You're correct about the way motion samples are stored.
There is a feature like this in the latest builds:
- Please try running CUDA-Z to verify if you have a working CUDA installation.
- Do a complete uninstall of the NVIDIA driver and then use "Perform Clean Install" when installing the driver from the NVIDIA website.
No need to mention your name in comments, and use /* */ style.
I suggest to keep Interface at the top for now, and to rename Save User Settings to Save Settings in the header.
This is really a known to do item and not a bug, so closing, feature requests are handled elsewhere.
I can reproduce the difference with the latest attached script, it's due to this bug fix: rBe7aad8fd0e28: Fix T53559: Auto texture space for text and font is wrong in Cycles
The code here is clearly wrong, so committed the patch.
Looks good to me, thanks for the fix!
Closing due to lack of information to investigate this further, if we can get a .blend file to reproduce this problem we'll reopen the report.
Please specify all information requested on the bug submission form:
Setting priority back to incomplete, please do attach a .blend file. I can't see any difference with a material created by this script.
Sat, Mar 17
Ok, let's see how the seam-merging code will work before making a decision then.
I find this quite difficult to understand, and would rather not have so much specialized code for this.
Seems generally fine.
Fri, Mar 16
Thanks for testing, my best guess is rBd454a44e9, but doesn't really matter.
Thanks, I'll commit this with some additional fixes.
Thu, Mar 15
Thanks for the updates, it builds for me at least but I did not look more deeply into the code since it's WIP anyway.
This patch just removes the OIIO image reading code, which we don't want to do, Blender, Cycles standalone and other integrations rely on it. It should become optional with a CMake build option instead.
Looks good, I'll do some further tweaks and commit.
Looks good to me and works well here, just needs one fix.
Multiple small patches are much easier to review and are less blocking, I much prefer them.
Sounds like a good plan.
Seems reasonable, since it uses dlopen it's not an invasive change.
Thanks for the report, this appears to be another variation of a known issue that we are trying to address in D1212 for example. Small features far away from the original are problematic for raytracing.
Does changing the glass from Multiscatter GGX to GGX work around the problem for you?
Wed, Mar 14
Right, but that's what I intended. If you want to repeat the bevel, you can use Shift R or F3. I can't see a good reason to be inconsistent between typing in a value and move the mouse to set the value.
@Sybren A. Stüvel (sybren): the EXR bug did not affect 2.79a, it should not be included.
Tue, Mar 13
It seems the dirty page purging totally changed:
I can't reproduce the problem on macOS 10.13.3 with a Wacom tablet.
Please describe the exact steps you are doing, for example :
- File > Load Factory Settings
- Change manipulator to scale
- Press and drag one of the axes of the manipulator
render_image_update_pass_and_layer() already sets sima->iuser.layer to the layer currently being rendered in the image editor, If we do nothing in render_image_restore_layer() in case of single layer rendering, isn't that enough, so that we don't have to worry at all about figuring out what the active layer is?
If you want to have press and hold behavior for tools, then indeed the keymap needs to change to accommodate that, not just here but in many other places. R key rotation and most other tools were not designed for that though, and I'm not sure why you would expect them to be?
Probably a duplicate of T54269, which revision did you test?
@YAFU (YAFU), does running Blender 2.79a like this cause the reported memory usage to be reduced after rendering?
Mon, Mar 12
It can be scene related also if it works normally in 2.79, since there can be a new bug in 2.79a that is only triggered by specific scene setups. So ideally we want statistics from a scene that we can test ourselves, to figure out where exactly the difference is.
Sun, Mar 11
Thanks for the patch, some comments.
I couldn't build with this patch. It's not applying cleanly on the latest Cycles repository master, and referencing files named alembic.h and alembic.cpp which are not included?
The committed fix is not exactly what I had in mind, I suggested to only change the behavior when the -W option is used (so WIN_OVERRIDE_WINSTATE is set).
I'm fine with renaming --window-borderless to --window-fullscreen. Maybe @Campbell Barton (campbellbarton) has an opinion on it?
So first the confusing thing is that "borderless" actually means fullscreen . Fullscreen at some point meant making a borderless window that covered the entire screen, and combined with the -p option you could get a smaller borderless window. But macOS now has native fullscreen support which works quite different.
Sat, Mar 10
I can't reproduce the problem with the fishy cat scene on Ubuntu with 2.79a.