BF Blender: RegressionsProject
ActivePublic

Details

Description

2.79 Release

Project to tag bugs that are release blockers and other tasks that are planned to be included in the upcoming Blender release.

Revisions to be ported to the release branch
(range XXXX..XXXX, sorted in order of cherry-pick)


For 2.79rc3/release

SHA1XXX
SHA1XXX

From @Campbell Barton (campbellbarton) in range (2ff9c8a3bc51142a1d7ad19d8b772b58e4dcc890..3789ab4166c40d254ab1cf9f07d429730650efdb)

  • 2ff9c8a3bc51142a1d7ad19d8b772b58e4dcc890
  • f6825d333bfe2d58227e5e7421cd8719d89cd59a
  • 361c7cbbc57720db3b04698c7e45fe72297e2b24 & 592dd9dea6aa6e8f954e7604b497bdd9c2387e21
  • f09dee5aed35ede00bdbaedba171ce24edf5e0b0
  • 46b9f89f5e46109ea811ccf474acb09616e7c33b
  • b8d77c44f112cae1e98308ff33ce055ea4fb46ff
  • 55861cb2346bdb4b1346c6999b99ff00f27e7d4b
  • 3277bd4031d08f750526168a50e1856d2936cf69
  • 76d695f7611dfe3f2560525771399008f0bf9f40 & 0b5b464e823660f4ce47107d63004e762372122e & e47629808404cd03019ed6de2c5d6fd5307630fa
  • fa365d5f2f9d2e4bbe0b5c7800093723c7d09de6
  • b6b27b067113184093c3023ac7c9bf9a6ad5914b & b07dcb8fb06f622c8521878df6cbef01d98d6cc4
  • 0b5cabdca5dfcc61bb06b0edc1a55c44d041b615
  • cb7f1f8160606e92a4f6d216f03254e870b8eca6
  • 2ca1f297486804dcdb90c376e68dce656a73c726
  • 3750389ce3f3b6b386986c33608d6aed348e6f7e ???

From @Joerg Mueller (nexyon)


Candidates for review:

Recent Activity

Mon, Feb 19

Sergey Sharybin (sergey) closed T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node as Archived.

Discussed with other developers about ways to process here.

Mon, Feb 19, 2:49 PM · BF Blender: Regressions, BF Blender
Sergey Sharybin (sergey) closed T53032: Image sequence works bad with Displace modifier as Resolved by committing rBdc65a6f63b54: Fix T53032: Image sequence works bad with Displace modifier.
Mon, Feb 19, 2:37 PM · Images & Movies, Rendering, BF Blender: Regressions, Modifiers, BF Blender

Mon, Jan 29

Campbell Barton (campbellbarton) merged T53941: 3D Manipulator Bug into T53850: 3D manipulators unselectable while “Lock to Cursor” is turned on.
Mon, Jan 29, 8:09 AM · BF Blender: Regressions, BF Blender

Wed, Jan 24

Brecht Van Lommel (brecht) closed T53886: cycles regression: increased noise in overlapping transparent shaders as Archived.

I'm aware of this, it's an intentional trade-off we are making. With Russian roulette you get more noise in simple cases, but it helps in more complex cases like transparent hair. The two are difficult to distinguish for the renderer and it's generally not helpful to do so in real world scenes where you have a mix of these.

Wed, Jan 24, 7:36 PM · Cycles, BF Blender
Philipp Oeser (lichtwerk) added a project to T53886: cycles regression: increased noise in overlapping transparent shaders: BF Blender: Regressions.
Wed, Jan 24, 7:19 PM · Cycles, BF Blender

Jan 22 2018

Campbell Barton (campbellbarton) added a comment to T53850: 3D manipulators unselectable while “Lock to Cursor” is turned on.

Fabien, you need to recompile blender - best build blender-v2.79a-release, this will be the official stable release soon.

Jan 22 2018, 10:25 AM · BF Blender: Regressions, BF Blender
Fabien (CaptainSweet) added a comment to T53850: 3D manipulators unselectable while “Lock to Cursor” is turned on.

Thanks for your help
How can I apply the fix? Can I simply copy and paste fixed files?

