Waylandvk runs fine for 720p but once I try 1080p the video starts stuttering even with all shaders disabled. I have a RX560. Same problem across multiple distros and I'm using git master ffmpeg libplacebo mpv. what do
Hunter Lee
Have you checked your resource usage? Maybe your CPU can't keep up the decoding, in which case you might want to try hwdec. I think you could also enable some shaders in that case. Or maybe it can't cache fast enough, which would be a disk/network issue, though I doubt that.
CPU usage is vo/gpu: Failed initializing any suitable GPU context! >cplayer: Error opening/initializing the VO window.
Aiden Rogers
By the way x11vk works without a problem
Ryan Wright
I have this in my mpv.conf since waylandvk gives artifacts sometimes # When running on Wayland # gpu-context=waylandvk # vulkan gpu-context=wayland # opengl
I really don't like that there's subjective choices in the OP advertised as "advanced setup". Yes, it's better than lanczos, but who says it's better than spline36, or haasnsoft, or ewa_hanning? I'd just omit any specific cscale.
Jacob Cruz
I plan on using this with Raspbian, how is the support? Especially with the hardware encoding the PI 4 has.
Gabriel Walker
>ewalanczorsharp A slightly sharpened version of ewa_lanczos, preconfigured to use an ideal radius and parameter. If your harware can run it, this is probably what you should use by default.
Cooper Perez
...still no gui??
William Brooks
>corrupts your playback
Noah Ramirez
That entry is about luma scaling though.
Zachary Reed
>but who says it's better than spline36, or haasnsoft, or ewa_hanning? Literally one person. Some guy recommended it a few threads ago and next time it was included in the OP. >I'd just omit any specific cscale. So you'd recommend spline36 for most people here?
>So you'd recommend spline36 for most people here? No, I'd recommend that for people that want a "it just werks" solution. I'd recommend people to try out different scalers and pick the one they like. And see if krigbilateral runs, in which case you should use that. But I don't think that should be in the OP.
Jose Phillips
retard scalers on every thread, don't expect that shit to change.
Anthony Baker
Why are you doing this? Why copy the reply chain from the other thread.
Robert Sullivan
>Still no neural network scalers? Look for FSRCNN-TensorFlow.
Don't steal my posts. Besides, that would be based on the "objective" manual which states that as the one you should use. Also, it comes out the best in objective upscale testing.