- User Since
- Mar 20 2010, 4:06 PM (483 w, 1 d)
Apr 3 2017
I think it probably needs to over sample the surface (extra samples between where it will place dabs) and maybe even consider the derivative of the slope and do a proper quadrature to predict how much it should change spacing, instead of the simple change it does now (sorry if the implementation has been improved over my original simple one, I haven't had a chance to go over this patch)
I hate to bike-shed my own feature, but I hate the name "adaptive spacing". It seems like this should be the default behavior and the old screen-space based spacing should be called "spacing *based on screen distance" and adaptive spacing should just be "spacing". If you have to call this anything it should be "spacing *based on object distance"
Feb 25 2015
Was thinking that we only use the PFD from ChoosePixelFormat as a last resort, so we actually take full control over picking a context on Windows.
Very interesting if you have to explicitly add the PFD flag to get the directx wrapper. Means that there are probably very few programs in the wild that actually ever use it.
Feb 24 2015
OK, found the docs here: https://msdn.microsoft.com/en-us/library/cc231189.aspx
It is surprisingly hard to find info on what the values of GL_RENDERER is when you get a Microsoft context that supports OpenGL 1.4 through Direct3D.
Feb 23 2015
EDIT: The spec you linked, in the second paragraph, says multiple flags may be set due to potential parallelism in the implementation.
The performance problems I was concerned about were due to other checks that you did not put in this patch.
I'm create a patch that allows an external engine to use a separate context, but please feel free to send me a personal email if it looks like I've forgotten about it.
Feb 10 2015
I was looking at how FreeGLUT handles what it calls "subwindows", which have an option to be created with their own context, even though they are child windows of a top-level window (or of another subwindow).
Feb 6 2015
By sharing I mean objects created in one context are available in the other.
Feb 5 2015
Made several inline comments.
@rdb (rdb), still not a great example because uniforms are a part of program state and if you implement your own matrix it wouldn't affect Blender and I don't think that Blender should make any assumptions about glUseProgram after calling a foreign renderer.
The example given, of a matrix, is actually an example where there is a well know convention of pushing and popping a stack on order to preserve state.
Now I cannot get CMake to even configure WITH_CYCLES_STANDALONE, so I'm going to go ahead and voice my doubts that the build bot is actually building it.
Feb 4 2015
I will check and see if the problem still exists.
I cannot say because I don't know what the differences are.
Jan 24 2015
@rdb (rdb) I was actually talking about BGE, but I guess things are more compkex. I still need to look at it.
Jan 22 2015
I fix this problem in my SoC patch. I do it by making sure the Game engine follows a reasable GL state setting policy and (maybe) calling the gpu reset function when done (don't remember if its needed, cause ideally it shouldn't be).
Jan 12 2015
Dec 3 2014
@Sergey Sharybin (sergey) ok, I'll just fold this in with the combined patch I'm currently working on
Nov 28 2014
This reminds me about possibly renaming "draw normal" to something else like "draw default" since that always tripped me up, but I'm not even sure if that is relavent after this patch.
Nov 10 2014
This problem still exists. Would this be an acceptable fix for the problem?
Oct 14 2014
I think I could commit all of these remaining changes. My only question is whether the change to the type of embedder window should also be changed for msvc when compiling for 64 bit.
I think I've fixed these issues now.
Oct 13 2014
Getting rid of GLEW would actually be pretty cool.
When I read Blender's OpenGL code I find a lot of mistakes made because somebody did not look up exactly where an entry point is defined. There is a failure to check for the right version or extention or to even check at all.
Oct 11 2014
Does this occur with multiple Blender windows open?
My first guess would be a problem with display list sharing.
Sep 14 2014
Sep 10 2014
I added this now because it doesn't really have any dependencies. Doesn't really matter when it is done, but it is part of viewport.
Aug 11 2014
Did a review of my own code to make a todo list.
Aug 5 2014
In https://developer.blender.org/D706 I have a GPU_print_error function that prints to the console, but maybe a hook could be setup to send errors to a place a user can see more easily.
I don't mean to spam, but the glreport.py script that used to be in the tools directory will be useful here since it will tell you the version of every gl function that is used in blender.
One thing that I mean to do in the above patch is do further checking to make sure that extensions and versions are always checked, but it already creates a central place to handle this.
Aug 3 2014
small optimization of the MX_profile_* macros
Aug 2 2014
Aug 1 2014
Jul 31 2014
This patch fixes a problem that stops the cycles standalone from compiling on MSVC.
Jul 27 2014
Jul 26 2014
I have also run into this problem. I did think at the time that recompiling jpeglib would be a better solution than adding the flag, but I have been using the flag as a workaround so I can work.
Pushed some minor changes to the SDL context.
@cambellbarton The sdl context looks exactly like I imagined it. I've got a bit on insomnia, so I'm glad I didn't start working on it myself.