D
Deleted member 225844
[Deleted.]
Last edited by a moderator:
were491
Member
it might be your media player screwing up decoding, that's my best guess.
other than that, then i have no idea other than trying a different computer entirely. (side note: you can use CRF instead of VBR because CRF can dynamically adjust bitrate to keep a relatively constant quality, saving bitrate in low-motion video and using more on high-motion video. a CRF of ~12-15 will probably work fine)
e: it looks like your video log says something about "2 frames left in the queue upon closing" which might be the issue. i guess you could try a fresh install of windows / obs (preferably on a different drive so you don't delete your data if it doesn't work)?
other than that, then i have no idea other than trying a different computer entirely. (side note: you can use CRF instead of VBR because CRF can dynamically adjust bitrate to keep a relatively constant quality, saving bitrate in low-motion video and using more on high-motion video. a CRF of ~12-15 will probably work fine)
e: it looks like your video log says something about "2 frames left in the queue upon closing" which might be the issue. i guess you could try a fresh install of windows / obs (preferably on a different drive so you don't delete your data if it doesn't work)?
Last edited:
Just to throw this out there... Didn't Linus mention in a couple of his recent review videos something about certain Ryzen motherboards or PCIe controllers being bad for streaming/capture? I could have sworn that he's mentioned a couple of times that it's the reason why he still uses Intel builds for his gaming tests. I think it had something to do with the USB controllers as well... maybe you could fire this one to him or some of the other top guys and see if they could think of something. I get the feeling they like challenges! ^_^
callphoenix
New Member
Hello SJP176 :)
2 questions:
- Have you tested to see if a preview of a webcam (UVB USB) is OK with no stuttering?
- Have you tried to do the preview without the filters we can see on the log?
>> 'Capture Card':
17:14:09.283: - source: 'Magewell Pro Capture HDMI 4K Plus LT' (dshow_input)
17:14:09.283: - filter: 'Scaling/Aspect Ratio' (scale_filter)
17:14:09.283: - filter: 'Render Delay' (gpu_delay)
2 questions:
- Have you tested to see if a preview of a webcam (UVB USB) is OK with no stuttering?
- Have you tried to do the preview without the filters we can see on the log?
>> 'Capture Card':
17:14:09.283: - source: 'Magewell Pro Capture HDMI 4K Plus LT' (dshow_input)
17:14:09.283: - filter: 'Scaling/Aspect Ratio' (scale_filter)
17:14:09.283: - filter: 'Render Delay' (gpu_delay)
callphoenix
New Member
I see...
The PC you stream/record with is the same you want to cast?
>> Be cool with the answer, we're here to help and I don't care of the money (don't want). Like I'll not answer first or better to one because of money. This questions sometimes seems to be irrelevant but not necessarily bad.
Tomasz Góral
Active Member
Hi,
A you try save on lower resolution ?
- change audio sample rate to one value (44,1 or 48kHz), now default value is 48kHz
- set all moniotrs to one refresh rate (you have 120 and 60, set one lower)
- test recording on 720p, 1080p, 2160p and fps 30 and 60
A you try save on lower resolution ?
- change audio sample rate to one value (44,1 or 48kHz), now default value is 48kHz
- set all moniotrs to one refresh rate (you have 120 and 60, set one lower)
- test recording on 720p, 1080p, 2160p and fps 30 and 60
Tomasz Góral
Active Member
The problems is in hardware or drivers, because the problem occurs in various programs, i think some input signal is worst capture like typical resolution).
Your input signal is ? (resolution, fps).
Your input signal is ? (resolution, fps).
Tomasz Góral
Active Member
Input signal is from device (name) ?
