- User Since
- Apr 27 2010, 10:07 PM (326 w, 1 d)
Sat, Jul 23
Fri, Jul 22
Reusing i for inner & outer loops had already been fixed in BM_face_exists_overlap_subset.
Thu, Jul 21
Has already been fixed upstream in Eigen.
While not a "bug" we do appreciate the writeup. I'll go through the suspicious fragments and fix if needed.
Fri, Jul 8
Mon, Jul 4
Broken in the official 2.77 (no letter) release, so it was before that.
Confirmed broken on nVidia + Mac.
Works fine on Intel + Mac same system.
Yes my Mac is the previous year's model, with Intel HD 4000 + NV 750M. Will check it out.
Fri, Jul 1
Breaks build on Mac / clang.
Jun 27 2016
I don't understand the unification of Layers (for grouping objects) with Render Layers. To me they seem like totally separate topics.
Jun 11 2016
Sure, I'm familiar with this code & have some Wacom gear at home.
May 29 2016
Hey @Chau (BlenderNavi) I'm looking at this again. You are not forgotten!
May 27 2016
May 24 2016
May 22 2016
Apr 27 2016
Did you receive the email thread titled "[Bf-viewport] OpenGL refactoring tasks"? From 11 Dec 2015. We discussed flat shading with the dFdx approach. I still think it's a bad approach but some parts of this patch are good improvements.
Apr 22 2016
To clarify the scope of this design task:
Apr 7 2016
Looks good! I wanted to test it on Mac before giving the OK but have been busy. Go ahead and commit then we'll clean up any rough edges later.
I'll dig into the problem and provide test builds to @Chau (BlenderNavi) if needed.
Mar 28 2016
Going forward, code like this should target the Blender 2.8 release. That's when we'll have GLSL 1.5 and guaranteed geometry shader support. Blender 2.7x will stay at OpenGL 2.1 + extensions.
Mar 3 2016
@Kévin Dietrich (kevindietrich) This is your area, right?
Mar 1 2016
The GL state usage looks good to me. It's more important to set the items you mentioned to their default values. Restoring their previous values with glGet is a performance killer! And not needed most of the time.
Intel GMA950 is an OpenGL 1.4 part, but MacOS has a software fallback for whatever the GPU can't handle. That means OpenGL 2.1 is always supported and available but you're running in software.
Feb 27 2016
Feb 26 2016
Feel free to assign line thickness bugs to me as well. Clean up my own mess :|
Feb 19 2016
Feb 11 2016
This looks great!
Feb 5 2016
Are you sure about the 0.2 samples? Mine only allows 1, 2, 3 and so on.
Feb 2 2016
Viktor already installed the latest driver, so it's not a problem of being out-of-date.
Jan 30 2016
This doesn't point to anything specific... I'll stick an nVidia card (Quadro K600) in my Windows test machine and try to reproduce.
Jan 29 2016
I redid some of this code recently and am glad to see @Kévin Dietrich (kevindietrich) take it even further.
Jan 28 2016
Jan 27 2016
Jan 26 2016
Jan 23 2016
Jan 22 2016
Cool, I was unaware of this feature. The release notes mention it only works in solid shading mode. It also affects edit mode -- hidden parts remain hidden.
Sure I'll take this one. I think it's important to have zero performance impact when clipping is off (vast majority of usage).
Jan 18 2016
Jan 16 2016
Could not build on Mac after pulling the latest changes. Error points to here! Disabled CYCLES_OSL and am back in business.
Jan 15 2016
Disabling alpha test when U.glalphaclip == 0 is another approach. Visually it's the same as discarding alpha 0 pixels. I'll do some experiments & see what works well.
Jan 10 2016
Hrm. It just happened that one time for me. Image displaying fine now.
Jan 9 2016
@Willem Odendaal (dexon) Did you see the problem on official 2.76b or a custom build?
Yep, I'm seeing this on my Mac (geforce GT 650M).
Images appear white in trunk/master.
Appear normal in 2.76b on same machine.
Jan 7 2016
I was only able to track the GL_POINTS hack to January 2008. Wow... exactly 8 years ago!
Jan 6 2016
FYI -- I made small changes to image drawing code recently, after 2.76b was released.
Jan 5 2016
Jan 4 2016
Dec 27 2015
Also try exactly 1.0 since that disables the OpenGL alpha test.