These are the settings I use after new upgrade to Windows 10 Pro and I also have a GTX 970. And to point out I am using the K-lite Codec Mega Pack, nothing else.
Latest Nvidia Drivers 353.62. I must say that these are the best ever settings I have achieved yet and my CPU is only a dual core i3-4160. My CPU usage doesn't jump above
50% usage with Win 10, when using Win 8.1 Pro my CPU hovered between 60-70% when playing large bitrate 1080p content.
Scaling algorithms:
Chroma upscaling: NNDED13 256 neurons (activate Superes filter)
Image downscaling: Spline 3 taps
Image doubling: Everything unchecked
Image upscaling: Jinc
Upscaling refinement: Everything unchecked
Rendering:
General settings: Everything unchecked
windowed mode settings: Everything left at default
Smoothmotion: Unchecked
Dithering: Ordered Dithering
Trade quality for performance: Everything Unchecked
GPU Acceleration (Open CL) is ticked in SVP. Huge performance increase.
MPC-HC:
Internal Fitlers:LAV decoder: Cuvid (I actually don't know if this is being used as I have DXVA Native selected in external LAV Video Decoder)
External: ffdshow raw video filter, LAV Video Decoder
xenius wrote:johnnys wrote:CUVID is disabled in the lastest windows 10 Nvidia drivers....they are planning on using NVENC as the next API for encoding/decoding...
xenius wrote:Indeed, disabling CUVID works. Thanks! Just have to wait for updated drivers is suppose.
Well all i did was change my internal video decoders to FFmpeg instead of NVIDIA Cuda Decoder. Thats was the problem for me why my player crashed. Im also using the newest nvidia drivers 353.62.
Can also anyone give me an example to begin with to configuring madvr. I've tried some settings but i don't really see much of a difference tbh Atleast not in low res bad quality videos
Im using a i5 2500k at 4.5k with 8gb of ram on a GTX 970.
Atm these are my settings:
Chroma upscaling: NNEDI3 128
Image doubling: always use 128 luma & chroma double
Image upscaling: Lanczos 3 taps ARF
Image downscaling: Catmull-Rom ARF
Smooth motion: only if there would be jitter without
Dithering: option 2