Eevee - Objects in front of SSS (Subsurface Scattering) shaded objects cause bluish dispersion #65849

New Issue

KiJeon · 2019-06-16T11:46:03+02:00

KiJeon commented

2019-06-16 11:46:03 +02:00

System Information
Windows 10 64-bit

Blender Version
Broken: 2.80 June 16, 01:02:51 - 87de71a8aaa

Short description of error
In Eevee, any objects occluding SSS shaded objects has bluish dispersion appearing at the point of occlusion.

Here's an example:

At first I thought it was a hair problem, but it's any object:

Exact steps for others to reproduce the error
Should be easily reproducible by setting the values for SSS radius to be 1, 0, 0, and placing an object in front.

Also attaching the example file:2.8 MonkeyHairTest.blend

**System Information** Windows 10 64-bit **Blender Version** Broken: 2.80 June 16, 01:02:51 - `87de71a8aaa` **Short description of error** In Eevee, any objects occluding SSS shaded objects has bluish dispersion appearing at the point of occlusion. Here's an example: ![image.png](https://archive.blender.org/developer/F7105616/image.png) ![image.png](https://archive.blender.org/developer/F7105618/image.png) At first I thought it was a hair problem, but it's any object: ![image.png](https://archive.blender.org/developer/F7105622/image.png) ![image.png](https://archive.blender.org/developer/F7105624/image.png) **Exact steps for others to reproduce the error** Should be easily reproducible by setting the values for SSS radius to be 1, 0, 0, and placing an object in front. Also attaching the example file:[2.8 MonkeyHairTest.blend](https://archive.blender.org/developer/F7105626/2.8_MonkeyHairTest.blend)

KiJeon commented

2019-06-16 11:46:03 +02:00

Added subscriber: @0o00o0oo

blender-admin commented

2019-06-16 11:46:03 +02:00

#71608 was marked as duplicate of this issue

Clément Foucault self-assigned this 2019-06-19 23:01:41 +02:00

Clément Foucault commented

2019-06-26 13:59:47 +02:00

This is a limitation of eevee's SSS. The occluders have infinite thickness.

Clément Foucault commented

2019-06-26 13:59:57 +02:00

Changed status from 'Open' to: 'Archived'

Clément Foucault closed this issue

2019-06-26 13:59:57 +02:00

KiJeon commented

2019-06-26 21:23:14 +02:00

@fclem Is there a good way around this problem? Or are there plans to remedy it?

As it stands, in situations like this, the results are sadly unusable.

@fclem Is there a good way around this problem? Or are there plans to remedy it? As it stands, in situations like this, the results are sadly unusable.

Clément Foucault commented

2019-06-27 01:36:39 +02:00

We could add a thickness option for the SSS, similar to the one the SSR have.

Jean Da Costa commented

2019-06-27 01:51:55 +02:00

Added subscriber: @jeacom

Jean Da Costa commented

2019-06-27 01:51:55 +02:00

This is a terrible problem, it'll make SSS unviable for characters.

Jean Da Costa commented

2019-06-27 02:02:22 +02:00

@fclem I'm not into shaders but isn't there a way to mask out the non-sss material pixels influence on the blurring kernel?

Edit:
It seems actually like they are already being masked but if I get it right the more non-sss pixels the blur kernel has to ignore, the darker is the resulting pixel, maybe if you normalize the resulting color relative to the amount of SSS pixels in the kernel?

@fclem I'm not into shaders but isn't there a way to mask out the non-sss material pixels influence on the blurring kernel? Edit: It seems actually like they are already being masked but if I get it right the more non-sss pixels the blur kernel has to ignore, the darker is the resulting pixel, maybe if you normalize the resulting color relative to the amount of SSS pixels in the kernel?

Clément Foucault commented

2019-06-27 03:19:46 +02:00

@jeacom We cannot do this because we cannot distinguish between pixels that are unlit from pixels comming from another object at this point in the pipeline. The lighting is masked before doing the SSS passes.

But the idea is good and that's what we will eventually do if we can in the future.

@jeacom We cannot do this because we cannot distinguish between pixels that are unlit from pixels comming from another object at this point in the pipeline. The lighting is masked before doing the SSS passes. But the idea is good and that's what we will eventually do if we can in the future.

Jean Da Costa commented

2019-06-27 20:34:17 +02:00

This comment was removed by @jeacom

*This comment was removed by @jeacom*

Jean Da Costa commented

2019-07-02 16:14:15 +02:00

@fclem just noticed that if we enable SSS on the occluder's material and set the SSS radius to 0, the problem goes away, presumably because the lighting mask gets disabled on the occluders?

Clément Foucault commented

2019-07-02 16:17:02 +02:00

Because the lighting does "bleed" from the occluder to the SSS object. This becomes a problem when the occluder recieves more light than the occludee.

Jean Da Costa commented

2019-07-02 16:45:52 +02:00

In #65849#711500, @fclem wrote:
Because the lighting does "bleed" from the occluder to the SSS object. This becomes a problem when the occluder recieves more light than the occludee.

oh, I see.
So the color is being applied before the blur, not after.
If I enable separate albedo, the light bleeding disappears tho.
At least it looks viable to disable the light mask if sepparate albedo is enabled, this would fix the problem at the cost of the texture blur right?

I personally think the result looks better with separate albedo enabled anyway.

> In #65849#711500, @fclem wrote: > Because the lighting does "bleed" from the occluder to the SSS object. This becomes a problem when the occluder recieves more light than the occludee. oh, I see. So the color is being applied before the blur, not after. If I enable separate albedo, the light bleeding disappears tho. At least it looks viable to disable the light mask if sepparate albedo is enabled, this would fix the problem at the cost of the texture blur right? I personally think the result looks better with separate albedo enabled anyway.

Jean Da Costa commented

2019-07-02 17:14:13 +02:00

I am convinced that if separate albedo is on, the best is to not mask the lighting.

![image.png](https://archive.blender.org/developer/F7562134/image.png) ![image.png](https://archive.blender.org/developer/F7562137/image.png) I am convinced that if separate albedo is on, the best is to not mask the lighting.

Germano Cavalcante commented

2019-11-20 21:44:59 +01:00

Added subscriber: @Stober-3

KiJeon commented

2019-11-21 07:59:58 +01:00

It's unfortunate this task was closed as "invalid". Shouldn't it be left open until a solution can be found?

As it stands, it's a clear visual problem. I'm afraid being marked as "invalid" simply buries the issue and prevents a prompt to resolve it (unless this limitation is noted somewhere as problematic).

It's unfortunate this task was closed as "invalid". Shouldn't it be left open until a solution can be found? As it stands, it's a clear visual problem. I'm afraid being marked as "invalid" simply buries the issue and prevents a prompt to resolve it (unless this limitation is noted somewhere as problematic).

Raimund Klink commented

2020-12-03 13:08:33 +01:00

Added subscriber: @Raimund58

Jon Henry M. Faltis 9077149 commented

2020-12-21 17:44:09 +01:00

Added subscriber: @geocentric_wage

Jon Henry M. Faltis 9077149 commented

2020-12-21 17:44:47 +01:00

Removed subscriber: @geocentric_wage

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Eevee - Objects in front of SSS (Subsurface Scattering) shaded objects cause bluish dispersion #65849