Jan 22 2018, 10:11 AM · BF Blender: Regressions, BF Blender
Campbell Barton (campbellbarton) closed T53850: 3D manipulators unselectable while “Lock to Cursor” is turned on as Resolved by committing rBfb4582134ae1: Fix T53850: Lock to Cursor breaks 3D manipulators.
Jan 22 2018, 10:03 AM · BF Blender: Regressions, BF Blender
Campbell Barton (campbellbarton) triaged T53850: 3D manipulators unselectable while “Lock to Cursor” is turned on as Confirmed priority.

Found the cause, looking into a fix.

Jan 22 2018, 7:57 AM · BF Blender: Regressions, BF Blender

Jan 21 2018

Campbell Barton (campbellbarton) closed T53843: RMB menu Online Manual call disfunctional, throws Attribute Error as Resolved by committing rBd9c962a36780: Fix T53843: Error opening online manual.
Jan 21 2018, 11:53 PM · BF Blender: Regressions, BF Blender
Fabien (CaptainSweet) created T53850: 3D manipulators unselectable while “Lock to Cursor” is turned on.
Jan 21 2018, 8:45 PM · BF Blender: Regressions, BF Blender

Jan 20 2018

Philipp Oeser (lichtwerk) assigned T53843: RMB menu Online Manual call disfunctional, throws Attribute Error to Campbell Barton (campbellbarton).

there might be better ways to do this, but how about D3013: possible fix for T53843?

Jan 20 2018, 4:11 PM · BF Blender: Regressions, BF Blender
Philipp Oeser (lichtwerk) added a comment to T53843: RMB menu Online Manual call disfunctional, throws Attribute Error.

caused by rBf4fb329808a7

Jan 20 2018, 2:46 PM · BF Blender: Regressions, BF Blender
Philipp Oeser (lichtwerk) triaged T53843: RMB menu Online Manual call disfunctional, throws Attribute Error as Confirmed priority.

Can confirm this [this worked in 2.79], having a look if I can find the issue...

Jan 20 2018, 12:36 PM · BF Blender: Regressions, BF Blender
Philipp Oeser (lichtwerk) triaged T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node as Confirmed priority.

caused by rB04800ed43434d053359cb3035eb082902fe80e4b

Jan 20 2018, 1:07 AM · BF Blender: Regressions, BF Blender

Jan 19 2018

ronald engel (ronald8e) added a comment to T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node.

So maybe disabling the render layer in properties panel should mute the renderlayer node too? If I mute the renderlayer node and leave the renderlayer enabled it still renders the renderlayer, so in 2.79.1 I had to disable in properties panel and mute in nodes to turn the layer off?

Jan 19 2018, 4:48 PM · BF Blender: Regressions, BF Blender
ronan ducluzeau (zeauro) added a comment to T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node.

In Master, blender respect what user asks ; it saves an Image for a renderlayer that have not been rendered.
It is the set-up shown, here. A disabled renderlayer but a renderlayer node that is not mute.
If the set-up was a disabled renderlayer and a renderlayer node that is mute. Master still respects that and doesn't save a blank image.
Mute property can be animated. But it is not exposed in UI of properties of the node.

Jan 19 2018, 3:44 PM · BF Blender: Regressions, BF Blender
Sergey Sharybin (sergey) added a comment to T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node.

It is an exactly design issue. And any decision here is kind of arbitrary.

Jan 19 2018, 2:23 PM · BF Blender: Regressions, BF Blender
ronald engel (ronald8e) added a comment to T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node.

Tested it with some blur nodes between Renderlayer and Composite - in this case 2.79 also starts to save empty images. So maybe its more a design decision: should disabled Renderlayers be skipped in compositing or behave like empty/white/transparent image.

Jan 19 2018, 1:08 PM · BF Blender: Regressions, BF Blender
ronald engel (ronald8e) added a comment to T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node.

I made two Screenshots for better explanation ...

Jan 19 2018, 12:55 PM · BF Blender: Regressions, BF Blender
Sergey Sharybin (sergey) reopened T53835: Disabled Renderlayers still get saved (blank) through File Output Node when connected to Composite Node as "Open".

Well, didn't notice this worked before. But then still, what if you've got other nodes in between...

Jan 19 2018, 12:35 PM · BF Blender: Regressions, BF Blender

Jan 11 2018

Campbell Barton (campbellbarton) added a comment to T53455: Track To constraint makes object disappear.

