I'm not sure I'm posting in the right area - if it's not, mods please move to the correct place

So I often I use SVPtube 2 to play YouTube videos with interpolation. A problem has been happening where the video will freeze for a few seconds while the audio continues, and then the video comes back, but out of sync with the audio by the amount of time of the video freeze.

Also (unrelated), when I jump forward a few minutes in a video, sometimes playback stops. The counter keeps going, but it's black and silent. When this happens, I cannot get the same quality video to play again (eg. say I tried jumping forward while playing the 1080p 4.7mbps version - once it goes black I can't open that video, I have to pick 1080p 2.1mbps, or 720p 1.5mbps)

Thank you,

Michael

2

(9 replies, posted in Using SVP)

Did you mean OLED settings for tonemapping? If so, I couldn't say because I generally like the job my A8G does with HDR.

If you're talking about smoothing, if you can send 4k120 10/12 bit to your C2 (if memory serves the C2 has HDMI 2.1), what I'd do is set

120fps (of course... why else would I mention 4k120?)
Frames interpolation mode: 1m
SVP Shader: Complicated
Motion Vectors Precision: Half
Motion Vectors Grid: 6px. small 2
Decrease grid step: By two with global refinement
Search radius: small
Wide search: strongest
Width of top coarse level: I choose large, but I don't think I've noticed much of a change
Same with search radius... I don't know that I can tell it's doing anything. I suspect small search radius is best because there's a "small (fast)", but now I think I could be wrong about that, because Complicated > Average > Simple > Simple Lite > ... > ... > fastest (slow PCs)

I have some questions but don't want to commandeer or derail OP's thread, so I'm going to make another one

3

(1 replies, posted in Using SVP)

I hadn't used SVP for 4 months, and the 4px mode didn't work in nVidia Optical Flow before (it made everything look blocky and jerky). But it appears to work well now!

Is it working now?

Was it broken before? Or was it just my installation?

Hey, were you able to find the cause of what I described? Will there be a fix?

EternalStudent wrote:

Which app was that in?  I haven't used the transcoding feature much myself.

https://www.svp-team.com/wiki/Manual:SVPcode

Would be my guess.  That page mentions ffmpeg.  I don't know what options are passed down to ffmpeg, but this might be an issue for them to fix.  If I was you I'd try to see what options were set when you run things.  And what settings ffmpeg offers.

I wonder if lowering the threads to 8 from 16 would get you a little gain too?  Optimum thread counts always sounded like black magic to me.  "Take your actual core count and add 1" was something I'd seen before for video transcoding.


The problem is in the "SVP 4 Pro" app for sure, the one you'd find by typing SVP in the start menu. The video transcode function found in the Transcoding tab at the top is what I was doing, using the CPU to generate the interpolated frames at the highest quality settings (so SVP Shader: Complicated, Half pixel motion vector precision, Grid: 6 px. Small 2, Grid step: By two with global refinement, Small radius, Strongest search, Small top coarse level), and using the GPU to decode the video file for processing (Rendering device NVIDIA 3080) and to encode the processed video to 264 (H.264 AVC) by NVENC as interpolated frames were calculated by the CPU.

In describing this possible issue, the CPU has always been the limiting factor within SVP for the transcode - my GPU can encode this at up to 160fps (I found this by lowering SVP Shader, Grid, Grid Step etc. settings until the GPU saturated)

What I did to improve performance by 15% was, like you said, lowered the thread count to 8 manually, but not just any 8 or the first 8, I chose specifically to use cores # 0, 2, 4, 6, 8, 10, 12, and 14 within Windows Task Manager. Doing so is like disabling simultaneous multithreading in the BIOS for the application. This resulted in the transcode going from about 78FPS up to 90FPS, and CPU power consumption doing almost the exact same thing.

I'm not near that system right now, but I remember after adding the cores above, I also added 1, then 1 and 3, then 1, 3, and 5, checking performance each time. I think after cutting back to the proper 8 and gradually re-enabling virtual cores one at a time until 3 or 4 were available, the performance increase was less than one percent

Considering this, I very strongly believe this is a SVP software bug. 15% is a LOT of performance left on the table.
There is always the possibility that my computer is having a complicated interaction with the software though, so as always, someone with an AMD system will have to replicate the issue, record the performance, modify applicable settings, see if/how they affect performance, then review applicable code pertaining to multi-threading. They could skip diagnostics, but without the dry runs, the fix might be partial or not work

Hey,

I have an AMD 3700x and I noticed that when I changed the processor affinity in Windows Task Manager to just the odd CPUs (which are the real cores, not the SMT/HT virtual cores), core power consumption rose ~15%, along with the frame rate of the active transcode.

