to use the subdivision method to create the planks making up the roof.

Again, the story. Add some extra fortification to the structure of the wooden shed:

We clean up and this is where we are now:

Using the same techniques we add the final house…

It's obviously owned by the Johnsons… And they took a round window along when moving in here.

Then finally, the mine.

We can work far more organically here due to the nature of rocks.

And ofcourse, the nature of weathered wood.

I actually don't recommend using the straight line tool too often.

Getting out our original sketch again:

Let's make rocks! Use the original sketch as a guide.

Again, the keyword here is story:

If you draw mountains, they are not straight lines diagonal, because they have been weathered over the years. So add a little bit of squigly to those lines. Similarly, if your drawing has a lot of sky, give it some clouds.

Use the subdivision trick to determine how to segment a rail coming from the mine. (After all, how would those minecarts get out. (I briefly forgot that rails are hammered over the wooden beams. I added them later, but this is why reference is important :) )

And that's the last of the technical objects that needed to be drawn.

Which means it's time for people… and their stories.

The supervisor doesn't want to stand up all the time when he's checking out everyone. So I gave him a chair.

Some people checking out…

A guy comforting his friend up front.

And some guys walking into the village.

Then finally, we add the last details. Some texture and grass popping out between things:

Some trees and plants

And a personal favourite:

Birds in the sky.

And we're done with the sketch.

Doing this in Krita

So, as you may have noticed, it took me a while to create this tutorial.

This is mostly because it was very exhausting to do this with current digital methods.

So... I kind of started coding in Krita to create my ideal perspective assisting tools, which consists of an assistant that functions as vanishing point, and one that gives you parallel lines.

The demonstration of said feature is here:

(link: https://www.youtube.com/watch?v=OhEv2pw3EuI)

And, I developed it specifically so that you would not necessarily have to use snapping. This allows you to freehand lines, and especially, make little dents and scratches more easily.

We'll be using Krita for inking now. (You do not have to ink it, but I have some wise advice to impart on inking backgrounds.)

So, you set them up, by aligning the knots to existing perspective lines.

In the case of black and white linework, let me give you a tiny optimisation tip: Set the layer to the grayscale colourspace. This'll make the layer take up half the required memory, as grayscale layer only need two channels to register the colour of a pixel in, while colour laers need four. Do this via layer->change layer colour space. Not image->change image colourspace… I did the latter, and it'll become obvious why later.

Use a thick brush to draw the silhouette lines. Avoid using the straight line tool if you can. It can feel artificial.

Then use a thinner brush to draw the details and angles facing toward you. The latter, try not to make them continuous. The reason for this is that this can segment the shape and make it too complex. Suggestive, broken lines make this less of an issue.

Here's a few examples of the same box, first inked with all lines the same thickness, then with the silhouette lines defined more, then with broken lines for edges, with the last fully hand drawn. I am more a fan of the lower two. The second is sorta acceptable, but avoiding the first is a good habit.

I also underlaid the inked part with a white silhouette layer. This helps me see the drawing more clearly, and make inking parts that are further into the distance with more ease.(The lines won't overlap into the forground parts, if I put their inking on a layer below the white :) ) This is just done with regular filling methods.

I kinda half-assed these gentlemen because I did not sketch them out properly… let this be a lesson. I use a thicker brush than I would normally...

… this in conjunction with using thinner lines the further away, results in the two miners at the front becoming the subject of the drawing.

To simplify getting the details on the ground to decrease in line-weight the further away they go, use a perspective grid assistant. If you set the size to the 'perspective' sensor, it'll decrease the closer it gets to the grid's vanishing point.

This only works on flat planes, so the railroad track and the ground, not so much on things like the wagon.

I also took the oppertunity to improve the logic of some elements, such as the construction of the shed and the attachment of the pipe in the foreground. I however missed others like the relativesize of the men to the mine-opening itself. Always take the chance to improve your drawing with each iteration.

We then proceed to use g'mic recolorize[interactive].

There's a little bit of athmospheric perspective going on here, but not much. The mountains become more 'sky-coloured' than the foreground. Unfortunatly, I didn't have time to really dive into the matter of athmospheric perspective this time around :/

Of course, because I set my whole image to grayscale instead of only the layer… I ended up with a greyscale output… so set only the inking layer to grayscale :)

Fixing the grayscale (with recolouring), and adding some quick shadows via the hard-light method, and I'm done.

Notice how though the drawing has a lot of detail, the ones that have the thickest linework get more attention, and you are able to identify different elements quickly.

This could of course be better if I spended another hour or two on it, but I want to finish this tutorial :)

Conclusion and afterthoughts:

Perspective is… time consuming, but it's one of the most rewarding draughting techniques to master(It's far easier than figure drawing). I hope that I also expressed the mind-set required to draw a good background. The latter is especially important, because taking the time to tell these little mini stories can give a lot more enjoyment to the viewer, who can return to a drawing time and time again to find it present them a new story.

I have to admit, I always tend to be a little more fluid about drawing perspective details, I don't always draw all vanishing points for example, but rather interpolate the line. Either way, I do enjoy perspective-drawing a lot, and I hope you will too!

Next time: The second krita kickstarter tutorial.

Sources:

on multiple horizons and fables of traditional perspective:

http://slgallant.com/tag/horizon-lines/

For this tutorial, I will be using Blender's raytracer 'Cycles' to illustrate some basic lighting concepts. A raytracer is a type of 3d renderer, a method to visualise 3d models on your screen, which uses a model very similar to the real-life model of light, where photons come from a lightsource, and bounce off objects, with some ending up in your eye, so you end up seeing those objects. Raytracers, unlike real-life, do a reverse, and start from the pixel on the screen and move towards the light sources, to determine what the colour of the pixel on the screen is.

It's not extremely important to understand how it really works, but it's definitely fun to play with. Blender, also an open source program, can be downloaded for free here:

http://www.blender.org/

As well, the demo image for this time around has been modeled and rendered with blender:

The idea being you can focus 100% on the material and not worry about the shape.

This tutorial will also be done with Krita 2.9, as it's a reward for the Kickstarter for that program. You can get the first beta for Krita 2.9(or later versions when the time comes) from krita.org. Now, onto the tutorial.

Review of previous theory.

So, we have…

Self shadow. Self shadow happens due to the light only bounding off certain sides of an object, and not being able to reach others.

Cast shadow. This is the shadow created by another object blocking the light.

Indirect light. This is the light that is bouncing off other objects, and thus appearing in areas where the direct light does not reach.

Highlights. These are a type of reflection. We'll go into these a little further in this tutorial.

Overview of basic materials:

In this tutorial, I will make use of a method that separates the shadow and light from the colours. This is important because when it comes to materials, the use of your shadows is 80% of the work, so you should spend extra time on the shadows for this.

Diffuse materials:

The traditional diffuse material is chalk, it's called diffuse because it spreads out the light into all directions.

Diffuse materials are pretty easy to draw: we only require a self-shadow, cast shadow and indirect light to make it realistic.

I've used the g'mic recolorize filter to quickly colour this in.

Then, I split those up, though this step is not mandatory:

The split layer function is underneath layers->split layer(at the bottom).

Remove/Hide the background, we won't need it:

Now we make a new layer called shadows. We fill it with 50% grey:

and then we set said layer to hard light(you can find this under catagory 'light'.

Now, if you set the shadow layer to multiply, or to Hard Light. Multiply means that it will multiply the numeric values of the colours, always resulting in a darker colour.

Hard Light, on the other hand, will have 50% grey as affecting nothing. Everything below that set to multiply(so black results in black), and everything above that set to Screen(the inverse of multiply). This means that lighter colours will lighten the overal colour, and darker colours will darken the overal colour.

Finally, use inherit alpha.

Inherit alpha takes the alpha of lowest layer in the stack, and uses this as a transprency mask. This is not unlike Photoshop's clipping mask functionality, with the difference being that clipping masks uses the first not-clipping mask layer as it's transparency mask, and inherit alpha uses the lowest in the stack.

For our object, we'll set down the basic shadows. For the colour, we'll use a 30%-ish grey, so we can still see a little.

Use the fill tool with 1px growth to simplify this process:

Now, we're going to add subtle high-lights with a similar technique.

Select the areas that are directly in the light, and go over them with white.  If you used multiply, this can have a strange effect, hence why I preffer using hard light.

Now we're pretty close to our diffuse material, but the sharpness makes it look a little cartoony.

So what we're gonna do first is the indirect light. Find all the areas affected by indirect light, and use an airbrush to colour them a little with indirect light colour. Here I use blue, as usually the sky is a secondary light to an object, and yellow for the watertank as indirect light.

Without lines, it still looks a little weird, so we're going to add in some spots where the light definitely can't reach. These are creases and the like.Go over with a small black brush to get this right.

Now, while this is a perfect diffuse material, it still feels unnatural:

It's missing story.

First off, we need to consider where the material came from. In case of Chalk or Limestone, it would be coming from a quarry. Limestone or chalk get shaped by careful chiseling, breaking away the unneeded stone, and then sanded into smoothness.

This means we cannot have this perfectly flat shading, so we should add a little bit of roughness to it.

Use the block-basic brush to gloss away the hard-edges. Krita allows you to pick the colour of a single layer, which is something you should definitely do here. The default hotkey is shift+ctrl+left-mouse click, but you can change it in settings->configure krita->canvas input settings.

Doing this picking from one layer keeps out shadow layer seperate from the colours:

After you're done with that, it should be quite far ahead:

Then, let us play with textures:

Specifically, chalk is very uninteresting, but Limestone has a lot of impurities from fossilized sea life, giving a nice texture to it from time to time. You can draw these yourself, or look up an image on the internet.

Now, the nice thing about us separating the shading from the colour of the material itself through using hard light, we can now paste or draw this texture underneath the hard-light-layer. This means that the shading will apply itself correctly.

We're not putting it in the colour group, instead put it in a group layer above it.

And now for a secret Krita trick: Go into the layer properties for the group you put the texture in, and disable the alpha.

It will act like Inherit Alpha!

Use the layers in the colour group to quickly select a specific color. (Right-click the layer->select opaque)  Then select the texture-layer, and right-click->add transparency mask. The selection should now be used as transparency mask.

Use the perspective transform to correctly deform the texture, make sure to turn off the transparency mask so you don't deform it!

Apply this to all the surfaces of the figure. For the circular parts, like the wheel, use the cage tool.

For genuinely round parts, either use liquify or warp.

When using warp or the cage tool, you can do ctrl+click to select multiple nodes at once! Then use move(within the selection), rotate(outside the selection) or scale(ctrl+outside the selection) to transform them.

Play with the settings, such as going for affine instead of rigid. This will change the way how warp in particular deforms the texture.

