#igcsdof faq

Hi!

I'll try and cover this in more detail in a video at some point, but in the meantime what you're probably running into is a result of the frame wait setting near the top.

You have the option to choose classic or fast. As it suggests, the fast option will render igcsdof more quickly than the classic option, but the classic option will often be sharper depending on other factors.

I try to use the fast option wherever I can, because I'm very impatient.

You can change the number of frames waited. For Cyberpunk I believe I have that set to 5. You'll know whether you have it set correctly as soon as you start rendering the image. If it's wrong, the in-focus areas will immediately be out of focus, to greater or lesser degree. If you manage to get this setting just right, the in-focus areas will be sharp.

If you're using classic mode, adding frames to wait will increase the render time. If you're using fast mode it generally doesn't as much.

If you're using any kind of upscaling, like DLSS, this is where you can still lose a bit of the sharpness, even if you've got the frame wait number as good as it can be.

So, play around with choosing either classic or fast, and then with how many frames to wait. Hopefully you'll be able to find a setting that works for you. It'll likely be different in each game.

Now, even knowing all of that, I still end up with soft details sometimes in areas that should be in-focus. So I do exactly what you describe. I take a shot without dof, and a shot with dof, and I blend them together in Photoshop afterwards.

The easiest way to do that is to open both pictures up in Photoshop, paste the blurred version as a new layer over the top of the focused version, create a layer mask on the blurred layer, and then use a paintbrush set to black to paint out on the layer mask the areas you want to be sharp. This will let the bottom layer show through. Doing it as a layer mask means it's non-destructive, so if you make a mistake you can switch to a white paintbrush and paint the top layer back in again.

I find this method really useful beyond just making details sharp again. Because of the nature of igcsdof, it's hard to control which areas are in or out of focus, and often more will be out of focus than we want. You can bring more into focus using the layer mask method, giving you more control.

Hi anon,

IGCSDOF doesn't work like a traditional DoF shader. It takes a little getting used to.

I'm guessing you've read the guide on Frans' website about IGCSDOF? So you should understand the basics from there. Make sure you have the companion IGCSDOF.fx shader activated in your load order. You want it to be at the bottom of your load order. The only thing below it should be film grain.

Basically what happens is:

choose whether you want fast or classic mode (if you're using TAA/DLAA/DLSS/FSR you'll want to use classic mode) and choose the number of frames to wait (for classic you'll probably want 4-5 to be safe, for fast you'll likely want 2 but you may need to experiment to see which produces the sharpest in-focus results)

decide what you want to be in focus, and turn on the magnifying box and position it over that part of the image

move the focus slider until the double images overlay each other perfectly and sharply in the area you want to be in focus

move the intensity slider (above the focus slider) until the part of the image you want to be blurred has its double images a distance away from each other - the further apart they are the more blurred it will be (this is the hardest part to get used to, but you'll quickly learn what you need to aim for)

readjust the focus slider if it became slightly misaligned when you adjusted the intensity

make any adjustments to quality and bokeh shapes (if you're making a test render to check the bokeh/blur intensity, set the quality to low at first so it renders quickly, and then change it to higher once you're happy)

start the rendering and wait for it to finish

take your shot

end the session

Depending on the mode (fast or classic) and quality you chose, the number of frames you wait, and the fps of your game, the rendering part can take anywhere from a few seconds to several long minutes. That's completely normal.

While it's rendering, it takes hundreds of separate temporary shots from different angles. This is why it looks like it's moving around on the screen. It then blends all those shots together into one final one, and that's when you take your picture.

Hi anon, Happy New Year!

Yes, definitely. IGCSDoF is a game changer. It produces a much smoother image than CineDoF, especially around the edges of hair and any bright objects at the edges of characters etc. And it also helps smooth hair out, which can look a bit crunchy depending on the type of AA and other shaders you use.

Hi anon!

Try to imagine the space in front of you in terms of linear distance from the camera to the farthest point in the distance.

When you focus, you're not focusing on a specific object per se, but rather on a particular distance between the camera and the horizon. Anything in the scene that is at that exact distance from the camera will be in focus, whether it's in the very centre of the screen, or off to the left or the right, top or bottom.

Anything you want to be in focus has to be at that distance.

Let's assume the camera is at 0 and the horizon is at 1, and half way between them is 0.5. If Tav is standing at 0.2 and Shadowheart is standing at 0.4 you can either focus at 0.2 or at 0.4 to have one of them in focus, but one of them will be out of focus. You can't set two different focus points because you'd end up with blur from both of them messing them both up.

There are a couple of ways to work around this, and as always with ReShade both involve compromise.

The first would be to change your composition so both characters are the same distance from the camera. That's nice and simple, but can sometimes ruin the composition you wanted.

The second is to adjust a bunch of stuff to mimic how a real camera creates blurred and in focus areas. Zoom (FoV) will affect how strong the blur is and how quickly it goes from in focus to out of focus. Theoretically, the closer you zoom in (the lower your FoV) the stronger the blur will become and the quicker it will start after the in focus area. Additionally, lowering the bokeh size in the IGCSDoF settings will also create a softer transition between the in focus area and the blurred areas. This can help create a wider zone of space that is in focus, meaning your characters may not have to be at the exact same distance away from the camera.

Unfortunately, because of the way IGCSDoF renders everything, that zone will still always be somewhat narrow, and, especially if you wanted rather strong blur in the background, you'll likely find yourself battling against some areas that aren't in focus that you wish were.

Luckily, there's a secret third way that can come to the rescue: Photoshop*.

I frequently find myself taking a shot with IGCSDoF, and one without, and then opening them in Photoshop to blend them together so I get what I want in focus. I do this a lot with close portraits where I want a decently strong background blur but find as a result only the eye is in focus and the nose and sometimes even the mouth is slightly out of focus. I layer the screenshots over one another, add a layer mask, and paint out the areas of blur I don't want so the in focus parts come through.

There are always going to be limitations with ReShade and screenshot tools more generally because it's trying to mimic real world behaviour in an artificial digital game space, so we have to get crafty sometimes!