I thought I should let you know that SVP is probably limiting performance - at least with AMD's Zen 2 processors - treating virtual cores as real ones. Zen 3 (AMD's most recent processors) are very similar in design.

Also, Intel's newish 12th gen processors have two types of cores ("P" and "E") which are causing performance issues for many software devs

Still enjoying SVP daily. Thank you for your work!

Michael

As the title says.

Using a 3080 and 9600k, so I can't see it being caused by hardware.

I tested videos of 720 through 2160p resolutions, and no matter what I try, 24p content looks like it's playing at 24fps. I've tried reinstalling to no avail.

Is this a known issue? If so, is there a planned fix?

I want to make 24 frames per second into 60

for example, numbered frames from 24fps at 60fps are: 1, 1.33, 1.66, 2, 2.5, 3, 3.33, 3.66, 4, 4.5, 5, 5.33, 5.66, 6, 6.5 etc.

Which setting would I use to make this happen? 1m? I'm not sure, because the descriptions aren't too comprehensive (it might do what I described above, but it also might do something in addition to it - like randomize which frames get averaged twice for example. I don't know because there's not much info in the description, especially for 1.5.

Can you please describe how each frame interpolation mode turns 24fps into 60fps? For my interest, and maybe also for addition to the manual for others to see. Everyone would be able to know exactly what they are choosing from (and more quickly come to the desired result).

I think 24fps into 3 common refresh rates, say 50, 60, and 120, as well as 30fps into 50, 60 and 120 would be illuminating. 24 to 50 allows people to infer what would be done converting 24fps to 75. Doing 120 allows people to see what happens when a multiple of a framerate is chosen.


Also, random question... is making 24 into 60 with the 1m frame interpolation mode the same as doing making 24 into 120 with the 2m frame interpolation mode?
The final quality looks similar, but maybe a bit choppier with less artifacts with 2m 120 (both transcodes have artifacts masking disabled, standard SVP shader, and best CPU motion vectors options selected (half pixel 6px small 2, bu two with local refinement, small, strongest, small)

Please let me know, and what you think about the manual idea.

Also, is it common for Nvidia Motion Vectors Grid set to 4 to cause choppiness, but 8 and above work fine? Software/intelGPU I can set all the way down to "6 pixels Small 2", no problem.

Is using Nvidia optical Flow with its setting of 8 better than the customizable CPU options with 6px small 2? Is there another spot to customize Nvidia Optical Flow?

I'm relatively new to video encoding. I just bought your program to do some though, and am wondering...

I read in a couple places that a CRF setting of 17-18 makes a video that's indistinguishable from perfect for a lot of video types. I've decided to use 16 as I'm re-encoding videos and would like to preserve as much detail as possible without taking up excessive space.

Problem: I've read QuickSync makes lower quality videos compared to software. Will an Intel QuickSync transcoded video with CRF setting of 16 make a video of the same as Software with CRF 16?

If not, what's the software equivalent to QuickSync CRF 16?

Is the QuickSync limitation to CRF 16 an implementation of your softare, or a limitation of the device? Is its reputation for lower quality videos because of its minimum CRF of 16?

Thank you!

I like motion interpolation on my TV, and in my reading found your software online. I downloaded it onto my laptop (8250U) and found I could do real time interpolation to 60fps with 720p24 material, but had trouble doing 1080. I tried transcoding and found it took about 110-120% longer than playback. Ideally I don't want to transcode everything in advance, and my HTPC is faster than my laptop (and already hooked up to my tv - 9600k @4.5 with 3080), so I installed it there. I found it could do 1080 easily. I went to check how much faster my HTPC is by transcoding a 1080p movie, but I couldn't! In the drop down menu the option isn't there, and the transcoding tab is also missing! I checked the versions and they're exactly the same, and I installed them the exact same way and on the same day. Did you disable transcoding in the evaluation version with fast GPUs present? And offer different options in the "Motion vectors options"? There are like 8 on my Intel 620 graphics, 3 on Nvidia.

It seems odd that you would, so I'm wondering if I'm doing something wrong. I like your software, the way it allows you to optimize video quality by balancing artifacting and smoothness when interpolating to a new frame rate of your choice. I like 60fps for TV and movies the most -to my eye it's almost as smooth as 120 most of the time, and when it isn't, the extra smoothness of 120 is just making the artifacts twice as big and twice as obvious. Together, 4x worse. Anyway, I'll probably be buying your software, but it'd be nice to know first exactly how customizable it will be using my 3080, and how fast each PC can transcode. Am I doing something wrong, or is there an extra functional limitation for PCs with fast dedicated graphics cards?