Create a transparency mask, and done:

At this point I realised the creases were not defined enough, so I used select-opaque on the lineart:

(I also set the layer-blend mode from overlay to normal, it looks better)

Then I painted over the dark creases with black, and the creases that would catch light with white, and deselected:

I then added new textures for the rest of the objects:

The highlights here accidentally had some colour in them, so I desaturated them by selecting grey and using the saturation blending mode:

Finally, we'll paint in little creases and cracks by painting on the shadow layer again:

It's these cracks that finalise the sense of material:

As last detail, I found that the shadow layer is too lacking in contrast, so I use the levels filter(under filter-> adjustment) to make the shadow layer darker:

Scale the end result:

So that's a typical diffuse material.

Metals and Shiny Materials.

Metal is not just drawn by making random squigglies: It reflects, so therefore you should draw the reflection.

But, for a proper reflection, you need an enviroment.

So in short, a piece of armour flying in a white void is just going to look white with exceptions of the little bevels which create light-traps…

Therefore, draw a simple background for your piece so we may have something to reflect upon.

(Yes, that is a simple background, it just looks complex because it improves the composition a lot)

I'm making the object I want to turn into metal black in a seperate layer group above our previous ones. Again, I'm using the inherit alpha trick on the layer group.

Try to draw a wire-frame over your object, we're gonna use this as a reference.

So, officially, the way to do reflection is to find the normal, then the line of sight, and have the line of reflection have an equal angle to the normal as the line of sight has.

So, something like this:

This does not need to be 100% correct, but taking some time to do this right pays off.

Try to paint the reflection in as properly as you can.

Darken it a little:

And add highlights on the edges.

Again, story, so dents and scratches, improve the quality of the render.

Roughen up the reflection with bristle_block

Painting in a dent helps out the naturality:

Shiny or Glossy materials use both methods, with the reflection toned down considerably.

Translucent materials, volumes, and skin:

These materials don't just have the light reflect off them, but they also have the light pass through, in different ways.

A piece of colour glass may only have a few particles that light can reflect off.

Glass is the last material you should technically be drawing, because it's dependant on the materials surrounding it, as well as behind.

Technically, we could copy the background and then distort it.

However, you can, of course, always try to draw it.

Make a black layer covering the part you want to make into glass.

And carefully paint in the distortion. Usually, you can get away with painting any kind of distortion:

Add highlights to the edges that catch light:

And use selections with the block_basic brush to create some reflection:

A cloud however will have far more particles for light to bounce off of. So it prevents the light coming through the middle, but not near the edges.

Drawing clouds and fog tends to be different the further away you are from it.

For clouds, it best to think of them as large piles of cotton that is highly affected by the colour of the sky due to how far away you are from it:

So with that in mind, first set up a set of greyscales, each blending into the sky:

Slowly start from a blue and work your way up to white:

Start smoothing the colours out:

Define the lightsources better:

Make the patches of colour less blurry:

Add more white:

Skin, surprisingly enough, is similar to the cloud, and light that is filtered through skin is blood red(because our blood is red, obviously). This can be observed best if you hold your hand in front of a bright light source in a otherwise dark enviroment. The thin parts of your hand will let light through, and you can study it quite well.

But on top of this light-filtering property, there's an oily layer on the skin that reflects as well.

That skin both reflects as well as lets light through is actually kinda important. Especially when drawing people from different races.

Shadow, which is affected by this subsurface scattering, is super obvious on white skin, but not so much on black skin. Similarly, that skin is shiny is super-obvious on black skin, but barely noticeable on white skin. None the less, all human skin has similar amounts(the difference is abstractable) of shininess and subsurface scattering if you would take into account the pigmentation.

Which means, that to draw skin well, you need to master both it's shininess as well as it's quality of scattering of light.

(In the above picture, I use the hard light layer technique again. I use copy layers so that all versions would be exactly the same, and overlaid it over a couple of skin tones. As you can see, the shadows look more detailed and red on paler skin, whereas the highlights are far more detailed looking on darker skin. To make this complete the underlying colour layers should have more variation(more veins in the pale hand, more pigment variation in the dark hand), but to demonstrate the effect of shadows this is sufficient.)

In that manner, handling skin between races is a matter of finding the right contrast than it is about the actual shade.

Glowing objects:

And last but not least is the glowing object.

Glowing objects don't have shadow of their own, unless they are lit by a far more brighter object, but in that case they will not feel like glowing objects at all.

Glowing objects also are, obviously, a lightsource. Which means all other objects get affected by them. Due to this, glowing objects are the first to be treated in a workflow.

(In this case, I added layers with white and set to addition above the shadow layer, which in turn was set to normal. Both were grouped, and the group was set to hardlight, to get the desired effect. )

Some people insist on putting glow-effects onto glowing objects, and I think all these people wear glasses, for you cannot observe lens flares with the naked eye. But, glow effects are easy to implement.

Whether you choose to make a copy of the layer, blur it, and then set the mode to add, while adjusting the opacity for subtlety…

Or you go over the original with an airbrush set to colour dodge…

(Used first yellow, then white.)

Or you use the dodge tool…

There's many roads to glow.

There's one more thing I would like to address: Colour of the light.

Now, this one is tricky, but unfiltered light sources, like flames or stars, have a different colour depending on their warmth, going from red(the coolest) to yellow, to white to light blue and then bright-blue(hottest).

So, a candle will look somewhat like yellow, while a weak fire will have a lot of red in it. Gas-torches and stoves on the other hand are super efficient, and will be very close to blue:

Similarly due to this, the light that's casted is also affected, we go warmer in colour the further away from the fire:

(Gas flames, again, are super efficient, so they're not very good for lighting things up. In fact, for some types of gas flame, you don't see them unless the room is dark: they're that efficient.)

To paint flames, it's easiest if you start with the 'coolest' colour, so for a regular fire, that would be red. Then build it up to yellow, and finally white.

Case Studies:

Now we've looked at the basics, we should have some specific ones to look at!

1. Water.

Water is shiny on top, but cloudy underneath.

This means you can add a gradient to indicate how deep the water is.

The surface can be handled with some highlights or a lot of reflection depending from what angle you looks at the surface. If it's head-on, you'll most see the gradient.

If it's very strong from the side, you'll only see the reflection

2. Foliage.

Not quite a material. But you can create it quickly by using proper brush-masks.

So, let's make a new animated brush.

First off, we make a new document:

And we paint on the first layer a bunch of leafs…

In fact, we paint several variations on several layers…

Now, go into the brush settings, and select the brush tip. Then go to stamp:

You'll see you brush. Set it to animated, and then have selection mode be random. This'll make it use all layers, and select them randomly to use as brushmask.

Also tick 'use color as mask' to make sure it paints with color instead of black.

Name it, and add it to Predefined tips to save it.

Make a new brush with this brush tip, and set rotation to fuzzy and turn on scatter.

It should give this result.

Then pick the darkest colour of green:

And start layering:

Simple foliage.

3. Pearls.

Pearls are known for their ability to shine in all colours of the rainbow. So, to draw them, one has to use all colours of the rainbow.

Let's start simple:

Make a light toned sphere.

Add in colour variation.

Add in shadows and highlights.

Put in a shadow underneath.

blur out the shadows and simplify the colours on the pearl.

The simplifying is so that you can have better control over the shadows and add in more colours.

Once done, blur them out again.

4. Satin.

Cotton and most frabics are difuse materials:

Satin however does something weird to the light which makes the highlights smaller and brighter, and the shadows more prominent. Therefore to transform out cotton into satin, we first increase the contrast by painting large shadows and brighter highlights:

Then we start blurring these sideaways to simulate the effect of the fabric on the light:

Just turn the canvas with shift+space+dragging the mouse, and use block_bristles to make your life easier:

Then, use '5' to reset the canvas, and a color smudge brush to blur away the sharpest details:

Increase the highlights that got lost:

And done.

Conclusion and after thoughts:

Okay, so this was a rather summation like tutorial, but I think you may have a gist of how to create realistic depictions of materials, as well as learned about some neat tricks in Krita.

However, this does not always cover stylistic depictions. As style is very personal, I encourage you to find those out for yourselves. In the meantime, you've now got a couple of nice tools in your arsenal, and hopefully they'll come of good use.

Next time: Maybe the perspective tutorial, maybe the other kickstarter tutorial. Who knows.

Sources:

Skin:

http://graphics.ucsd.edu/~henrik/papers/skin-analysis/skin-analysis.pdf

Foliage:

This idea comes from a pixiv tutorial which… I can't find anymore. If anyone knows which one I'm talking about, please sent me a message. The animated brush part is new though.

Everything else: Wikipedia or general information that seeped into my head over the years… which is probably on Wikipedia as well...

Glossing

You may have heard of glossing in an art context, but we also use it outside of a artistic context, such as ‘glossing over something’ or when referring to the shiny ‘glazing’ on pastries. In an artistic context glazing refers to using half-transparent layers of one colour over the other.

This technique can create fine blends, and has been used for a long time by Oil painters and water colourers.

Above: The Mona Lisa is well-known for many things, amongst which the fine using of the glossing technique. Here the use of glazes allows for the description of the fine fabrics she’s wearing, like her veil, but it is also aiding the soft glow in her skin.

The following gifs are made in Gimp, and we’ll use that as basis for this part.

The logic is as following. Use a brush and draw one colour with it. Now overlay said colour with another colour, but then semi-transparent. You can do this through a brush that allows transparency based on the pressure, or a brush that has the opacity set lower than 100%.

Overlaying one colour with a semi-transparent layer of the other results in a mixture of both colours, which you then can colour-pick. (of course, in traditional painting you aren’t able to do this, instead you are very careful with the opaqueness of the glaze)

Repeating this process a few times over results into a nice gradient:

Now, you would be tempted to think that Gimp’ s paintbrush tool is ideal for this, but the truth is that you’re better off with the Airbrush tool.

The reason is the Airbrush tool’s ‘flow’ parameter. This makes it so that there’s a subtle Dab offset being made. This causes a texture of varying opacities and makes it easier to blend with.

It is much more effective and easier to control than say a paintbrush with a soft brush-tip:

Now, this can feel a little abstract, so here’s a demonstration of our soldier being painted in this manner:

http://youtu.be/H7EgVcZ26hI

(No music on the video)

You can use glossing to mix up easily a coherent colour palette as well:

First, we set our brush blending mode to ‘multiply’, and take a light-pink(for example).

