|
That does shed some light on some things. Plain EVR and EVR CP with default settings just use the same old rendering methods in consumer products as over a decade before. The fact that these use 8-bit intermediate storage in textures makes them very prone to producing banding.
Setting limited ranges of whatever type and delegating some filtering to the display device may cause it to apply dithering. This isn't really a good thing. Dithering is a renderer task. (At least in consumer-grade devices; professional systems don't have to dither) There should be only one renderer in the playback chain.
As you are experiencing banding problems, did you try the black and white point calibration pictures (in a normal image viewer, not in a video renderer)? Many consumer display devices are set up to expand some type of 16-235 to full range by default.
The 10-bit options in MPC-HC are a bit disorganized.
-The 10-bit input option is fake. My predecessors did not know what they were dealing with. Do not use. (When I started to edit the renderer this was pretty much the first 'feature' I got rid of.)
-The 10-bit output option is real. It requires decent quality rendering surfaces, Windows 7 or newer and D3D fullscreen exclusive mode. It currently works on the AMD/ATi display adapters with HDMI, DP and analog outputs, and on Nvidia workstation cards (although I only received few reports about this, though).
-The 10-bit rendering surfaces option is a poor choice compared to the other options. 32- and 16-bit floating point or 16-bit integer surfaces are better.
The AMD Catalyst Control Center features video 'enhancement' filters in the video section. By default some of them are enabled. The pulldown option and allowing DXVA are perfectly fine to enable, but I can certainly advise to test stuff first with the other filterng options disabled.
The mentioned playback chain of "MPCHC - softw.dec. - avisynt (denose + sharpness) - Madvr" isn't too bad. I just wonder why you don't delegate the denoise and sharpen steps to Madvr. The artifacts when using DXVA video decoding are caused by the filtering I already described earlier.
|