GPU renders have warmer colors than CPU renders. #47812

Closed
opened 2016-03-16 01:41:35 +01:00 by Mike · 23 comments

System Information
2013 MacPro 8-core, 64GB RAM
Titan X

Blender Version
2.77 rc2

I also noticed that rendering this fire on the GPU in 2.77 has a different color tint than the CPU renders in 2.77 or other versions. The colors are slightly warmer/more yellow on the GPU renders in 2.77. Our renderfarm is a mix of CPUs and GPUs so this would be a problem for us, making a large portion of our renderfarm useless if we have to choose GPU rendering or CPU rendering to ensure the look is consistent. Ive attached 3 frames, with the middle frame (#623) being the one rendered by the GPU.

SANCT_XX - Flyover AoS_0622.png

SANCT_XX - Flyover AoS_0624.png

SANCT_XX - Flyover AoS_0623.png

**System Information** 2013 MacPro 8-core, 64GB RAM Titan X **Blender Version** 2.77 rc2 I also noticed that rendering this fire on the GPU in 2.77 has a different color tint than the CPU renders in 2.77 or other versions. The colors are slightly warmer/more yellow on the GPU renders in 2.77. Our renderfarm is a mix of CPUs and GPUs so this would be a problem for us, making a large portion of our renderfarm useless if we have to choose GPU rendering or CPU rendering to ensure the look is consistent. Ive attached 3 frames, with the middle frame (#623) being the one rendered by the GPU. [SANCT_XX - Flyover AoS_0622.png](https://archive.blender.org/developer/F293135/SANCT_XX_-_Flyover_AoS_0622.png) [SANCT_XX - Flyover AoS_0624.png](https://archive.blender.org/developer/F293137/SANCT_XX_-_Flyover_AoS_0624.png) [SANCT_XX - Flyover AoS_0623.png](https://archive.blender.org/developer/F293136/SANCT_XX_-_Flyover_AoS_0623.png)
Author

Changed status to: 'Open'

Changed status to: 'Open'
Mike self-assigned this 2016-03-16 01:41:35 +01:00
Author

Added subscriber: @LMProductions-1

Added subscriber: @LMProductions-1

Added subscriber: @ThomasDinges

Added subscriber: @ThomasDinges

Please attach a blend file demonstrating the issue, as clearly asked for in the bug report guidelines! Also it would be more helpful to have the same frame number from CPU and GPU, this way we can directly compare the image.

Please attach a blend file demonstrating the issue, as clearly asked for in the bug report guidelines! Also it would be more helpful to have the same frame number from CPU and GPU, this way we can directly compare the image.
Author

Sorry. I thought maybe you guys would've heard of this before and therefore might be able to explain it from just the images. Ive attached the .blend file.

I will also attach a .png of frame 623 rendered with the CPU when its done rendering.

Thanks for the help!

SANCT_XX - Flyover AoS.blend

Sorry. I thought maybe you guys would've heard of this before and therefore might be able to explain it from just the images. Ive attached the .blend file. I will also attach a .png of frame 623 rendered with the CPU when its done rendering. Thanks for the help! [SANCT_XX - Flyover AoS.blend](https://archive.blender.org/developer/F294238/SANCT_XX_-_Flyover_AoS.blend)
Author

Here is the rendered CPU image of frame 623.623 (CPU render in 2.77).png

Here is the rendered CPU image of frame 623.[623 (CPU render in 2.77).png](https://archive.blender.org/developer/F294467/623__CPU_render_in_2.77_.png)
Author

Any thoughts after seeing the .blend or the images Thomas?

Any thoughts after seeing the .blend or the images Thomas?
Mike was unassigned by Aaron Carlisle 2016-03-20 19:24:43 +01:00
Author

I see this case is assigned to "none" and priority is "incomplete." Is this bug submission a lost cause?

I see this case is assigned to "none" and priority is "incomplete." Is this bug submission a lost cause?

Added subscriber: @mib2berlin

Added subscriber: @mib2berlin

Hi Mike, I take the freedom and edit your file to get it better handled on slow connections and so.
The color error ist still visible but render now much faster to check.
Can you please check on the file?
Btw., the image ashes.jpg is missing in your .blend but does not really matter.