We then use this light-pink to mark the main-shadows:

We then change our light-pick to a light-blue:

And mark the darker shadows:

We then blend everything together like above:

Scumbling

Scumbling rather resembles glossing. But rather than being semi-transparent, a scumble layer is very textured. You see scumbling happening with dry media like pencils and crayons, but also in liquid media like paint and ink with usage of the Dry Brushing technique. The mixing in this case is one created by optical illusion: The colours are so close together that the eye can’t really distinguish between them and will instead observe a mixture.

Dry brushing is when the artist tries to put a very little amount of ink on a brush and paints with it. The result is an extremely textured stroke.

Above: Various forms of scumbling in real-life media: 1. HB pencil, 2. Cheap oil pastels, 3. Charcoal, 4. Colouring pencil, 5. Dry brushing with blue ink and a watercolour brush, 6. Drybrushing with a dried out alcohol marker, 7. glossing with alcohol markers(not an example)

In digital media, scumbling is often mimicked by putting a textured dab on the brush. This is for example how one would handle it in Gimp:

In Krita, next to the textured brush dab, you can also set the texture separately. The system will then simulate how scumbling works in real-life by using the texture as a height-map.

(On top, a texture dab, on bottom, Krita’s texture feature)

The use of scumbling is to create plasticity: rocks, rough fabrics and vegetation is easily made using scumbling.

The second use is that textured shapes are less intimidating to the viewer than evenly coloured shapes.

Mixing

Of course, you can mix paints in real-life, whether this be on a palette on on the canvas itself.

Digitally, though glossing and scumbling can be very powerful mixing tools by themselves, some brush engines add a mixing option right into the engine. In Mypaint such a thing is possible and in Krita this is referred to as the ‘color smudge brush’ engine.

(Forgive me, I constantly forget the name of the color smudge brush engine, here I have deemed it the ‘mix-brush’)

Above: Various forms of mixing available in Krita.

Working with a mixing brush is mostly a lot quicker than the glossing method, and it’s fun to push around the colours on your screen.

If you have a colour smudge brush without a color-rate, you get something akin to the smudge and soften brushes of photoshop. Add a bit of jitter to these brushes, and you will find you have a good brush to crudely mix two contrasting colours with.

Hatching

Hatching is probably one of the most difficult manners of mixing colours. It requires a lot of patience and consistency for the right feel.

The theory is anything but difficult:

You make a row of strokes, and if you need a sense of darker tone, you add a row of strokes perpendicular to that. In computer programs, it helps selecting or masking the region you wish to be hatching in before hand, to prevent making too many mistakes.

But mostly it’s patience and consistency. Having a rotate-view option on your drawing software also helps.

Hatching demonstration in Krita:

Disclaimer: I’m no actual hatching expert.

Our soldier hatched in Gimp, with help of the masking and selection tools. One important thing to note about hatching, and using hatching in faces is that the face can look really old really quickly depending on the amount of lines visible in it. It’s therefore recommended to hatch faces with the thinnest most consistent hatch pattern you can come up with. In fact, it’s better to try and use the next method and avoid hatching faces:

Dithering/Halftone

Back in the day, computer screens couldn’t show off many colours. Printers as well had trouble with high-colour depth: both instruments had only three colours to their disposal!

So the solution was another form of optical mixing. One that could be programmed and be expected to be highly consistent, unlike hatching and scumbling.

Strangely enough, the solution they came up with had been in use for a few years. A few hundred years in fact. If I could find a nice clean picture of it, I would be showing an image of a mosaic of a vase in Ravenna.

The idea behind dithering and halftone is that a strict machinal pattern is used to provide for various amounts of optical mixing between colours.

A set of dithering patterns in Krita at about 200%

The same set at 100%. Notice how the different square seem to be different tones.

Now halftone patterns are similar to dithering, specifically we’ll be talking about the halftone dot pattern, officially known as ‘AM-screentone’ pattern. The other variation would be called the ‘FM-screentone’ pattern, which looks like noise to the human eye.

Modern printers can make the halftone pattern on the fly, and often print at such a high-fidelity you can’t even distinguish the pattern without a magnifying glass.

However, you might want to reproduce the halftone effect for artistic reasons: A texture is just less intimidating than a large flat colour, and it can give a certain sense of plasticity to a drawing as well.

Many greyscale comics like Manga employ halftone effects for these exact reasons.

In Gimp you can reproduce this effect quite easily:

Draw something greyscale:

Go to Filters -> Distorts -> Newsprint.

Set the angle of red, green and blue to the same value so you end up with black dots. Lock the channel. And turn up over-sampling for antialiasing.

Hit OK.

Using selections you can have different settings applied to different parts of the drawing:

In Krita you’ll have to do with G’MIC’s half-tone filter, or use the hatching brush and texture brushes:

The halftone and hatchbrushes. Halftone is based of a texture, hatch is a brush engine.

Our soldier coloured in half-tone in Gimp:

(Note the slapdash hatching and spotting(spotting is placing purely black shadows): Despite it being slapdash it does add to the quality of the drawing, and my recommendation is to always combine the two.)

Conclusion and After-thoughts:

After two highly theoretically based tutorials, another technique focussed tutorial.

These techniques, especially Glossing were techniques I wished I had known myself when I was starting out. Hence me compiling them for you.

Next time: The basics of perspective.

Sources:

Mona-lisa photograph from Wikimedia commons.

When programming, you are writing instructions for the computer to follow.  A set of inter-related instructions is a program.

So, in modern day software programming, you don't actually write the exact instructions for the computer. Rather, you write a recipe that is then translated to machine code. This act is called compiling or building.

This is like you writing a recipe for cookies for a friend, and that friend gathering the ingredients and baking cookies out of them.

The recipe that you write is called the 'source'.

Unlike cookie recipes however, code can be copyrighted and patented. Many software companies do indeed lock their source code into a legal vault. You're not allowed to share the source, you're not allowed to modify the source, hell, you're even not allowed to look at it.

In open source however, there is a different kind of license applied to the code, where you ARE allowed to look at it, share it and modify it.

Open source usually happens when:

A person starts writing a program, but realises it's too much work for one person.

A company realises that if they pool resources with the whole industry including their competitors they can prevent a monopoly happening in their disadvantage.

An institute realising that sharing the technology that they have will fertilise their area of expertise, so they can continue to do work in that area.

A company going bankrupt, and realising that all the hard work that has been put into their software will be lost.

A person writing a program that they consider tedious to write and not wanting anyone else to go through the same tedium.

The source code of open source programs is really easy to get, it's often free to download on the internet. However, you've still got to 'build' it to get it to work on a machine. In fact, with many open source licenses it's perfectly legal to have someone charge you for building the software.

Thankfully for us, many programmers think that the building of a program is such an insignificant price to pay for people actually using the program that they often build it for free. The goal being you start using it and are maybe inspired to help out with the project.

Think of this as a bakery giving away free cookies with the sign: "Free Cookies! (plz help me share and improve the recipe)"

You are in no way forced to help out with an open source project if you start using the outcome. However, even if you don't know programming there's a lot of ways that you can help. Primarily, all open source projects have a bug-tracker where you can report bugs. Writing manuals and sharing tips and tricks with friends is great too. And finally, you can donate to the development fund. This fund is used to provide people who can program with traveling funds, money to acquire specific hardware(think drawing tablets) and sometimes even providing a programmer with a salary so he can work on the project full-time instead of spare time. And it isn't uncommon for people to start learning how to program exactly because they want to help improve an open source project.

That said, let's get to a list of programs I tend to use:

Gimp(Gnu Image Manipulation Program)

Site - Unstable Windows and Mac builds - Bug Tracker

Often pushed as 'Free photoshop', Gimp isn't great for painting, but is excellent at the nitty gritty technical stuff that is required for photo manipulation.

MyPaint

Site - Unstable Windows build - Bug Tracker.

A program focused on distraction-free sketching and painting, not much unlike say Sketchbook. Very interesting because it tries to come up with innovative tools like infinite canvas, procedural brush system and interesting colour pickers. Also very lightweight, and I've used it for both sketches as well as full-out paintings successfully.

Note: Version 1.1 didn't appear on windows because of tablet bugs, the unstable 1.1 build above tries to fix this. So if it works for you, tell them!

Krita

Site - Unstable Windows builds - Unstable Win XP build - Bug Tracker.

A program that focuses on sketching and painting like MyPaint, but has powerful editing functionality like Gimp. Most unstable of all three, but also most rapidly developed.

Krita Sketch is a variation made for  slate tablets, and Krita Gemini is a variation which combines both, for hybrid devices.

Inkscape

Site - Unstable windows builds - Bug Tracker.

A vector editing program. I have not gone into vector much, but I will in the future. Inkscape is not as powerful as Adobe Illustrator or Corel Draw, because it tries to maintain compatibility with the internet vector standard SVG. However, it has been sufficient for my needs and I recommend checking it out.

IMPORTANT!: The unstable builds are called that way because they change a lot and thus are PRONE TO CRASHING AND BREAKING FUNCTIONALITY. They are for the ADVENTUROUS and SAVE-HAPPY. However, when you come across such a bug when using them, you'll be a great help in reporting it at it's respective bug tracker!

My initial reason for using open source software for these tutorials is because I would like to offer alternatives to pirating expensive software. People in very specialised fields still have needs for their commercial software because specific functionality hasn't been implemented yet in the open source alternatives. However, there is a LOT that can already be done through open source software and I want to document that carefully.

Next to that, a lot of these projects have interesting and innovative tools I would like to show off, especially when they solve tedious tasks.

I hope this clears up a bit of confusion, and I hope you'll be able to find as much enjoyment in using these programs as I do :)

Primaries:

So. Colour theory is a mess. Over the years, there have been many, many colour systems.

You may know one of them:

If you recall your elementary school days, Red, Yellow and Blue are THE PRIMARY colours. You can mix all other colours from it. We have been acknowledging these as the primary colours since medieval times.

It’s also horribly outdated.

As I said, there’s been many many colour systems over the years, and both David Briggs’s “Huevaluechroma.com” as well as John Gage’s “Colour and Culture” try to make sense as to what the logic was behind these systems. Briggs doesn’t really give us any explaination why it is Red, Yellow and Blue that was favoured by so many artists, but Gage points out that Vermillion Red, Gold Yellow and Ultramarine Blue were the most expensive pigments in Medieval times, and thus revered. Later artists never quite disputed this, and thus kept Red-Yellow-Blue as basis for their colour systems, amongst which Johannes Itten, who has had a lot of influence into how colour theory is taught. Considering how the ‘basic paint set’ that most art stores sell usually consists out of Vermillion red, Gold yellow, Ultramarine blue, Titanium white and Lamp black, I feel Gage’s explanation is pretty good.

What would be a closer estimation is to accept there’s about two or three different colour wheels.

