- User Since
- Jul 9 2014, 7:45 PM (293 w, 6 d)
It was surely because of a constant defined twice (crash was happening here on linux+Mesa). Maybe fe41c1ec0b79 fixed it. Can anyone confirm the fix?
It does sounds like a hack. Instances don't have DrawData for examples. How does this patch is expected to work in this case?
A new option was added to allow high bitdepth (32bit) in master for 2.83.
- Fix inverted test
Mon, Feb 24
- Change default blur setting, rename background prop, change roughness remapping
@Antonio Vazquez (antoniov) It seems the vertex color modifier is no longer working with the latest patch. It colors the fill strokes with the same color and does nothing on the normal strokes.
I checked with @Julien DUROURE (julien) Kaspar (JulienKaspar); The default setting for the blur is 0.0. We should make it blurry by default (1.0)
I'm pretty sure most people doing lookdev want the opposite. So I think default could be 0.5f, a mix between context and readability.
I have no idea why this happen. I don't have a Mac to reproduce.
Sun, Feb 23
The gpencil world lighting only supports the world color not the world nodetree. This is a known limitation, not a bug.
@Matias Mendiola (mendio) I can accept to do the UI changes in master.
Ok I have reviewed every single modifier. Since we are breaking workflow for GPencil I think it's a good time to do a review of what is in there and the quality of them.
Sat, Feb 22
One compiler error but the patch looks good! No visual regression!
Alpha handling is totally dependent on colorspace. In your example 0.07(linear) ^ (1.0/2.2) is almost 0.3(srgb). Because everything is now rendering in linear space all blending is also done in this space. Which is why brighter colors are more blended by alpha blending than before (visually speaking but linear values are correct).
Fri, Feb 21
Shader compilation does not reset the accumulation of the renderpasses.
@Campbell Barton (campbellbarton) The blending IS done properly. This was done on purpose. The issue is that now we are doing blending in linear space for correctness which change pretty much all UI colors which use alpha blending. So we should patch the theme instead of the blending function.
Thu, Feb 20
Here you have the patch where I fixed the glass and some other small discrepencies: P1266
I'm going to commit a slightly different version where I optimized the matrix mul and only do that if custom bones are effectly used.
Wed, Feb 19
I've added support for detecting the half floats for EXR and PSD as it seems they are the only format that can support them. The other formats are using 16bits normalized int encoding or RGBE encoding (hdr). I don't know about dpx/cineon but I would guess they are normalized ints.
- Add support for detecting half float exr and psd files
It is a misinterpretation of what normal maps are. If you use a bump node and a heightmap it should work but with lower performance.
- RNA: add use_ prefix to boolean property
- RNA: Fix case in properly title and description wording
- Add compressed tangent
- Rebase on top of master
Looks good to me! Will commit that tomorrow! Thanks for the patch!
- Introduce new Performance subpanel and High Quality Normals option.
Can I commit this as it is in the patch or should I rename the variables?
I would say it's worth to rework the UI python a bit for that now that you are changing it. It's done in a matter of minutes.
Tue, Feb 18
I only have found a two issue:
- When clearcoat is not 0.0 and metalic is 0.0 (so possibly using node_bsdf_principled_clearcoat as entry point) the diffuse color/light changes with more energy moving to the diffuse color pass.
- The Specular node shader does not output the specular light the same way principled shader does.
Fixed by rB9491e7d7c218
Ok just an addition: It is not working in my screenshot but it works if using the UVs as coordinates. You cannot do normalmapping with tangent space that is not the one that was used to bake the normalmap. To do tilling texture like you want, you need to align the the texture space with the tangent space and if you use Box mapping it's impossible to do.
Your normalmap is not valid. I tested with a normalmap I did years ago and it's working. I don't know how to generated this normalmap but I think it is incorrect because not representing unit vectors.
few nitpicks but it's good to go.
Mon, Feb 17
This is not the same bug. And it is valid.
Fixed by 852cdd476b55