These are built from the current development sources (Git master branch).
If you notice anything wrong with theses, complain to lachs0r on Freenode or Rizon, or report the problem at mpv’s GitHub issue tracker.
Notable changes since 2015-01-20:
[02df988, 6c25050, f5e48f0] vo_opengl: performance improvements
4-tap EWA Lanczos is now roughly an order of magnitude faster and thus quite usable on most modern GPUs.
[27261be, e872852, 1ec7721, 8eb9ddd] vo_opengl: remove/rename some options
Removed scale-sep and indirect. These are now auto-detected sanely; and enabled whenever it would be a performance or quality gain (which is pretty much everything except bilinear/bilinear scaling).
Removed sinc/(ewa_)lanczos/blackman aliases. Just set the radius with scale-radius if it's really needed. The default for this is 3.0, which should be optimal for most of these scalers. In other words, lanczos3 is now lanczos and lanczos4 is lanczos:scale-radius=4.
Renamed suboptions: lscale → scale, lscale-down → scale-down, lparam1/lparam2 → scale-param1/scale-param2, lradius → scale-radius. Same thing for the c options, i.e. cradius → cscale-radius.
[2d182fd] vo_opengl: implement naive anti-ringing
Controlled with the scale-antiring suboption (default off). Only works with ewa_lanczos. A value of 0.7 should yield decent results.
This is not quite the same thing as madVR's antiringing algorithm, but it essentially does something similar.
Porting madVR's approach to elliptic coordinates will take some amount of thought.
[39231e5] vo_direct3d: disable shaders if unavailable
Apparently, extremely crappy graphics drivers don't allow you to use shaders. Simply disable use of shaders if this happens, and use the "old" method instead.
[724f722, e349579] remove vo_opengl_old
At this point, there is probably no hardware left that doesn't do OpenGL 2.1, and at the same time is fast enough to handle video.
Notable changes since 2015-01-08:
[286340d] video: Add sigmoidal upscaling to avoid ringing artifacts
This suppresses ringing artifacts caused by upscaling in linear light with vo_opengl, i.e. when using ICC profiles.
[26baf5b] vo_opengl: add ewa_lanczos upscaler (aka jinc)
Still a little slow due to GPU memory bandwidth/latency limitations. This can be made much faster, but the shader code really needs to be cleaned up. Default for opengl-hq remains spline36 for now.
[61f5a80] vo_opengl: get rid of approx-gamma and make it the default as per BT.1886
Since mastering studios have historically been using only CRTs, the practice adopted for backwards compatibility was to simulate CRT responses even on modern digital monitors, a practice so ubiquitous that the ITU-R formalized it in R-Rec BT.1886 to be precisely gamma 2.40.
As such, we finally have enough proof to get rid of the option altogether and just always do that.
[1883b7c] player: add --autofit-smaller option
Similar to --autofit-larger. Ensures a minimum window size.