The first one is the quadratic psychological colour wheel:

Now, I wrote something incorrect in an earlier tutorial, which by the time this is posted is corrected. This is because I had been taught a wonky model of how colour vision works in highschool: That Humans can only see red-green-blue.

This is not the case, rather our cones have variety of sensitivities to specific colours. The signals being sent to our brains however, come in four main variations: Red, Yellow, Green and Blue.

With the opposite signal of Red being Green, and the Opposite signal of Blue being Yellow.

Together, these colours make the psychological colour wheel.

The second(and third) colours wheels are based on the most efficient colours to mix colours in light: Red, Green and Blue:

This is also called the ‘additive colour wheel' because mixing all colours results in the most energetic colour: white.

Now, the most efficient colours to mix pigments in is similar to the above colour-wheel: Magenta-Yellow-Cyan:

This is called the “Substractive colour wheel”, because mixing all colours results in the lowest energy colour: Black. (Well, usually it’s a muddy gray, but in theory it should be black, pigments are difficult).

The substractive colour wheel is slightly different from the additive colour wheel, because you can mix a smaller amount of colours from it than the additive colour wheel. This range of colours is called a ‘Gamut’.

And the visible gamut for humans is described in the CIE XYZ system(So often described as ‘a horseshoe shape’, that it might as well be known as ‘The Horseshoe’)

(Because your screen can’t show all visible colours, this picture is only a mere representation)

The Weird Horseshoe shape is caused by how our three cone types are differently sensitive to different colours.

Attempting to pick three ‘Primary colours’ from this would result  in three colours that are outside the visible gamut. Three so-called ‘Imaginary colours’, because we can only talk about them in theory but not actually observe them in reality.

Now typical computer screens have a smaller gamut than the XYZ gamut. You need a profiling device to find this gamut, and the information for it is stored in an ICC profile, which in turn can be used by your operating system and graphics software to show you the best possible colours. Otherwise, your system assumes a profile called ‘sRGB’.

Colour Spaces:

Going further into colours spaces you may have noticed that some graphics software allows you to choose a ‘colourspace’.

So, you already know RGB, which is the colour space your computer screen uses standardly.

CMYK is Cyan, Magenta, Yellow, Key(black), this is a subtractive colour circle as used by printers.

Then there’s XYZ(The HORSESHOE!)

YCrCb, which is… interesting. It tries to approximate human vision by setting Red versus Green and Yellow versus Blue, alongside Luminance(Linear lightness, explained a little later one), used in JPEG compression. Unfortunately, to use it in Krita you need an YCrCb ICC profile, and these are hard to find.

And finally, there’s La*b*. La*b* is another approximation of XYZ, with L standing for both lightness as greenness, a* being Green versus Magenta and b* being Blue versus Yellow. This one is often used by photographers to convert their photos from one colour space to another. There’s similar colour spaces to this one, like LUV and CAM, but these last ones are not very common.

I recommend you either work in RGB when the output will certainly be for screens, such as webcomics or sprites for video games, when working for print, try using CMYK. XYZ, YCrCb and La*b* are a little broken in Krita at the moment.

The second way of approaching colour spaces is by looking at the precision. Early computer screens could only display two colours. Black and White. In computer science something that can be either one thing or the other is called binary, and the storage is called a bit. So this was a one bit display.

In modern times, most displays are 24bit and can show about 16.5 million colours. Most images are 32bits, or 8bits per channel. There’s also 16bits per channel images, which are used in photography and rendering for optimum precision. Even in painting 16bit per channel can help with getting the smoothest gradients. Usually however, you’ll be using 8bit per channel.

Terminology

Now, while going into colour selectors, let’s look over some terminology.

Hue. The hue is a description of the colours along different wavelengths/frequencies. Basically, whether something’s red, yellow, green etc.

Brightness(Often called Value) is how close a given colour is to black(officially: “attribute of a visual sensation according to which an area appears to emit more or less light”).

Saturation is in HSV how close a given colour is to White(officially,”colorfulness of a stimulus relative to its own brightness”), in HSL it refers to ‘Chroma’.

Chroma is how much a given colour is close to grey(officially, the “colorfulness relative to the brightness of a similarly illuminated white”).

Lightness is where a colour sits between White and Black, it’s occasionally referred to as the Tone.

In using this triangle, to say, get the Lightness properly, you need to imagine a line going from mid-grey to the purest red. Then you draw lines from black to this imaginary line and then to white:

The above is often described as a HSV(hue, saturation, value)-colour picker.

We can also just morph this triangle into a square:  

Which gives us a Lightness-Chroma square. This is what is used as a HSL(Hue, Saturation, Lightness) colour picker, with the Saturation being a renamed ‘Chroma’.

We can also do this triangle-to-square with the saturation and value as basis:

This is one where the Saturation(horizontal value) is most important.

And here the Value (vertical) is most important.

All three screenshots are taken from the Krita advance-colour selector where they are all options.

There’s a difference between nonlinear lightness, where the fullest chroma of a colour is mapped to the middle of the lightness spectrum.

Generic ‘Lightness’, Luma and Intensity are examples of Nonlinear lightness, which means they are based on perception of where humans see a difference between different tones.

Luminosity on the other hand, is linear, and is solely based on how much a given colour radiates light.

The MyPaint HCY colour selector. The interesting thing about both Luminosity and Intensity is that they acknowledge that Yellow looks lighter than Blue. In Luminosity’s case this is because it is based on a measurement of how light the colour yellow is. In Intensity’s case, they note that screen yellow (255,255,0) has on average a higher value than screen blue (0,0,255)

Krita’s Artistic colour-selector with ‘HCY’ on the left and ‘HSI’ on the right. The difference is subtle, but there.

Colour harmonies:

Opponency Complementaries:

We’ve mentioned opponency before, it’s two colours where one gives an opposite signal to the other:

These can be confirmed through the after-image experiment.

Triadic Complementaries:

A complementary harmony is when the colours are opposite to eachother on the rygcbm colour wheel:

You also have a split complementary which comes in two variations:

Both complementary harmonies have the effect of creating a balance.

Analogue colours on the other hand have the effect of strengthening each other. Analogue colours are colours that are close to each other on the colour wheel.

the complementary colours have a large effect on the colour of grey:

If you surround grey with blue, and if you surround grey with red, you have the feeling you’re looking at different shades of grey.

This because grey is fairly yellowish in contrast with blue, and grey is fairly greenish in contrast with red. This is interesting, because it allows for grey to stand-in for yellow or green in blue or red dominated palettes. Which bring us to our next topic:

Gamut Masking

James Gurney realised that pictures with strong colour harmonies are using an even smaller gamut of all the colours they can use. So he decided to try and systematically make this smaller gamut: Gamut Masking.

The most basic gamut is a triangle:

The outmost points are then your primaries:

Between those you have your secondaries:

Tertiaries:

Map those to a variation of tones:

Tada, the rainbow(Gamut masked).

There’s more shapes you can use, like for example complementary:

All these examples were made in MyPaint 1.1, which has gamut-masking build into the palette. For other programs you will have to mix this palette yourself.

Relation to shading

(why shadows are blue)

So, around the 18th century, a Dutch scientist, Hans-Christiaan Huygens, discovered that you only needed yellow and blue light to end up with white light. This was promptly adopted by the Dutch painter Johannes Vermeer, who continued to use yellow to tint his lighter areas and blue to accent his shadows. Indeed, out of the four psychological primaries, Yellow is the lightest colour and Blue is the darkest colour. So no wonder that many colour palettes are based around dark tones being blueish and light tones being yellowish.

However, in reality, shadows are just absence of light, so black versus white. On a sunny day, when the sunlight is so bright it does indeed make things look yellow, you do see blue in the shadows. However, this is not because blue is the complementary AND the opponent of yellow, but rather because our blue skies also light us up:

Indirect Light and Realistic colours:

We looked at indirect light last time, however, what I didn’t mention is that not just does light bounce off surfaces, it also loses energy. To put it differently: It takes over the colour of the thing it bounces off. (Rather, an object having a certain colour is because this specific object reflects that colour most efficiently)

Indirect light is the main cause for ‘coloured shadows’, and simulating the reflecting of colours is the best way to get realistic colouring.

Because of it’s dependency on the lighting, realistic colours are determined per image, and per the lighting situation of the image. Stylised colours on the other hand can be both pre made on a palette, as well as be designed like realistic colouring is.

Basic ramping formations and Stylised colours:

A ramp, in this case, refers to a set of colours going from one colour to the other. These are for the creation of stylised colours.

The simpelest ramp would be a white-to-black ramp:

Then comes the time to introduce colour: a lightness ramp.

Now we can have variations in the colourfulness(Chroma/Saturation):

And we can have variations in the hue(hue-shifting):

The most common hue-shifting goes from yellow to blue along red for ‘warm’ colours and along green for ‘cold’ colours. And through the middle for variations in saturation:

The leftmost ramp.

The middle ramp.

The rightmost ramp.

The second most common hue-shifting are the analogue ramps:

Have you ever noticed how Cyan, Magenta and Yellow look lighter than Red, Green and Blue?

If you use two of these colours aligning in a colour ramp, you get a much brighter effect than if you were to use regular lightness for the colour ramp.

The analogue colour ramps are ideal for lighting effects such as fire, or sparkles.

And last, but not least is the spiral:

The spiral goes inwards towards grey as it gets darker, and outwards towards the saturated colours as it gets lighter.

Above example spirals on HCY colour selector.

The right-most spiral on a HSV colour-selector.

Practical

For this tutorial I will solely use Krita, because I’m building on top of the shading tutorial of last time. For Gimp, combine this with the shading tutorial of last time.

First things first, we’re going to check our tone-balance.

The tone balance is how the black-white contrast of our image. Humans are actually more sensitive to difference in tone than differences in colour, so checking your tone-balance can be great in balancing the colours of your drawing.

So how does this work? We’re going to turn our colour picture into a greyscale image. You can use the desaturate filter for this, in Gimp under Colors->Desaturate and in Krita under Filters->Adjustment->Desaturate. However, I’m going to suggest using a different method. Make a white layer that is above everything in the stack. Set the Blending mode to either ‘Color’ or ‘Saturation’(Both can be found under HSY tab in Krita). everything looks grayscale now.

Now, as you can see, the hair still feels darker than the skin, so does the armour.

What you want to avoid here is this situation:

An extreme case, but the grayscaled version might as well be a statue.

This one’s not too great either but still better than the previous example.

This one’s okay.

These darker-skinned versions are okay as well.

Based of my greyscale, I’m going to make the eyes a little darker to get a more balanced feel, I’ll be doing this by placing a darker greyscale layer above my colour layer for the iris, and set the blending-mode to luminosity(In Krita under HSY):

Darker eyes: More balanced tones.