SANCT_XX - Flyover AoS_edit.blend

The switch to "Incomplete" was before you add your file, an admin has to change back.

I can verify the color difference between GPU and CPU.

Blender 2.77
Opensuse Leap 42.1 x86_64
Intel i5 3570K
GTX 760 4 GB /Display card
GTX 670 2 GB
Driver 361.28

Mib

Hi Mike, I take the freedom and edit your file to get it better handled on slow connections and so. The color error ist still visible but render now much faster to check. Can you please check on the file? Btw., the image ashes.jpg is missing in your .blend but does not really matter. [SANCT_XX - Flyover AoS_edit.blend](https://archive.blender.org/developer/F297656/SANCT_XX_-_Flyover_AoS_edit.blend) The switch to "Incomplete" was before you add your file, an admin has to change back. I can verify the color difference between GPU and CPU. Blender 2.77 Opensuse Leap 42.1 x86_64 Intel i5 3570K GTX 760 4 GB /Display card GTX 670 2 GB Driver 361.28 Mib

We get a lot of reports every day. We try to handle them quickly, but sometimes it just takes time. Please have some patience. ;)

We get a lot of reports every day. We try to handle them quickly, but sometimes it just takes time. Please have some patience. ;)
Author

In #47812#365240, @mib2berlin wrote:
Hi Mike, I take the freedom and edit your file to get it better handled on slow connections and so.
The color error ist still visible but render now much faster to check.
Can you please check on the file?
Btw., the image ashes.jpg is missing in your .blend but does not really matter.

SANCT_XX - Flyover AoS_edit.blend

The switch to "Incomplete" was before you add your file, an admin has to change back.

I can verify the color difference between GPU and CPU.

Blender 2.77
Opensuse Leap 42.1 x86_64
Intel i5 3570K
GTX 760 4 GB /Display card
GTX 670 2 GB
Driver 361.28

Mib

Thanks Wolfgang. I'm not sure what you did to try to make it faster aside from removing the grate on the model? I rendered your .blend on my CPU at 100 samples as you had it, and it took 1min and 12 seconds. On the GPU it took 7mins and 21 seconds. So 7x slower on the GPU still.

> In #47812#365240, @mib2berlin wrote: > Hi Mike, I take the freedom and edit your file to get it better handled on slow connections and so. > The color error ist still visible but render now much faster to check. > Can you please check on the file? > Btw., the image ashes.jpg is missing in your .blend but does not really matter. > > [SANCT_XX - Flyover AoS_edit.blend](https://archive.blender.org/developer/F297656/SANCT_XX_-_Flyover_AoS_edit.blend) > > The switch to "Incomplete" was before you add your file, an admin has to change back. > > I can verify the color difference between GPU and CPU. > > Blender 2.77 > Opensuse Leap 42.1 x86_64 > Intel i5 3570K > GTX 760 4 GB /Display card > GTX 670 2 GB > Driver 361.28 > > Mib Thanks Wolfgang. I'm not sure what you did to try to make it faster aside from removing the grate on the model? I rendered your .blend on my CPU at 100 samples as you had it, and it took 1min and 12 seconds. On the GPU it took 7mins and 21 seconds. So 7x slower on the GPU still.

Hi, just for info, alongside Samples and Dimension I change Step Size to 0.5 in Volume Sampling (Please check 0.5 - 2).
I remove objects/meshes not involved in render time/color error to shrink size of .blend to 1/3 of original file.

Mib

Hi, just for info, alongside Samples and Dimension I change Step Size to 0.5 in Volume Sampling (Please check 0.5 - 2). I remove objects/meshes not involved in render time/color error to shrink size of .blend to 1/3 of original file. Mib
Author

I see there has been no activity here for awhile. Just dropping this post to keep the issue fresh. I know a lot of bugs get submitted every day. I did a render last night of a scene with a fire & smoke, and the smoke is a different color on each frame rendered with the GPUs than with the CPUs.

Im using 2.77a now. Issue persists.

I see there has been no activity here for awhile. Just dropping this post to keep the issue fresh. I know a lot of bugs get submitted every day. I did a render last night of a scene with a fire & smoke, and the smoke is a different color on each frame rendered with the GPUs than with the CPUs. Im using 2.77a now. Issue persists.

