2021-02-01, 03:15
I wonder, is there a preset out there for my GPU, that I can easily copy? I don't really trust my current settings to be the best, because I don't really understand what I am doing. What's the easiest way to share my settings with you guys? Perhaps you can help me optimize my settings, that would be great!
CPU: AMD Ryzen 2600
GPU: nvidia GTX 1060 6gb
monitor: 2560x1440
I have profiles for 576p, 720p, 1080p, 1440p (do I even need this?) and 2160p; each 30FPS and 60FPS.
Chroma upscaling: everywhere NGU, AA LQ
Image downscaling: everywhere SSIM: 1D 100%, scale in linear light, activate anti-ringing filter (relaxed)
Image upscaling:
CPU: AMD Ryzen 2600
GPU: nvidia GTX 1060 6gb
monitor: 2560x1440
I have profiles for 576p, 720p, 1080p, 1440p (do I even need this?) and 2160p; each 30FPS and 60FPS.
Chroma upscaling: everywhere NGU, AA LQ
Image downscaling: everywhere SSIM: 1D 100%, scale in linear light, activate anti-ringing filter (relaxed)
Image upscaling:
- For over all 1440p and 2160p profiles: Jinc + activate anti-ringing filter
- For both 1080p:
- NGU sharp,
- algorithm quality:
- luma doubling -> high
- luma quadrupling -> let madVR decide
- chroma -> let madVR decide
- luma doubling -> high
- activate ...
- doubling -> let madVR decide
- quadrupling -> let madVR decide
- doubling -> let madVR decide
- more scales:
- upscaling algo: bicubic60 AR
- downscaling algo: bicubic150 AR LL
- upscaling algo: bicubic60 AR
- NGU sharp,
- For both 720p:
- NGU sharp
- algorithm quality
- luma doubling -> high
- luma quadrupling -> direct quadruple high
- chroma -> high
- luma doubling -> high
- activate ...
- doubling -> only if 1.2x or bigger
- quadrupling -> only if 2.4x or bigger
- doubling -> only if 1.2x or bigger
- more scales
- upscaling algo -> bicubic60 AR
- downscaling algo -> bicubic150 AR LL
- upscaling algo -> bicubic60 AR
- NGU sharp
- For both 576p
- NGU Standard
- algorithm quality:
- luma doubling -> high
- luma quadrupling -> double again high
- chroma -> let madVR decide
- luma doubling -> high
- activate ...
- doubling -> let madVR decide
- quadrupling -> let madVR decide
- doubling -> let madVR decide
- more scales:
- upscaling algo -> bicubic60 AR
- downscaling algo -> bicubic150 AR LL
- upscaling algo -> bicubic60 AR
- NGU Standard
- For over all 1440p and 2160p profiles: nothing
- For 1080p:
- 30FPS: soften edges 2
- 60FPS: nothing
- 30FPS: soften edges 2
- For both 720p:
- enhanced detail 1
- LunaSharpen 0.65
- AdaptiveSharpen 0.5
- activate anti-bloating filter 100%
- activate anti-ringing filter
- refine the image after every ~2x upscaling step
- enhanced detail 1
- For both 567p:
- nothing
- nothing