.

I remember like, years ago watching this video by Keinan Lafferty (I think) on youtube, and used the 'banana method' or whatever he called it in that, which is essentially making a curved 'banana' shape on your colour picker

essentially, the idea is, using your base colour (or, mid tone), you progressively shift the colour lighter and more de-saturated for your lighter tones, and inversely shift progressively darker and more de-saturated from your mid-tone for your shadows.

You can see in my colour-picked example here that it is not a perfect semi-circle, but it was generally the basis for how I'd pick colours, and then would tweak them from there.

How I'd tweak the colours usually is with the help of overlaying gradient maps. I'd also tweak colours in especially shadows, where I'd intentionally use a slightly lighter grey for the illusion of 'cooler' tones in shadows, and darker more saturated tones along the edges where the shadow meets the light.

I've more recently started trying out using triadic colour schemes, where I pretty much jump to the opposite side of the colour wheel for my shadows and light.

This image on twitter sort of sparked my inspiration to experiment more with that (image is linked to the original post) but I am still trying to get the hang of it still 🥲

Black Menswear modelled by Black Men

Flexible feather armor

The Tube Post has been updated for 2026! Endless thanks to @andstuffsketches for the amazing art (!!!), @rowenabean for checking this over for me (all remaining mistakes are my own), @kelpforestdwellers for extra tips, and @appleyblam for an editing pass. Thank you all!

which is cool. and it would be cool to say, find a granny square cardigan pattern with 24 squares, knit these squares, make a sick cardigan. but then i realized i don't know how to knit or anything. and once you accept that there isn't really a clear "application" and this concept lives on a screen, you open yourself up to more possibilities. a la birth of venus.

step 1: python script that looks at the original image and generates an excel spreadsheet the same dimensions (793 x 1322 pixels = 793 x 1322 cells), and each cell is populated with the hex code of the color that appears in that pixel of the original image

step 2: excel macro to generate list of every unique hex code that appears in the excel spreadsheet.

step 3: excel macro to calculate the R, G, B values of each of those hex codes.

step 4: excel macro to fill each cell with the color of that hex code (not necessary, i just like to do it).

step 5: I add in Saturation (the difference between the largest and smallest RGB value) and Lightness (average of all RGB values).

step 6: pick a color palette. i always find myself gravitating towards groovy seventies palettes with warm reds and oranges, so i decided not to do that this time. i looked on coolors and found a color palette that was all dark greens that were similar to each other. there were only like four colors or something in this palette. and to make it truly different from the other project, there should be a small gradient. so i determined the smallest possible change between colors and used an excel macro to color it. i was going to stop here and do the entire image in shades of green (inspired by that guy on tiktok that paints using only one color) but then. idk. i realized the "opposite" of each color was an equally subtly changing pink. so i imagined that the end of this process would be an "abstract" image, with subtle variations of pink and green, that would end up suggesting birth of venus.

so all told, i had 502 unique replacement colors, 251 of which are green, 251 of which are pink. (793 x 1322) / 502 = either 2088 or 2089 of each color.

step 7: find some method for finding the difference between the original colors of the image and my new color palette. I use a method of comparing, R, G, B, S and L:

((abs(R1 - R2) + abs(G1 - G2) + abs(B1 - B2)) / 3) + abs(S1 - S2) + abs(L1 - L2)

and you come up with something like this. on the left, those are colors that appear in the original image. across the top, those greens are the colors i'm replacing it with. in blue, that's the number of each new color i have to work with (it's just blue for contrast). and in the center, this pink area, that's a giant spreadsheet with the "objective" difference between each original color and each replacement color. it's pink because i have some conditional formatting applied, ignore that part.

and in this situation, you have some choices to make. in the original image up there, i used a schema prioritizing light and dark--i.e., i looked at the darkest color (pure black) that appeared in the original image, then found the closest replacement color (i.e., the replacement color with the smallest number). then did the same with the lightest color. then the next darkest, next lightest.

but i'm going to do it slightly differently this time. and i don't know how this image will come out looking.

if you look at the "first" green, closest to the left, and sort by smallest to largest:

you can see that these colors on the left are closest to the "first" green i've decided to work with. that might seem odd. i mean, #7F9800--> #00a94f are pretty close, but #A95400 is red. but that's just a difference in hue. really, #A95400 and #00a94f are very similar in lightness and saturation.

and this also calculates the number of times that color actually appears in the original image. that first specific green, #7F9800, only appears twice. but some colors, like actual black #000000, appear something like 46,000 times. and if you add all the numbers in the "frequency" column, it should exactly equal the sum of each replacement color (2088 ish x 502).

