Closed chrispable closed 7 years ago
OK! I solved the issue! It is indeed caused by (Merge pull request #47 from augustg/bugfix/patch-data-not-mountable)
The patch folder cannot be mounted and as such, the fixed note skins are not loaded. You can work around the issue by copying the noteskins to the root itg directory structure.
Do NOT touch the rendering pipeline, it does work fine. It was purely a resurgence of that 2008 issue not getting the updated assets to use the latest rendering pipe for d3d.
Thanks a ton for looking into this.
Is it that it can't be mounted or that the VFS priority for the patch file/dir got fucked (log.txt should have a mountpoint list at the beginning somewhere)?
Sometime between Beta3-WIP2: http://openitg.gr-p.com/wips/wip2-2013-01-25/
And Beta 5 release on github, the D3D backend was borked for the notes and now they always display as "flat" (OpenGL is NOT affected). I suspect it was the D3D 9 migration. SM5 does not exhibit this problem and the backends are very similar. I will try to track it down but have never dealt with any graphics api before.
Update: I tested a reverted DX8 backend and the problem persists. Based on old reports of the issue from 2008, it may have to do with vertex shaders, however, it did work in 2013 at some point so it was fixed for a bit at least. Bug report from 2008: http://www.boxorroxors.net/viewtopic.php?p=7988&sid=72c2b23126dc10aa7665b8e1851609b7#7988
And it seems to have been mitigated with revision 626 (not on git to see changes): Revision 626:
RageDisplay_OGL: noteskins now work without the vertex shader, so ween away from that and support older cards
Half fixed?: Found revision 625, I applied the mesh hack in the function: void Draw( int iMeshIndex )
Setting the addition condition of: iMeshIndex > 0 in the if statement selects the proper note colors BUT the texture itself scrolls.
Trying to track down the commit that did this -- The git timeline is a bit fucked because branches were deleted so I can't see where they were forked from. So far I have found it has not appeared in the branch work of the dx9 commit, but was in the merge of the dx9 commit. I've also seen the bug as early as April 15th 2016 (Merge pull request #47 from augustg/bugfix/patch-data-not-mountable has the bug). Will Keep editing this with a narrowed date.