You can also check the Chroma/Saturation balance by setting the colour of our white colour-bended layer to ‘#803F3F’(or a mid-tone red)

Set this one to luminosity:

As you can see though, checking chroma/saturation is a lot more difficult than checking tone because we don’t perceive the difference that strongly. That said, it’s better to have more desaturated colours in an image that highly saturated parts, because saturated colours can hurt one’s eyes if used too much. Grey, as a natural colour, doesn’t make us suffer.

However, some works do require low contrast in tone or extreme saturation, and these are merely diagnostic tools.

Making colour ramps!

First, we get our base colours(or midtones, or flats… etc.) and we make fully opaque stripes on a separate layer:

You then mimick the shadow layers by setting the brush to black and 50% opacity. Draw a horizontal strip with this over all vertical strips. Do the same for 25% opacity black, and 50% opacity white with the blending mode set to ‘overlay’ :

If you did everything correctly, you should have the same colours that you have on the shaded drawing now. These are your basic ramps.

For the first type of ramp, I’ll be showing how to determine colours when you’re making the indirect light blue, similar to the colour of a sunny day.

Now make a new layer and add light blue strips on top of everything:

Set the blending mode to color.

Now change the opacity to say… 40% (Maybe a little darker, maybe a little lighter)

And from that we’re going to build our colour ramps:

You can follow this through with the other colour-ramps. The metal gets to be blueish earlier because it’s very reflective. Skin, I have two variations here, one which is more reflective than the other.

Applying these colours to the base drawing with separated shadows first requires that you set the layers to 100% opacity and ‘normal’ blending mode.

Then set the alpha-lock on the toolbar(the checkers):

And use Shift-backspace to fill a whole layer with the foreground colour.

And here’s an example with the two different skins, as well as the armour less reflective. In the end, I decide to have the skin less reflective and the armour very much so:

Now, we’re going to do a yellow-blue hue-shifting variation. Pick a colour from the base ramps, and depending on whether it’s lighter or darker than your midtone, shift the hue towards yellow or towards blue.

Hue shifting requires a lot of tinkering, and you need to experiment a little to get the desired colour ramps. (For example, I’d probably tone down the blue in a hair a little more)

Here’s an example where colour ramps were made with the MyPaint colour selector:

Here’s an example with colours straight from the candy palette:

It should be obvious as to why I named it the candy palette. It should also not come as a surprise that I typically only use the candy palette diluted.

Conclusion and Afterthoughts:

Colour theory is one of the muddiest areas of art theory. This is partially because we’ve been using outdated systems, but also because colour is very intuitive: Despite all these ‘outdated’ systems, there have been many great artworks that used colour extremely well.

Because of that, I encourage you to experiment a lot with colour. It’s very common that people stay stuck in the shift-towards-blue ramp building, while experimenting with harmonies and different manners of approaching light can really broaden your visual vocabulary.

I’ll probably return on this topic as well, mostly because I didn’t cover symbolism in this (already waaay too long) tutorial, as well as examining special cases, like fire, pearls and skin.

Next-time: Hatching, Mixing, Glossing, Dithering, Scumbling

Sources:

On Opponency: http://www.huevaluechroma.com/032.php (The site is overall a very good read), http://en.wikipedia.org/wiki/After_image http://en.wikipedia.org/wiki/Opponent_process

On Blue-Yellow: John Gage “Colours and Culture: Practice and Meaning from Antiquity to Abstraction”, paperback edition 1995,  pg 154

On red yellow and blue being chosen for their expensiveness: John Gage “Colours and Culture: Practice and Meaning from Antiquity to Abstraction”, paperback edition 1995,  pg 158

On HSV, HSL, HSI and HCY terminology: http://en.wikipedia.org/wiki/HSV_color_space http://www.huevaluechroma.com/081.php

On ICC profiles, RGB gamut and XYZ gamut: http://ninedegreesbelow.com/photography/xyz-rgb.html

Image of the XYZ gamut: http://en.wikipedia.org/wiki/File:CIE1931xy_blank.svg

YCbCr: http://en.wikipedia.org/wiki/YCbCr

Shadows: Cultural theory.

This doesn’t have to be so. A lot of early Egyptian and even medieval art didn’t use shadows. An easy explanation to assume would be because they didn’t know how to draw shadows, but this might not always be the case. One thing to remember is that a lot of art from these periods was religious art, and graced buildings. Often religious art is primarily symbolic than it is realistic, and the importance of specific colors and composition outweighed shading. Aside from that, carved relief are thought to be commonly painted as well, though there’s few reliefs left over today that are still painted. The result would be a sense of dynamic shading happening through the shape of the relief.

A relief from the Abu Simbel temples. If you look carefully you can see the reliefs are painted. As well, you can see how the use of relief here creates shadows dynamically.

Mosaics from Ravanna, definitely showing shadows on the cloth, but hardly any cast shadows.

What they probably were more interested in was conveying the materials that they were trying to depict: Showing saints clothed in fine silks and delicate gold. Shading may have been the best way to describe these materials visually.

It was during the renaissance times that we start seeing a lot of complex shading happening in paintings. Carrevagio in particular was well known for his use of shadows, being more or less the trend setter of using shading to create strong dramatic effects.

Wikipedia attributes him with starting the Baroque movement, an aesthetic style which is largely dictated by the rule of drama rather than the rules of realism:

"Caravaggio "put the oscuro (shadows) into chiaroscuro.”[38] Chiaroscuro was practiced long before he came on the scene, but it was Caravaggio who made the technique a dominant stylistic element, darkening the shadows and transfixing the subject in a blinding shaft of light. With this came the acute observation of physical and psychological reality which formed the ground both for his immense popularity and for his frequent problems with his religious commissions.”

To these religious painters, to paint the divine meant to paint light. And ever since, the best way to paint lights is to paint shadows.

The most realistic use of shadows came around the end of the ninteenth century, when the impressionists attempted to catch what they saw as objectively as possible. While before them shadows were often dramatic and theatrical, the impressionists showed very subtle though messy shadows, caused by the ever changing light. If you can get past how messy the impressionist era paintings were, you can see the striking amount of realism they contain.

Edouard Manet’s ‘At the Café’, one of my favourite pictures to show off that impressionism mattered: Aside from the unusual intimate composition(for the time), the use of shadows results into a very photo-like feeling painting.

So that would be some reasons for you to paint shadows:

*To showcase material. (Plasticity)

*To paint lights effectively.

*For dramatic effect. (Symbolism, communication)

*To mimic reality.

Shadows: Technical theory.

Now we have that little bit of art theoretical thought out if the way, lets continue on in physics.

Previously, I had asked you to think of Light as waves, like sound. Now I’m going to ask you to look at the alternative way to look at light: little particles. Not that you can’t model shadows convincingly without thinking of them as particles, but it is a little easier.

So, typically, we model shadow like this:

One light with parralel rays casting a perfect shadow. Paralel rays though, are typically only sunlight. And even that’s because the sun is so far away from us.

Most light sources have divergent rays, meaning they start at a single point and cast outwards:

And that source can be pretty large as well:

Finally, light has another thing it does: Namely bounce more than once, causing indirect light:

So in effect, even in a single lightpoint scene, and assuming there’s a surface for the light to bounce off of, an artist will always have to assume an object is lit by more than one light: Direct and indirect light. A situation without indirect light is one where the light doesn’t have something to bounce off from, like outer-space. But even outer-space situations have places where indirect light occurs.

That said, aside from paralel and divergent light, there’s one final light-type: Convergent light.

It’s very rare though, and the most common case of it involve lenses.

Now we’ve established the different types of light and the cast shadows they create, let’s look at the form shadow and how different materials reflect light.

If we assume that a material reflects perfectly, you get a mirror effect, like chrome.

There’s also materials that let light through, like glass, but also air.

Finally, a little less known, there’s materials that are a little of both, called diffuse, caused by an effect called subsurface scattering. I subsurface scattering, the light hits the surface, but because the surface is slightly transparent, it goes through it, bouncing off the non-transparent parts and bouncing out again.

This is evident in most materials, but most obvious in wax and skin. The notable difference between an roughened up perfectly reflecting material and a sub surface scattering material, is that glossy materials carry a highlight with the colour of the lightsource, and diffuse materials carry highlights with the colour of the material. We’ll go into colours interaction with shadow deeper next tutorial.

Practical

For this tutorial, we’ll stay in the realm of cel-shading, because the most important thing is to determine where the form shadow and cast shadow is. Cel-shading is an abstracted type of shading which was often used in Cel-animation(animation where the characters and other moving parts were drawn on transparent celluloid sheets and overlaid on top of a static background). Abstracted shadows were much easier to animate than more realistic shadows, and for us it’s easier focussing on the basics this way.

We’ll delve only a little into material rendering. We also won’t be delving into shadow colour much yet, that would be next tutorial.

So, first I’m going to showcase a method using Krita and putting the shadows and highlights on a different layer than the base colour. For this method I’ll only use pure black and pure white to draw these shadows and highlights.

The second method will also be reproducible in the gimp, but the method is a little more destructive.

Alright, for this method, we’re going to organise our layers:

With masking

Krita can make use of grouping. Set your layer in  layer group.

Now right-click and ‘select opacity’, then right click and select ‘layer mask’

Now click the blue arrow to put the layer mask into the group.

Then, between the layer mask and the base colour, add a new layer, if you draw on this, you’ll notice you don’t affect the transparency.

Alpha inheritance

Alpha inheritance, or layer clipping, as Photoshop calls it, is a slightly sped up version of the above. Put the base colour on a layer and add a layer above it. Now toggle the a symbol.

Alpha inheritance makes a layer ‘inherit’ the alpha of the lowest layer in a group.

While fast, it doesn’t affect groups inside the group, while masking does.

(Note: Photoshop’s version abstract the grouping, Krita doesn’t, so it’s very important to make you you have a group set.)

Now we have covered that:

Let’s shade!

Let’s find the cast shadow.

The first thing you need to decide is where the light is coming from. For our sake, we have one  direct white light. And opposite, we have the indirect light. (Supposedly our light is hitting a white wall or something).

Next up, we look for the terminator of the shadows of the direct light. That’s where the direct light can’t reach.

In computer graphics, the terminator is found by taking the outmost point of where the light can reach:

And forming a silhouette with that. That would be the terminator. Everything away from the light on the terminator is cast in shadow.

Try to abstract these more complex shapes as the above stacked together.

Now we’re going to make the cast-shadows.

Similarly, we can find the cast shadow by taking our terminator and using the light to pass by it till it hits a surface.

Of course, we also need to think of the indirect-light. She’s not in outer space after all. But let’s abstract it a bit. Set all shadow-layers to around 50% opacity along side drawing the formshadows.

