On my last post, I had forceware version 77.77 installed and using a GeForce 6800GT
I just updated to version 81.95 and also updated my TheaterTek to version 2.2 so I could have the newest nvidia codec that they supply.
Some interesting findings. I think it is bad, but for others, it could be good.
One of the new features for this driver release is: PureVideo high definition MPEG-2 de-interlacing support. This pretty much is true.
No matter which Video Decoders I attempted to use in ZoomPlayer, the video card was doing its own de-interlacing. I was about to write 3:2 pulldown, but I am not sure if it is doing this or not. All the clips I was testing were .TS 1080i movies that had interlacing artifacts. Even though the card was now doing alot of the work, it was not 100%. I would day maybe 90-95% of the interlacing was gone, but you could spot some here and there. Notably during scene transitions.
Ocean's Eleven was now without interlacing artifacts when using theatertek's nvidia codec, DScaler5 alpha codec, powerdvd codec and windvd codec. Before this, movie had interlacing artifacts when these codecs were being used. When choosing kzeuh's Dscaler plugin, the video card and the plugin were fighting over who should handle the video. There was alot of jerkiness and weirdness going on.
One other strange thing, after installing the new drivers, I had to change the Output Color Space of the Dscaler plugin from YUY2 to YV12. YUY2 had a major blue tone to it. Skin color looked blue like an Andorian. The intro to Spider-Man 2 was not red, but blue. Changing it to YV12 fixed it.
I have since uninstalled 81.95 and installed 77.77. Ocean's Eleven is back to its interlaced self. For the rest of the movies, DScaler now applies 3:2 pulldown without something else doubling the work (ie. the video card).
I thought I would prefer the video card to handle the IVTC, but if it cannot do a perfect job (smooth as silk, no interlacing at all, etc) then no thanks.
Anyone else have a 6xxx/7xxx series card with driver version 81.95 installed not happy with the results compared to kzeuh's method?