Thanks for tracking down the commit. This is quite an edge case - while the vector is normalized it's so close to zero to begin with, a second call is needed to make it really normalized (which fixes the bug).

Jan 11 2018, 12:55 AM · BF Blender: Regressions, Animation, BF Blender
Campbell Barton (campbellbarton) closed T53455: Track To constraint makes object disappear as Resolved by committing rBac2ebf9c54a9: Fix T53455: Object disappear w/ TrackTo constraint.
Jan 11 2018, 12:52 AM · BF Blender: Regressions, Animation, BF Blender

Jan 10 2018

Philipp Oeser (lichtwerk) reassigned T53455: Track To constraint makes object disappear from Philipp Oeser (lichtwerk) to Campbell Barton (campbellbarton).

caused by rB16fbb47c8860b61694b2e08270d198f8d173b32a

Jan 10 2018, 9:19 PM · BF Blender: Regressions, Animation, BF Blender

Dec 13 2017

Campbell Barton (campbellbarton) closed T53529: Rip tool on single edge crashes Blender as Resolved by committing rBbe54868ec3ff: Fix T53529: Rip crashes w/ wire edge.
Dec 13 2017, 6:14 AM · BF Blender: Regressions, BF Blender

Dec 11 2017

Sergey Sharybin (sergey) assigned T53529: Rip tool on single edge crashes Blender to Campbell Barton (campbellbarton).
Dec 11 2017, 3:32 PM · BF Blender: Regressions, BF Blender

Dec 2 2017

Brecht Van Lommel (brecht) added a commit to T53349: Cycles - difference between OpenCL and CUDA with AO simplify: rCaf120f146aaf: Fix T53349: AO bounces not working correct with OpenCL.
Dec 2 2017, 6:11 AM · BF Blender: Regressions, Cycles

Nov 26 2017

Brecht Van Lommel (brecht) closed T53349: Cycles - difference between OpenCL and CUDA with AO simplify as Resolved by committing rB83e80db56ec9: Fix T53349: AO bounces not working correct with OpenCL..
Nov 26 2017, 4:00 PM · BF Blender: Regressions, Cycles
mathieu menuet (bliblubli) added a comment to T53349: Cycles - difference between OpenCL and CUDA with AO simplify.

@Brecht Van Lommel (brecht) I don't have commit rights, could you commit it please?

Nov 26 2017, 9:57 AM · BF Blender: Regressions, Cycles

Nov 25 2017

mathieu menuet (bliblubli) added a comment to T53349: Cycles - difference between OpenCL and CUDA with AO simplify.

here is the fix

Nov 25 2017, 9:32 AM · BF Blender: Regressions, Cycles

Nov 21 2017

Mai Lavelle (maiself) reassigned T53349: Cycles - difference between OpenCL and CUDA with AO simplify from Mai Lavelle (maiself) to Brecht Van Lommel (brecht).
Nov 21 2017, 9:21 AM · BF Blender: Regressions, Cycles

Nov 20 2017

Bastien Montagne (mont29) assigned T53349: Cycles - difference between OpenCL and CUDA with AO simplify to Mai Lavelle (maiself).

@Mai Lavelle (maiself), think that one is for you? Thanks. :)

Nov 20 2017, 8:38 AM · BF Blender: Regressions, Cycles

Nov 19 2017

mathieu menuet (bliblubli) updated the task description for T53349: Cycles - difference between OpenCL and CUDA with AO simplify.
Nov 19 2017, 5:37 PM · BF Blender: Regressions, Cycles
mathieu menuet (bliblubli) updated the task description for T53349: Cycles - difference between OpenCL and CUDA with AO simplify.
Nov 19 2017, 5:24 PM · BF Blender: Regressions, Cycles

Nov 6 2017

Brecht Van Lommel (brecht) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

Ok, if the code was modified then indeed a new log would be useful.

Nov 6 2017, 4:44 PM · BF Blender, Cycles
mathieu menuet (bliblubli) updated subscribers of T53249: [regression] OpenCL performance becomes very random with big scenes..
Nov 6 2017, 4:28 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

Where is the "Free mem AMD" print coming from? I can't find that code in master or earlier revisions. In master, the split kernel global size is determined by max_buffer_size and num_elements, which from the logs don't appear to change. Yet the global size is reported as being different.