step 8: excel macro again. this one is complicated. basically it sorts that first "green" column (column E in my spreadsheet) from smallest to largest. then it adds each cell in the "frequency" column until it reaches or surpasses the blue cell above column E, which for this particular color is 2089. it copies those "original image" colors and their respective frequencies over to another sheet. for the color that surpassed 2089, it splits in two. then it deletes that column E. Then it makes sure "frequency" and "replacement color sum" still total. then it runs again on the new column E, until the whole spreadsheet is used up. and it generates something like:

[color from original image] [number of times that color appears] [replacement color, filled in]

and there's approximately 8000 lines of that.

i have the replacement colors in the order above. starting with vivid green, slowing transitioning to dark green, switching abruptly to bright pink, slowly transitioning to pale pink.

step 9: another excel macro. this one looks at original image broken down into hex codes, then looks at the generated list and replaces each [original] color with the replacement color, that exact number of times.

end result of these macros, following different "rules" of assigning replacement colors to original colors, is this:

which looks different, obviously. but it is the exact replacement colors, and same number of each replacement color, as the original up there.

The blurb said the shot was taken in Granada, Spain, by Agneta Fondén.  No other info, so I have no idea how old, etc.  There's a game (from 1988) that uses a similar pattern on one of its pieces, but this could predate it by at least a thousand years  — or not.  But the pattern intrigued me, so I made a texture map and used Blender's geometry nodes (no generative AI) to set up a hexagon grid with random rotations for the tiles:

That's all done with a single design:

You'd think this would have a name, right?  (For its historical use as an architectural / decor tile — although I've found out more about its use in games, that's not what I'm looking for.)  Like the Penroses do (and no, it's not one of those).  But I've had no luck finding it, or any other info.  Any (human only, please) help?

Hi Oliver! Between the two of us YOU are the one with a debut graphic novel published so I'm going to go ahead and link to that lol. It's Space Story!

Okay, here's the post with the Lynda Barry exercises. I can't recommend Making Comics enough for idea generation & general comics motivation. Barry is a master at getting you to loosen up and then tighten back up. I've also gotten a lot of mileage out of this Before/After worksheet from Tom Hart. The random image can be anything (movie screencaps, memes you've saved, other people's comic panels, etc).

Lately I've been working on Junk World #2 which is really long. I don't worry about sticking to a daily routine because my day job is physically taxing. But I do at this point have a set "formula" for finishing pages that feels very natural & doesn't stress me out.

First (image 1) I doodle a scene with ZERO regard for panel layout or page composition. This kind of drawing is just little guys talking to each other. Like literally floating heads. Everyone can do this kind of thing, I did it in the margins of my school notebooks. Then (image 2) I go back & start chopping up the story beats into panels, pages, etc. I only do this 1-2 pages at a time. Any more and I get bored!!!

Then I lay out the panels on 11x14 bristol and pencil/letter/ink. This is a "draw the rest of the owl" ass thing to say and it takes the most time, but because I've worked out most of the problems in sketch form it doesn't feel stressful or like I'm just staring at a blank page.

I think this is different from the standard "thumbnail" advice because the first part, the part where I'm just doodling/playing around, is fundamental to the actual flow & feel of the story. I can't skip it. And it's an extension of all my favorite exercises where the focus is generating silly images/characters just for fun and not worrying about the finished Thing.

But I don't know if that answers your question really... TL;DR:

In order to get in the groove, I pretend there's no causal relationship between "doodling a little cartoon animal" and "crafting something purposeful that people will see." And the core of that is getting to a place where I can just make stuff up without worrying about it :D

its kind of the opposite for me, my roughs look so ass😭 but i learned that its okay because each new pass is an opportunity to improve things and fix mistakes

its always best not to overthink things and just plow through the work, at least for me bc i know if i stop to plan, nothing will ever get finished

also if youre worried with about your cleanup being too stiff, the solution ive found for myself is to have a bigger gap between it and tiedown, so youre not just tracing over the same lines. its more tricky tho because you have to figure out a lot of things like perspective and volume while also drawing clean, but this way your linework will be more intentional

to make it easier to maintain consistency and good spacing, animate body parts (or even parts of body parts) separately, like go through the entire movement of the bottom side of the elbow, then return to frame one and do the top side, etc (you can kind of see it in the video)

also, if something is bothering you, like if a drawing seems kinda off, or you feel like the animation is jittery in some spot, my advice is to get in there and fix it! even if it seems like its gonna be too much unnecessary work. at least for me, i know if something is bothering me, it will bother me every time i watch the animation, so its better to fix it than to have an artwork i cant look at.

Some older examples of commissions!