In videogames this technique is used to determine the cast shadow, this is called volumetric shadow mapping. Most artists deal with shadows fairly intuitively, and don’t worry about drawing all the help lines.

Work out the form shadows a little more. Detail the hair a little, and add or remove form-shadow from places. (Don’t forget that pressing ‘E’ sets the brush you’re using to eraser-mode)

Alright, got the basic shadows done.

Now let’s add high-lights. I’m going to avoid highlights on the skin, just like how I am going to abstract away the translucency of the skin.

Highlights, specular highlights specifically, have a manner of scematically being determined as well. However, just like shadows many artists apply them intuitively.

Set the highlight layers transparent as well. Also set the blending mode to overlay to make the highlights look less washed out.(Don’t set the highlights to overlay on the eyes, having it pure-white like this is for the best).

We’re almost there.

Now, let’s put in the ‘indirect-light’ hotspots. You’d typically just erase from the shadow layer, but we’ll use a layer with an erase-blending mode.

Set each shadow-layer you use this on, in a separate group(this is where the masking comes in handy) now add a layer on top of it and set the blendmode to ‘Erase’. You can set this one half-transparent as well. Now draw on that layer and see how it erases from the shadow layer.

(Red is direct light, green is indirect light)

Use this to draw spots of indirect light in the shadow without messing up your shadow-layer.

Don’t forget to simulate the indirect light being cast from the subject themselves.

Tada, looks quite nice, huh?

Now, indirect light is much more complex than what is depicted here, but we also abstracted a lot of it by setting the shadow layer to semi-transparent.

Doing this in the gimp:

First off, we take our image in the gimp.

We select the alpha-lock over here.

And notice that we can’t change the transparency of the layer :)

For this I’ll be using the Ink tool because it gives nice anti-aliased lines. I’m setting the mass high, which means I have less control over the lines but they’ll end up smoother.

Set the colour to black and the transparency to 50%, and paint shadows:

You’ll soon notice something very annoying though:

Going over the same place with your brush makes it become darker. Maybe not what you’d want with cel-shading.

So, we’re going to pick the shadow colour. In most graphics software, there’s a so called sticky-key for the colour pick tool when you are using the brush. Meaning that if you press this key, you go into colour pick mode, but unlike a regular hotkey, releasing this sticky-hotkey returns you to brush mode.

In photoshop, much to my left-handed annoyance, it’s hardcoded to the alt-key. In Gimp it’s set to the CTRL key(yay!). In Krita it’s standardly CTRL, but can be configured, like Mypaint. Painttool Sai and many other Japanese graphics software have it set to the right-mouse click, much to my preference.

Use ctrl+click to pick the colour of the shadow.

Set the opacity of the brush to 100% again.

And continue drawing the shadows like we discussed in the Krita example.

For the highlights, set the colour to white, the brush blending mode to ‘overlay’ and the opacity to 50%:

Draw some spots:

Pick the colour of the highlights again:

Set the brush opacity to 100% and blending mode to ‘normal’ again.

And refine the highlights further:

For the indirect lights, pick the mid-tone(the one we used for the flats) and set the opacity of the brush to 50%, and start drawing:

Pick the resulting colour again. Set the brsuh to 100% opacity again:

Do this process with the other layers as well:

Now, use the colour picking to switch between colours as you detail the shadows.

For certain areas though, you’ll find you only want to edit the shadows but not the highlights, or only the highlights and only the shadows, like this area:

We have something for that!

Set the Blending mode to ‘Lighten’ or ‘Lighten only’:

Drawing with the mid-tone should now only affect the shadows:

The ‘Lighten’ blending mode makes it so your brush only draws if the colour you’re drawing with is lighter than the colour you’re drawing over.

Similarly, the ‘Darken’ blending mode will only paint if the painting colour is darker than the colour you’re painting over. Using these brushes to you advantage can help you fine-tune your shadows a lot easier.

Finte-tune the shadows to your hearts content, at some point when you feel it’s finished, it’s finished:

This method is a lot more ‘destructive’ than the layer-method: You can’t go and select an area and change the colour easily. The advantage is that you end up with less layers, and so less heavier files as well as less finding of the correct layer to shade. It’s also a better way to set up you’re shadows if you’re preparing for soft-shading or painterly works.

Conclusions and Afterthoughts:

Shading is one of the big elements of drawing and painting. This can be seen through that it’s one of the subjects of drawing that you never stop studying. Similarly, taking your time and laying down the forms shadows and casts shadows, highlights and indirect lights carefully, can really improve the overall quality of your work, as well as making yourself more efficient in the long run.

However, it’s not always necessary to draw shadows, and you can wonder for yourself why you feel a particular work needs shadows or not.

As it’s a subject you can keep on learning about, I’ll be going into shading more in the future, amongst which it’s interactions with other elements of drawing and painting, such as colour theory!

Next week: Colour theory and building ramps.

Sources:

Basics of light and shade: http://www.huevaluechroma.com/021.php (Definitely visit this site if you want to go into the deeper basics of drawing shadows)

Images Carravagio, Abusimbel, Manet and Ravanna: http://en.wikipedia.org/wiki/Caravaggio

Theory

For this tutorial I’ll be using Krita. Gimp has good layer functionality as well, but some layer types are yet to be available in Gimp. I’ll also describe how certain elements are named in Photoshop.

Most image manipulation programs have layers systems however, and I’m sure I’ll describe at the least one method you can use.

In Traditional art, it was not uncommon to first draw the full background before drawing the subject. Or to first draw a line art and then colour it in. Computers have a similar way of working.

In programming, if you tell a computer to draw a red circle, and then afterwards tell it to draw a smaller yellow circle, you will see the small yellow circle overlap the red circle. Switch the commands around, and you will not see the yellow circle at all: it was drawn before the red circle and thus ‘behind’ it.

This is referred to as the “drawing order”. So like the traditional artist, the computer will first draw the images that are behind everything, and layer the subject and foreground on top of it. The layer docker is a way for you to control the drawing order of multiple images.

Remember how I was talking about the Alpha channel last time? Without them, layering would be a lot less useful.

Each layer in a layered image is an image of it’s own and has it’s own Red, Green, Blue and Alpha channels. When you save a layered image, each of these images are stored as separate images in that file.

But there’s more than these layers being stored.

There’s the layer hierarchy: You can group layers together to make them less confusing.

There’s layer states. You can hide layers, or lock them, so you don’t accidentally edit them. Or only lock the alpha channel, so the transparency is preserved no matter what brushes you use.

Or, you can force the layer to inherit the alpha channel of an underlying layer(Alpha inheritance or Layer Clipping).

There’s operations: Filter layers, or “Adjustment Layers” as Photoshop calls them, are literally a layer of filter. So you can make a layer that shows a brightness-contrast being applied to the underlying image, but doesn’t destroy the underlying image. Hence this often being reffered to as “non-destructive”.

Similar to filters, there’s Blending modes. Because all colours can be expressed as a collection of numerical values, you can apply all sorts of maths to them. Blending modes are these maths formulas. I’ll be going into them briefly this tutorial and will dedicate some time to them in the future.

And finally you can make layer masks. In traditional art, a mask is a piece of paper or a layer of rubber that protects the parts of an image underneath from pigment. A stencil is a type of mask, that you reuse often, but there’s also masking liquid that is single-use only. Digitally, a layer mask or alpha mask is a greyscale image which adds transparency to a layer or layer group depending on the amount of black in the mask. It’s similar to a traditional mask in that you can prevent certain partsof the underlying image from being effected by the layer that contains the mask.

Preparing lineart.

This knight will be the drawing I’ll be using for this tutorial. It’s a black lineart on top of a white background. Feel free to follow the tutorial with her, just don’t remove the signature.

Now find you layers tab and add a new layer.(Hit the green plus symbol)

Put the new layer underneath the layer containing the lineart(drag and drop or use the blue arrows for that), and draw on it…

…And notice nothing happening. This is because the white isn’t transparent. You wouldn’t really want it to either, how else would you make convincing highlights? So what we first need to do to colour in our drawing is prepare our lineart. There’s several methods of doing so, each with varying qualities.

(If you used the colour to alpha method in last tutorial, you can skip preparing the lineart: Your whites should be transparent already(Indicated by a checker background in almost all image manipulation programs).)

1. Multiply

So, typically, to get a black and white lineart usable for colouring, you can set the blending mode of the lineart layer to Multiply.

Tick the word “Normal” to get a dropdown.

Krita has a lot of blending modes, which leads to the unusual grouping of them. You can find the ‘Multiply’ blending mode under Arithmetic. Tick the box to set it as a favourite, for easy future access. Click the name multiply to set the blending mode to multiply.

And you should be able to see the underlying layer come through the white.

Multiply is not a perfect solution however. For example, if through some image editing magic I make the lineart yellow, it results into this:

This is because a Multiply doesn’t make the white transparent. Rather, as a blending mode, it takes the values of the colours of the blended layer, and multiplies it with the colours of the lower layer.

If you double click the active colour-swatch:

You get to see the select colour screen:

It shows you the numerical values of a colour. So yellow in this case has a red value of 255, and a green value of 255, and a blue value of 0. These values are used to multiply the colours.

As you can see from the magenta-yellow example, Multiply is a great way to simulate the mixing of real-life pigments. In general, blending modes are a digital artist’s best friend, and hence I’ll be going into detail about Multiply and other blending modes in the future.

Advantages: Easy, can work to your benefit even with coloured lines by softening the look of the lines while keeping nice contrast.

Disadvantages: Not actually transparent. Is a little funny with coloured lines.

2. Colour channels

We went into this technique a lot in the previous tutorial, and like how we can remove the blue lines using the colour channels, we can also use the ‘channel to selection’ to only get the black lines out.

In gimp, make a transparent layer above the lineart.

Then select all(Ctrl+A or Select->All), and go to the colour channel docker.

Right click the red channel(or any of the other two, remember this is a grayscale image) and select “subtract from selection”.

Now fill this selection(Ctrl+comma, or Edit->Fill with foreground colour).

Deselect everything(Shift+Crtl+A or Select->None) And hide the lowest layer.

Advantages: Works in all image editing software with colour-channel selections, including Photoshop. Actual transparency.

Disadvantages: Fiddly, doesn’t work if the base-lineart is not black and white.

3. Masks

Masks are really easy to use in this case, because our image is already black and white, with one of those colours(white) indicating where we want it transparent. Now, the thing is that, unlike traditional art, where black is seen as having value, because it indicates more ink being used, in digital art it’s white that has value(light energy). And with masks that value is translated to opaqueness. So in both examples here, you’ll notice I’m inverting the values to make sure that the white is translated as being transparent instead of opaque.