Added subscriber: @Sergey

Added subscriber: @Sergey

Can see the issue with the new file now. Will investigate.

Can see the issue with the new file now. Will investigate.
Sergey Sharybin self-assigned this 2016-04-19 17:21:46 +02:00

Added subscriber: @bliblubli

Added subscriber: @bliblubli

Not sure this is related, but in some of my test, NVidia and AMD gave different results regarding max bounces. It looked like NVidia did trigger the russian roulette much faster than AMD (AMD renders where exactly like the CPU ones). Nvidia did some driver side optimisation for Luxmark, so it may be the same for Cycles. So the bug may not be on Blender's side.

Not sure this is related, but in some of my test, NVidia and AMD gave different results regarding max bounces. It looked like NVidia did trigger the russian roulette much faster than AMD (AMD renders where exactly like the CPU ones). Nvidia did some driver side optimisation for Luxmark, so it may be the same for Cycles. So the bug may not be on Blender's side.

This issue was referenced by 4cdd6b9bdd

This issue was referenced by 4cdd6b9bddac37d8a482cbf44cf4c94fb4d10135

Changed status from 'Open' to: 'Resolved'

Changed status from 'Open' to: 'Resolved'
Author

I see this issue is marked "resolved" "by committing 4cdd6b9bdd: Fix #47812: GPU renders have warmer colors than CPU renders"

Does that mean the issue is fixed? If so will it be in the next release?

Thanks! =)

I see this issue is marked "resolved" "by committing 4cdd6b9bdd: Fix #47812: GPU renders have warmer colors than CPU renders" Does that mean the issue is fixed? If so will it be in the next release? Thanks! =)

The fix will be in tomorrows daily build (builder.blender.org) and in the next official release.

The fix will be in tomorrows daily build (builder.blender.org) and in the next official release.

This issue was referenced by blender/cycles@7111ce6b61

This issue was referenced by blender/cycles@7111ce6b611e962f0c66655fdb13c58df496d70b
Sign in to join this conversation.
No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset Browser
Interest
Asset Browser Project Overview
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
Interest
EEVEE & Viewport
Interest
Freestyle
Interest
Geometry Nodes
Interest
Grease Pencil
Interest
ID Management
Interest
Images & Movies
Interest
Import Export
Interest
Line Art
Interest
Masking
Interest
Metal
Interest
Modeling
Interest
Modifiers
Interest
Motion Tracking
Interest
Nodes & Physics
Interest
OpenGL
Interest
Overlay
Interest
Overrides
Interest
Performance
Interest
Physics
Interest
Pipeline, Assets & IO
Interest
Platforms, Builds & Tests
Interest
Python API
Interest
Render & Cycles
Interest
Render Pipeline
Interest
Sculpt, Paint & Texture
Interest
Text Editor
Interest
Translations
Interest
Triaging
Interest
Undo
Interest
USD
Interest
User Interface
Interest
UV Editing
Interest
VFX & Video
Interest
Video Sequencer
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Blender 2.8 Project
Legacy
Milestone 1: Basic, Local Asset Browser
Legacy
OpenGL Error
Meta
Good First Issue
Meta
Papercut
Meta
Retrospective
Meta
Security
Module
Animation & Rigging
Module
Core
Module
Development Management
Module
EEVEE & Viewport
Module
Grease Pencil
Module
Modeling
Module
Nodes & Physics
Module
Pipeline, Assets & IO
Module
Platforms, Builds & Tests
Module
Python API
Module
Render & Cycles
Module
Sculpt, Paint & Texture
Module
Triaging
Module
User Interface
Module
VFX & Video
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Priority
High
Priority
Low
Priority
Normal
Priority
Unbreak Now!
Status
Archived
Status
Confirmed
Status
Duplicate
Status
Needs Info from Developers
Status
Needs Information from User
Status
Needs Triage
Status
Resolved
Type
Bug
Type
Design
Type
Known Issue
Type
Patch
Type
Report
Type
To Do
No Milestone
No project
No Assignees
6 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: blender/blender#47812
No description provided.