- User Since
- Sep 12 2004, 3:45 PM (774 w, 2 d)
@Emerson Rosa (e1412) You can discuss new features at https://blender.community/c/rightclickselect. There you can discuss ideas with others, and flesh them out so that they're in such a detailed state that a developer can pick them up and work on an implementation without having too many open ends to think about.
- Moved isEnabled check into BKE_object_modifier_use_time()
I can not risk delaying my project
Is that really the smallest you can make the file? The smaller/simpler you can make it (while still showing the problem of course), the less time developers have to spend digging through unknown stuff in order to figure out what's going on. Please spend some time making the smallest file possible -- you know the file, so you'll be able to do that much faster than us.
Thanks for the info; I've copied that page to the new wiki.
Please use the most recent daily build of Blender from https://builder.blender.org/download/, and fill out the bug report form completely. Without more information (that is, the exact steps needed to reproduce the issue, and a more elaborate description of the problem) we can't solve this.
Fortunately, this is not as big an issue as you describe. The code you linked is only used when enabling an add-on; the issue you describe seems to assume the way to reload an add-on is to disable and enable it.
Sun, Jul 14
Fri, Jul 12
We really need a minimal blend file that allows us to reproduce this issue.
- Using CTX_data_pointer_get_type() as per Brecht's comment
Look, we can't do anything until you describe, in detail, what the problem is. Those fields "[Please describe ....]" and "[Please fill out ...]" are there for a reason - to make sure the developers can see your problem for themselves. Without this, there is nothing we can do.
Thu, Jul 11
Thanks for confirming.
This looks like a feature request (support material IDs rather than just materials) for the FBX importer; not sure though, could be considered a bug. At least re-test it with the latest build of Blender 2.80 from https://builder.blender.org/download/.
I think it will be so smart that it will no disturbs the text. or limit the value...
It reads 'Volume' on my build. Try making the panel a bit wider and see if that resolves it.
This is why a minimal example is so important (and it still isn't minimal, it has twice the number of bones necessary to show the problem, no need to do both X and Y).
Please test with the latest daily build from https://builder.blender.org/download/. Also make sure that you've updated your graphics drivers to the latest version available from the Intel site.
I already have a fix waiting for 2.81 in D5191, which I'll commit when the repository is open for new features again (it's now locked in preparation of the 2.80 release). If that fix is in and doesn't help, then I'd really appreciate some files to test with :)
Wed, Jul 10
I checked with the UI team, and this is working as designed. When you pin in the mesh panel, you pin the mesh itself. The vertex groups and shape keys are defined on the object, so they disappear. When you pin from the modifiers tab, you pin the object, so you still have access to its mesh. The same is true for lamps, cameras, materials, etc.
All the development is focused on 2.80. We simply don't have the people to keep working on older releases too.
However VP9 CRF values should be quite higher. For example for quality NORMAL, default CRF for H.264 is 23 but for VP9 it should be around 28.
The screen cap you show is from an old version of Blender. Fixes don't magically appear in old versions of Blender. The new version will be released soon (hopefully next week), and before that you can test with daily builds from https://builder.blender.org/download/
Your Blender is more than a week old, and a lot of development happened since then. Please test with the latest daily build from https://builder.blender.org/download/
Brecht already told you that the problem was fixed in the upcoming release of Blender. Since this is a tracker for developers, and the issue is fixed, the developers can't do anything any more, which means that the issue is closed.
@david law (Ponlets) This is not a user support forum, it is a bug tracker for Blender developers. Since the bug has already been fixed we cannot do anything else.
Duplicate of T66656
- Removed return value of curvetomesh as per @Bastien Montagne (mont29)'s comment
- Handled feedback from Brecht + addressed another memory leak
The root cause was that Blender is unable to convert a single-vertex curve to a mesh, but didn't handle the resulting NULL pointer. This is a minimal example that just contains that curve:
Thanks for confirming it's fixed.
The script contains some mistakes (you shouldn't iterate over the collection of "selected objects" and then change the selection state; that changes the very collection you're iterating over). However, the crash also happens when you simply select Convert to → Mesh from the menu.
You're using the 2.79b release of Blender. Since then there have been a lot of changes, please test with the latest daily build of 2.80.
I copied the active scene from your example file (chose Full Copy), but can't reproduce the issue. The "sliders" still manipulate the camera in the same scene. Maybe re-test if this is still an issue in the most recent daily build?
Please provide us with exact steps to reproduce the issue. We don't have Clarisse, and I don't know what "combiner" is.
Please attach an as-simple-as-possible blend file that shows this issue. "Start with a simple character rig" is too vague a description to be 100% sure we're looking at the same issue as you are.
This is fixed in D5191.
I've fixed exporting loop normals as well; now things load properly in Gaffer as well as Blender, so I'm confident this is a proper bug fix (or two).
- Added comment about smoothing in SubD loading
- Alembic export: fix exporting of loop normals
Tue, Jul 9
Indeed, with a big chunk to begin with. Something along the lines of std::vector<...> v; v.reserve(10000).
@LazyDodo (LazyDodo) looks like a good fix to me, I'll commit that soon.
What about using a dynamic allocation scheme, but preallocating the same amount of memory as is done now? That should keep things fast up to the same limit, but instead of having the strokes disappear Blender just becomes a bit slower.
Please don't attach huge videos, but just describe it. A few well-chosen words are much easier to digest for us developers than having to watch a video.
If Blender is outputting multiple files, you must have something else going on than just a plain render to PNG. Please restart Blender to ensure you don't have any lingering persistent Python handlers. Then reset Blender to factory defaults (you can do this temporarily through File → Defaults → Load Factory Settings), run your script, and try again.
I can get it to crash when changing the "Set Origin" operator parameters in the redo panel.
I can confirm this. Here is a stack trace from a debug build:
I can't reproduce this on either Blender versions ad16af7a7e350a2164204b8af50ff00836d65386 and 805cabdf177af6c71da6239ff3fc5d2524286a18 running Kubuntu Linux 18.04 LTS.
Apparently this is caused by a limit on the amount of data buffered by Blender. The limit is set by the #define GP_STROKE_BUFFER_MAX 5000 line in gpencil_intern.h; decreasing this number causes strokes disappearing after less points.
Confirmed with Blender 805cabdf177af6c71da6239ff3fc5d2524286a18 on Kubuntu Linux 18.04.
I can reproduce this with Blender 516afd0162cd2e217ec50ce522f87317b73f0088 on Linux.
I can confirm this on Blender 516afd0162cd2e217ec50ce522f87317b73f0088.
It works fine here on Blender 516afd0162cd2e217ec50ce522f87317b73f0088:
I can't reproduce this (Blender 516afd0162cd2e217ec50ce522f87317b73f0088), for me the script doesn't do anything. Looking at the code, the 'wall slicer' doesn't intersect with any walls (I put a break in the code to stop it after the wall slicer was created):
Add-ons were moved between the addons and addons-contrib directory. Try removing Blender first, before installing the latest build.
Mon, Jul 8
I did a quick test with CommonMark, and it turns out it doesn't auto-link anything unless you use explicit markup:
Fri, Jul 5
@Bob Metcalfe (secretanorak) your report states "Broken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)" -- please fill out that information so that we know which exact version/build of Blender you're using.