Krita

Make a new layer, and fill it with black(shift+backspace, or edit->fill with foreground color).

Right Click the lineart layer ->Convert->Transparency mask.

Now the underlying layer will have the lineart layer as mask, meaning that the black parts are transparent. We want the white parts transparent, so we select the lineart-mask and invert it. (Filter->Adjustment->Invert or Shift+I).

Gimp

Right click the lineart layer-> Add layer mask ->Grayscale copy of layer(with invert selected).

Your layer mask should be in place now.

The image to the right of the layer image is the layer mask. The white border around the layer mask means that the layer mask is selected for editing. With a white border around the first image means that the layer itself is selected for editing.

->

To get rid of the last white artifacts(There are white artifacts, trust me) you fill the base-lineart layer with black(Ctrl+comma).

Advantages: Actual transparency, separating said transparency from the colour, and with colour channel-selections also reproducible in Photoshop.

Disadvantages: You can still get quicker. More entries in your layer menu. Doesn’t work without grayscale linework.

4. Colour to alpha

Available in both Krita(Filters->Colors->Color to Alpha) and Gimp(Colors->Color to Alpha).

We went over this filter in the previous tutorial.

Click the white bar to change the colour you want to remove. In our case, the default White is fine.

Advantages: Quickest way. Works with coloured linework.

Disadvantages: Not all programs have this filter. No strict separation of transparency and colour.

Flats

Again, with the actual colouring in there’s a bunch of methods. Depending on your software and/or lineart, you can apply these methods to their full benefit.

Before we start, I would like to recommend you to be careful with the colours you choose for your flats. It’s best to avoid over saturated colours in your base-colours, as these can be painful to look at. Typically, we’d only use very saturated colours when we try to really put in some special effects. The best practice is to have a balance in the variety of saturated, desaturated, light, dark and natural colours. By trying to limit the extremes, you force yourself to spread your colour-choice.

1. With a brush

Just like preschool, take any fully opaque brush and start colouring in. This method of colouring flats is basic but tedious, and I don’t recommend it outside of cases where the lineart is really fuzzy and/or detailed. Often you’ll be using it alongside the other techniques here.

If you’re using Krita, I suggest looking at the “Experiment Brush” type for this. Also, to use the eraser in Krita you can press “e” on your keyboard or press the Eraser icon in the toolbar.

Advantages: Maximum control.

Disadvantages: Tedious. You don’t want to do large areas with this.

2. Selections

Basically this method is making a selection with the selection tools and then flood fill it with the preferred colour.

First, make a new layer and put it underneath the lineart layer.

The most interesting selection tool has a different name in every software but often similar icons: fuzzy-selection(Gimp), Contiguous selection(Krita name) or Magic wand(Photoshop name) is a tool that can select a large area of similar colours, which is great for colouring in lineart. To use them to do so, select the tool and then click on a closed area.

Gimp’s fuzzy selection tool icon.

Krita’s Contiguous selection tool icon.

Then go into Select-> Grow (selection), this is to make the selection expand past the anitalising(soft border) of the lineart.

Now select your colour of choice(for me brown for the hair) and hit Shift-backspace or (Ctrl+comma for gimp) to fill the selection with brown.

In Gimp, when using a selection tool you can press “Shift” to add to the current selection, and Ctrl to remove from the selection. There’s also buttons in the tool options that allow this. Using these with the freehand, scissors, rectangle, fuzzy and ellipse select, you can easily put together complex selections.

In gimp, whenever you have a selection, pressing the quick select-button(small selection rectangle in the lower-left corner of the upper screenshot)(or Shift+Q) toggles the quick-select mask. You can then paint with black to remove from a selection, or with white to add to a selection.

Pressing the quick select button again(now a red-rectangle)(or Shift-Q) brings you back to selection mode.

For Krita you don’t have these quick keys, and have to keep to the buttons in the tool options.

Krita does have brush select and path select, which are very useful for small details and organic shapes.

(For path select in gimp, you first need to draw a path, select the path and then go Select->from path)

Saving selections.

In both Krita and Gimp you can save the selection you made for later use. In Krita this is a layer type(named local selection) in Gimp you do this in the channel docker. (Remember how I said that programmers really liked to stretch the definition of “Channel”? Well this is one of them. Photoshop works similarly, and apparently this was the way artists subdivided an image before the layers-docker was introduced).

In Krita you can right-click any layer while having something selected, and add a “Local selection”

In gimp you can go Select->Save to Channel. Hitting the red rectangle will make the channel to a selection again.

Advantages: Quicker than just using the brush. Most programs can use selections.

Disadvantages: The magic wand/fill type selection tool isn’t great for lineart with a grainy look or lots of gaps. Selections are sometimes fiddly to maintain(sometimes clicking wrong can make you lose your whole selection) and there can be a lot of artifacts depending on the software.

3. Geometric tools

Some image manipulation programs have geometric tools. So you have a tool for making rectangles or circles. And in the case of Krita, a tool for bezier curves.

Select the polygon tool.

Set the tool options to fill=foreground and outline=none. Make sure that your opacity is set to 1.00(fully opaque)

Draw a polygon by clicking on points on the canvas. Finish it with Enter, Shift or by clicking the begin point.

The path tool is my favourite in this case, because it’s pretty easy to use for organic shapes.

The path tool is pretty similar to the polygon tool, but it has a certain advantage: by clicking and holding, you can influence how curvy a line draw with the path tool is going to be. Letting go of the mouse button confirms the action, and then you’re free to draw the next point.

However, due a bug you currently can’t set the outline of the path and freehand path to none. To work around this set your brush size to “0”.

You can also erase with a geometric tool. Just press “e” or the eraser button(in red below).

->

Advantages: Quicker than using the brush or selections. Also decent with lineart that contains gaps.

Disadvantages: Fiddly details aren’t easy to fill in with this. So I recommend skipping those and filling them in later with a brush.

4. Fill tool

fill tool in Krita

The Fill tool’s usefulness depends on the options available. In most cases the fill-tool can’t deal with the anti-aliasing(the soft edge in your lineart to make it more smooth when zoomed out) In Krita you have the grow-shrink option. Setting that to say… 2 expands the colour two pixels.

Threshold decides when the fill-tool should consider a different colour pixel to be a border. And the feathering adds a extra soft border to the fill.

Now, if you click on a gapsless-part of the image with your preferred colour…(Remember to set the opacity to 1.0!)

Depending on your lineart, you can do flats pretty quickly.

Setting the threshold low can result in little artifacts around where lines meet:

However, setting the threshold high can end with the fill not recognising some of the lighter lines. Besides these little artifacts can be removed with the brush easily.

Advantages: Pretty darn quick depending on the available settings.

Disadvantages: Again, not great with gaps or details.  A lot of software even has trouble getting through the anti-aliasing of the lineart.

5. G’MIC Colorize

G’MIC is a filter framework. A framework being programmer talk for a base program that is designed to be expanded on, and filter being programming ‘magic’(a bunch of difficult mathematical and logical operations done on an image). Indeed, G’MIC is as acronym for “Greyc’s Magic for Image Computing”. G’MIC is interesting because it does a lot of groundwork so that any programmer can with little effort make their own filters. This means that there’s a lot of interesting filters that come with G’MIC that may not exist in other programs.

G’MIC can be used independently as a commandline tool, but there’s also a Gimp plugin for it, and Krita will shortly gain G’MIC implementation as well. For our purpose, we’re mostly interested in the ‘Recolorize(Comics)’ filter, that is available from G’MIC 1.5.6.0 and up. This particular filter is based on the same algorithm as “Lazy Brush”-plugin for tv-paint.

In gimp, you first need to make a layer, and make colour spots on this one single layer.

Then you go to Filter->G’MIC(make sure you have it installed). Go to Black-and white and then Colorize[Comics].

Set the first drop-down to the left(Input) to “All”, then at layers ordering set input layers as “Lineart + Color spots” and output as “Lineart + Extrapolated colours” or “Lineart + Color Spots + Extrapolated colors” if you want to retain the colour spots layer. Set the smoothness high if you have a lot of gaps in your lineart, and low if you have a lot of details in your lineart.

Hit “OK”.

You should now have two(three) layers: One with Lineart and one with all the colours.

The colours intersect without antialiasing, so we’re going to use this to our advantage.

Use the “Select by colour” tool(Gimp) or “Similar Selection”(Krita) to select all pixels of one colour. Click on the face. You should now have all skin-tone selected. Press “Ctrl+x” and “Ctrl+v” to cut and paste the selection. In gimp you now have a floating selection, hit new layer to fix that. Do this for all colours, and you’ll have your lineart coloured with layer in no time.

Advantages: Works with anti aliased lineart. Really quick to get the base work done. Can auto-close gaps.

Disadvantages: No anti aliasing of it’s own. Putting the colours on separate layers is takes some extra effort. You have to choose between getting details right or the gaps auto-closed. Only few bits of software can use the G’Mic filters in the first place.

Saving your work.

Not all file formats can save layers or groups or any of the other functionality used here, even though it be really useful to save them. It’s therefore important that you make sure that the file format you choose can save your images and be opened in the image editor of choice. The following options are good for saving layers.

Photoshop file, (*.psd)

The most common layered file format due to Photoshop’s age and widespread use. Much hated by programmers because it is evil, but you’ll find most programs support it anyhow. Can save layers.

OpenExr(*.exr)

A fileformat developed for the visual effects industry, OpenExr is intended for storing high bitdepth images and can store layers. Probably too heavy for regular use.

Openraster(*.ora)

An attempt at an open-standard Layer format. MyPaint uses these as working format, and both Krita as Gimp can open the format as well.

Gimp layer format(*.xcf), Krita format(*.kra), Painttool Sai(.sai) etc.

Both are program working formats. All graphics software has their own working file format, which ensures that everything that the program can save can be saved in these files. Often these files can only be opened with the program it’s the working format of, but in the case of .xcf, you can open these in Krita as well.

The following formats have no layers but are a lot more lightweight than their layered cousins, making them useful for showing off on the internet. Each format has their own manner of compressing and is thus useful for different kind of images.

Tagged Image File Format(*.tiff)

An old and a little strange format that can do many things depending on the software saving and opening it. Sometimes, a software can save and open multilayered tiff files, but this is unfortunatly not the the norm. Interestingly you can choose the compression algorithm to save in, in many programs.

Portable Network Graphics(*.png)

A rather favoured fileformat, intended to bring high quality images to the internet. PNG can be compressed very well and is great for lineart, pixelart and greyscale images. However, with paintings and images with a lot of colour information it’s often still beaten by JPEG in terms of compression.