Yes, I used another version to get the free memory reported and tried to see if limiting global size to make it all fit in memory would solve the problem, but it didn't. I can redo the log with vanilla master if you want. Here is the code:

Nov 6 2017, 2:25 PM · BF Blender, Cycles
Brecht Van Lommel (brecht) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

Where is the "Free mem AMD" print coming from? I can't find that code in master or earlier revisions. In master, the split kernel global size is determined by max_buffer_size and num_elements, which from the logs don't appear to change. Yet the global size is reported as being different.

Nov 6 2017, 12:47 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

@Brecht Van Lommel (brecht) thanks for P555 . Tried it but bug is still there.

Nov 6 2017, 10:24 AM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

@LazyDodo (LazyDodo) the GPU-Z log is wrong somehow, it ignores half of the memory. But it gives the impression that no memory leak happens on the GPU.

Nov 6 2017, 8:38 AM · BF Blender, Cycles
Brecht Van Lommel (brecht) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

Thanks for the tests! To be clear I'm not expecting anyone to work on this bug, and if no one else does I'll probably do it at some point, but the work is certainly helpful.

Nov 6 2017, 12:38 AM · BF Blender, Cycles

Nov 5 2017

LazyDodo (LazyDodo) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

Could be interesting to look at the output of gpu-z to see what the cards actual memory is doing in between those two runs.

Nov 5 2017, 8:34 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

actually, 2.79 has the bug, only the official one had the device selection bug and took the 1080Ti instead, which doesn't use system memory.
So it may be a driver bug, but then why is the first render always 30sec?
After some renders, I got up to 114seconds to render = nearly 3x slower... At this point however, the GPU was idling a lot, maybe waiting all the time for system memory access?
Here is a picture of the task manager with 2 consecutive renders on the same instance of Blender.


It may be a coincidence, but VS2013 builds had only +/-10% between first and consecutive renders (made 5 of them) while VS2015 builds go crazy with up to 3x the render time.
If someone could test on Linux with a RX480 to see if GCC or the Linux driver handles this differently. As said before, the RX480 can render this scene. On Linux, the Nvidia drivers destroy a part of the AMD driver and I couldn't find a solution to have both drivers side by side yet.

Nov 5 2017, 8:30 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

commit b53e35c655d4 already has the bug, so it's not due to the buffer patch.

Nov 5 2017, 6:44 PM · BF Blender, Cycles

Nov 4 2017

mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

got some explanations on IRC, sorry I didn't know the whole story.

Nov 4 2017, 9:02 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..
Nov 4 2017, 8:19 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

the bug was already there 24.08.2017, so my guess is that rBec8ae4d5e9f7 is the commit we look for.

Nov 4 2017, 8:15 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

just to give an idea of the mess to bisect:

  • cuda disables completly opencl in the majority of revision, so you have to rebuild without cuda
  • device selection changed, so userpref have to be modified depending on the revision you test and bisecting requires to go back and forth in time.
  • kernel compilation takes 1min50 for victor
  • scene preparation takes 2min04

so it takes about 5minutes of VS compile, then manual tweaks for user pref, then 2minute kernel compile+ then 2 renders at 2 (scene preps)+2(render)=8minutes of rendering. That's a quarter of an hour with 4 user intervention between which you can't do much.

Nov 4 2017, 7:59 PM · BF Blender, Cycles
mathieu menuet (bliblubli) added a comment to T53249: [regression] OpenCL performance becomes very random with big scenes..

rBec8ae4d5e9f7 only added support for more than 4GB of textures iirc.
You don't need HBCC support. On win7, even on Vega, there is no HBCC and my RX480 also renders full victor scene since a year on windows and since some months on Linux.
@Brecht Van Lommel (brecht) Is there a simple command to disable the context caching?
I could try to bisect, but @Mai Lavelle (maiself) should have better guesses of what could have introduced this bug. The scene preparation of Victor takes more than 2 minutes on my computer. With compile time on windows on top, shooting in the dark to bisect would take a lot of time.

Nov 4 2017, 7:27 PM · BF Blender, Cycles
Brecht Van Lommel (brecht) triaged T53249: [regression] OpenCL performance becomes very random with big scenes. as Normal priority.

I thought Victor only rendered on AMD after rBec8ae4d, not yet in the 2.79 release?

Nov 4 2017, 6:27 PM · BF Blender, Cycles