Yes, I didn't change other settings.
Code:
clip = core.std.SetFrameProps(clip=clip, _FieldBased=vs.FIELD_BOTTOM) # bff
That is archived by overwriting the scan type.
Quote:I note you select "fast". Would slower be beneficial to quality?
Slower modes are usually not worth it and tend to smooth more unless you adjust the settings.
This might be interesting. I usually stick with 'fast' as preset and only adjust it if I notice problems.
Quote:KNLMeansCL d=0 - is this distance?
Yes
Cu Selur
I'm receiving a crash. For what it's worth, AI is providing this analysis per attached.
The crash is with an FFV1 or H.265 encode.
Note that OPENCL=true. GPU-Z confirms OPENCL is on my system.
KNLMeansCL device type was set to auto.
device type setting "accelerator" had an output but had what looks to be reference blocks.
My available setting are only Auto, GPU, and accelerator. I'm missing all the others including CPU. I would think this would be the correct setting if present.
Device type:
Select the device which should do the processing.
CPU := An OpenCL device that is the host processor.
GPU := An OpenCL device that is a GPU.
ACCELERATOR := Dedicated OpenCL accelerators.
DEFAULT := The default OpenCL device in the system.
ALL := All OpenCL devices available in the system.
(Default: DEFAULT)
One of my debug reports attached.
I confirmed:
OpenCL 3.0 installed
All graphics updated Intel UHD 770 128 MB VRAM 16 GB Shared (I can't find info on if more vram is needed
I'm going to look at bios and see if I can increase)
C++ visual is version 14.44 (link provides 14.04 but i wouldn't think there's a backward compatibility issue)
This crash seems to come from FFmpeg rejecting the video metadata from Vapoursynth during piping.
In auto or gpu, the script below suggests that cuda is being utilized.
core.std.LoadPlugin(path="C:/Program Files/Hybrid/64bit/vsfilters/DenoiseFilter/KNLMeansCL/vsnlm_cuda.dll")
When I use accelerator it runs through but i don't have an accelerator and the result is defective.
core.std.LoadPlugin(path="C:/Program Files/Hybrid/64bit/vsfilters/DenoiseFilter/KNLMeansCL/KNLMeansCL.dll")
vsnlm_cuda.dll
KNLMeansCL.dll
Is there a way to keep from loading CUDA in the core? I think its safe to say this is the problem. I can't get Hybrid to recognize KNLMeansCL.dll to load in the core.
Could you assist with manual change instruction?
vsnlm_cuda should only be loaded when an NVIDIA GPU is found
=> looking into it
I think I found and fixed the problem.
Now KNMeanCL.dll instead of vsnlm_cuda.dll should be used.
=> Uploaded a new dev.
Let me know whether that fixes the problem for you too.
Cu Selur
Auto, GPU, and Accelerator fully process, but all 3 result in the image below. Previously this result was from Accelerator mode. I'm guessing this is a mapping setup for future processing.
Upload parameters are picky which is why I've attached zip folders.