JPEG(*.jpg)

A fileformat intended for images with little contrast and large amount of colour information, such as photos. JPEG is good for paintings and photos, but not too great lineart and pixel art.

Bitmap(*.bmp)

Avoid using BMP. The thing is, the format is really old, and few programs offer the option to compress bmp. Due to this it has a really bad reputation for being heavy, and often it deserves it’s reputation as well. JPEG or PNG are better options.

Conclusions and after thoughts.

Doing the flats is very elementary work. So it’s important to do it right, but at the same time due it being such basic work there’s the temptation to rush it. Try to figure out which of these methods work best for you to simplify doing this part, so you can focus on the quality.

This was a very tool-focused tutorial, and in some cases I probably want to expand on the topic in a more theory-focused tutorial. However, for the purpose of getting lineart coloured this tutorial is necessary, and I wanted to brief you about the many different methods to do so, before moving onto more exciting (theoretical) topics.

Next week: Shading!

Sources

Blending modes: https://en.wikipedia.org/wiki/Blend_modes

G’MIC: http://gmic.sourceforge.net/

G’MIC colorize[comics]: Gimp chat thread about the colorize[comics] filter Original thread on lazybrush algorithm at the mypaint forum Timothée Giet showing off how the filter works

PSD: On psd being evil

JPG: http://en.wikipedia.org/wiki/Jpg

EXR: http://en.wikipedia.org/wiki/OpenEXR

PNG: http://en.wikipedia.org/wiki/Portable_Network_Graphics

TIFF: http://en.wikipedia.org/wiki/Tagged_Image_File_Format

Theory

The fancy physicist name for light is 'electromagnetic radiation', and there really isn't quite anything like it in the universe. Depending on the way you look at it, it's either made up from tiny particles(like sand, or more specifically, atoms), or waves(like water-waves, or sound).

When looked at in a similar manner to sound, light too can have a frequency. And with different frequencies the properties of light change:

Like your radio is capable of picking up radio waves and your computer capable of picking up wifi, your eyes are capable of picking up the visible-light frequencies. Humans can pick up the full range of frequencies trough method of our three types of colour cones. These cones are differently sensitive to different colours, the long cone for example is sensitive to red, yellow and green, while the middle cone is sensitive to yellow, green and blue. By measuring the different signal-intensities between the info these two cones give, our brain is capable of knowing where a colour ranges on the hue range.

On the above image, Red(700nm) would get a modest reaction from the Long cone, but no reaction from the Middle and Short cones.

Green(+-550nm) would no reaction from the short cones either, but a lot from both Middle and Long cones(most from middle)

Yellow(+- 620nm) would get an avarage reaction from the Middle cones, a large reaction from the long cones, but none from the Short cones.

Blue(+-450nm) would get a lot of reaction from the short cones, but hardly any from the Middle and Long cones.

Specifically, in the case of the purple hues, we actually make it up.

To make an efficient approximation of the billions of colours humans can see,  monitors and screens made by humans use red, green and blue to display images and store images.

And we're going to take advantage of that today.

Practical

If you've ever used a radio, you might know that you can tune into a specific frequency: A channel. Because radio and colours are both forms of light, you can tune into a specific frequency of colour as well. In an image manipulation program(colloqially known as a drawing program) specifically, you are looking at the data that tunes into a specific colour channel.

Now, you can use Photoshop for this, but I'll be using GIMP. Any image manipulation editor that supports channel-manipulation(they usually have a separate docker for this) will do however. In the case of Krita, which I'll be using in future tutorials whenever I can, the channel-manipulation tools aren't very extensive yet. However, if you scroll down, there's some explanation of how to do this in Krita.

First off. We take our image:

It's preferable you scan it in(the former image), photographed pictures tend to have a colour distortion(the latter). If you do take a picture, try doing so during the day when there's plenty of light, and try to make sure the picture is evenly lit.

Secondly, we open up the picture in our image manipulation program of choice(in my case, the GIMP).

We find out where our channels docker is. In my case it was tabbed next to the layer docker, and that tends to be it's default spot in many image manipulation programs, if you can't find it, try to look under under window->dockable dialogues.

So each of these channels represents a colour, and how much of that colour is present in the end image.

Okay, now the channels docker is a little complicate, but let's play with it a bit to get to know it. See how all three channels are selected and visible?

Ticking on them removes them from the selection, but doesn't make them invisible. Which means that if we deselect the green channel and paint on the image with yellow(remember, in light, red+green=yellow).

And if we paint with white(red+green+blue)

So what's happening now is that only red and blue can be added to the image because you only have those two channels selected.

You can see it as well when you hide the green and blue channels.

Okay, so that's how the channels docker works. Remove all paint-strokes you made, and select the green channel again.

You may have caught on, what we're going to do is edit these channels to remove the blue lines. Paradoxically, we'll need to edit the red and green channels, because those contain the differences between white(red+green+blue) and blue(only blue).

Deselect the blue channel:

Before heading toward the next step, make sure you have black as foreground colours and white as blackground colour. Hit the little black-square over the white square to make sure that is the case..

=>

With only the red and green channel selected, go to EDIT-> Fill with foreground color(or CTRL+comma; the hotkey). If you did that correctly, you've made the red and green channels completely black, and your image should be black to blue.

Now right click on the blue-channel, be careful you don't select it, and click 'channel to selection' or 'add to channel'

Now, with still only the red and green channels selected, go to Edit -> Fill with background color.(or CTRL+ period) and then, deselect everything!

Depending on the quality of your picture, you should have lost the blue lines completely or partially. The partially usually has to do with the colour of the lines being slightly different in that part of the picture.

Because we made all colour channels exactly alike, our picture is grayscale. (Seriously, you can check) So we'll solve those last lines with the contrast tools. While you can use the Brightness-contrast for this, it's best to use the Levels-tool for this.

I prefer the levels tool here because it's a little more intuitive to use. You can also use the curves for this, but it's a little trickier to use. You can find the Levels tool under Colors -> Levels

So, levels tool. The little graph that is visible over the upper-bar is an outline of how much of a specific value is used in a picture. In most black and white pictures it usually spikes around the lighter-values.

You can play with the lower-bar(output) to see what it does, but you likely want to leave it unchanged because our interest lies in the upper-bar(input).

Drag the white arrow to the left, just past the spike in the graph. You should be seeing the leftover-lines gradually disappear. Continue till they're gone. If necessary you can also drag the black arrow to the right, to even out the black-lines of your drawing.

Click done and save:

Tada! You've removed the blue lines from your drawing.

Now, you may have noticed this little box with "Value" in the Levels-tool and if you click it, you'll see that you can edit the red-green-blue channels here as well. Which leads us to the following part:

Practical - Part 2:

The faster method.

So, if we return back to our drawing.

Surprising how much we changed huh.

Now we go directly to the Levels-editor.

We set our base-contrast.

We select value and go to each channel to fine tune the contrast there. It should be slightly higher on the red and green channels than on the blue.

You will really need to fiddle and tweak with these settings to get what you want, but eventually, it should be possible to get this:

So, how to do this in Krita?

Well, Krita doesn't have the ability to add a colour channel to a selection, nor does it have the ability to edit them in the levels tool.

However, it does in the curves tool, which is under Filters -> Adjustment -> Color Adjustment Curves.

The steepness of the curve is the contrast. You'll have to figure it out for each channel, just like with levels.

Now, to rid the last bit of discolouration, go to Filters -> Adjustment -> Desaturnate.

Done!

Manner #3: More channels, more fun!

So, while initially the Channels were named as such because they represent a frequency, programmers have since hijacked the term for anything using a greyscale picture to store data, amongst which, a term you may have heard of: the alpha channel. Or rather, the transparency.

Supposedly, the alpha channel is named as such because it was represented as the letter α in the original equations for calculating an image with an alpha channel over another image. Because this alpha can be represented as a greyscale image, it was decided to encode it similarly to the colour channels. And thus the blurring of the term 'Channel' bagan.

There's a filter in both Gimp and Krita that allows you to make one colour in a picture completely transparent. We're going to use this filter in a method to set the colour we pick to be the only one that isn't transparent.

For this one I increased the contrast, because for my lineart it's important the black was fully black. However, if you are doing the same to a pencil piece, you don't have to do this, the method will work without :)

We'll be using the colour-to-alpha for this and the layers docker for this, you can find the windows->dockable dialogues. In GIMP this can be found under the Colors menu, in Krita under Filters->Colors->Color to alpha.

By default, colour to alpha is set to white, click the white bar to get a colour-selector and set it to black.

This should be the result.

Now, we're going to make a new layer and fill it with our foreground colour(set the foreground colour to black.) You can add layers on the layer docker or under layers-> new layer.

You should now have a full black layer.

Right click on the lower-layer, the original picture. Below in the rightclick menu, you should see in Gimp "Alpha to Selection" and in Krita "Select opaque". Click that.

Now select the upper layer, and hit the delete key. Hide the lower layer.

Tada! It's all transparent now.

Conclusion and afterthoughts:

I hope you learned something from this tutorial, even though it starts at a completely different place than you end up in. This is not the last tutorial I'll write about colours channels, or even RGB itself, and I hope this tutorial is a good preparation for those.

In the meantime, I hope you can put these convoluted but computer-scientifically correct ways of removing blue lines to good use.

Next time: Flat-colouring and the layer system.

Sources:

Physics behind light: https://en.wikipedia.org/wiki/Light and every modern schoolbook on physics. Also http://www.youtube.com/watch?v=Q_h4IoPJXZw and http://www.youtube.com/watch?v=DfPeprQ7oGc

Frequencies light: https://en.wikipedia.org/wiki/File:EM_spectrum.svg

Pink is an imaginary colour: http://www.youtube.com/watch?v=S9dqJRyk0YM

Definition channel: http://www.merriam-webster.com/dictionary/channel (f :  a band of frequencies of sufficient width for a single radio or television communication ) and http://dictionary.reference.com/browse/channel ( "a band of radio frequencies assigned for a particular purpose, esp the broadcasting of a television signal")

Dog's vision: http://en.wikipedia.org/wiki/Dog, http://dog-vision.com/

Colorblindness: http://en.wikipedia.org/wiki/Color_blindness

Tetrachomats and human lenses blocking UV-light: http://en.wikipedia.org/wiki/Tetrachromacy

Vision in butterflies: http://www.butterflyzone.org/butterfly-articles/butterfly-uv-vision.shtml

Mantis shrimp vision: http://theconversation.com/mantis-shrimp-have-the-worlds-best-eyes-but-why-17577 https://en.wikipedia.org/wiki/Mantis_shrimp

Meaning Alpha-channel: http://www.peachpit.com/articles/article.aspx?p=1617518&seqNum=6