Tomasz Góral
Active Member
Looks good, device send audio 48Khz, set correct audio in grabber.
it's possible to test on other grabber card?
it's possible to test on other grabber card?
callphoenix
New Member
H
Hey, hi again SJP176 :)
Are you ok to test? If ok here what you can do to pinpoint the problem:
You use 3 monitors connected to your NVIDIA GeForce RTX 2080 Ti:
-Acer XB273K S
-Samsung (something)
-BenQ EL2870U
Try to create 3 separate scenes and in every scene put one of the screens (as source - screen capture mode), like so:
Scene: ACER / Source: acer
Scene: SAMSUNG / Source: samsung
Scene: BENQ / Source: benq
Now play a video on your Acer monitor and see if the preview of that one on OBS is stuttering and then do the same with the 2 others and if one is positive, then check how the monitor is managed by your system (perhaps by the software guiding your GC / Windows and sometimes even the BIOS)
There's more but this has to be out of the game :)
Hey, hi again SJP176 :)
Are you ok to test? If ok here what you can do to pinpoint the problem:
You use 3 monitors connected to your NVIDIA GeForce RTX 2080 Ti:
-Acer XB273K S
-Samsung (something)
-BenQ EL2870U
Try to create 3 separate scenes and in every scene put one of the screens (as source - screen capture mode), like so:
Scene: ACER / Source: acer
Scene: SAMSUNG / Source: samsung
Scene: BENQ / Source: benq
Now play a video on your Acer monitor and see if the preview of that one on OBS is stuttering and then do the same with the 2 others and if one is positive, then check how the monitor is managed by your system (perhaps by the software guiding your GC / Windows and sometimes even the BIOS)
There's more but this has to be out of the game :)
callphoenix
New Member
️ I know that feeling... Careful it makes computers fly (not for a long time) ️
Well... The last one is the killer...
You completely get rid of your actual OS and get a fresh install with an Ubuntu (GNU/Linux distro)... This is how I stream and record screencasts ️
I know that kind of situation can be very frustrating... Better laugh at it, life is too short...
You've got a very good setup, if the hardware is tested OK, then it's somehow the software... I know a case where the guy (a tec one) coudn't find any reason for the strange behaviour with his 2 monitors (1 Thomson & 1 Iiyama) and the last thing was to unplug all and test one at a time... And guess what... The faulty was the PLC he was connected on for both network and power...
koala
Active Member
A fascinating issue. I downloaded your first video from Youtube. I watched and saw occasional stutter now and then, but wasn't able to put my finger on what actually stuttered. Then I single-stepped frame by frame for the first 22 seconds. There was no duplicated frame, no lost frame. The background scrolled one pixel every other frame frame and the foreground scrolled one pixel every frame. Absolutely constant. For every single frame for the first 22 seconds. These should be flawless, but I visually observed stutter. If I rewatched, the stutter was different - at a timestamp I didn't observe stutter before, and vice versa.
koala
Active Member
The frames are perfect. Single step through every frame, there is not a single frame lost or duplicated - this is usually the source of stutter. Every frame scrolls the foreground by one pixel and every other frame scrolls the background by one pixel.
I assume this is some kind of sync failure between video and monitor refresh rate in the player. In the player, not in the recording process.
However, I tried on a 60 Hz monitor with vsync (slight stutter every few seconds), on a 144 monitor with gsync (bad stutter almost all the time), without gsync (the same), the 144 monitor to fixed 60 Hz without gsync (same slight stutter as on the 60 Hz monitor).
A video contains timestamps for each frame, that means it is a command to the player how long each frame must be shown. For a constant frame rate video, the timestamps must be increasing at exactly the same rate - at 1/60 s for a 60 fps video.
I extracted the timestamps from the video and built diffs between them. They are constant - almost. The frame intervals diff in 1 ms, but the average over 3 frames is 0. The first frame is in time, the next frame is 1 ms late, the next frame is 1 ms advanced, and the 4th frame is in time again. This sequence is repeated for 5 seconds, then there are 2 consecutive frames in time instead of 1, then the pattern is repeated. Might be this results in one stutter every 5 seconds. These by 1 ms variant intervals seem the result of rounding: 0.333 ms would be the correct timestamp diff, but in the video the precision is milliseconds, not microseconds. (0.333 ms are 333 us).
If this is expected for every 60 fps video, I don't know. Usually, I don't observe such lags in other material. May be the absolutely constant scrolling is making such lags so visible.
Unfortunately, I cannot really help if there is something to tune to make it go away.
I assume this is some kind of sync failure between video and monitor refresh rate in the player. In the player, not in the recording process.
However, I tried on a 60 Hz monitor with vsync (slight stutter every few seconds), on a 144 monitor with gsync (bad stutter almost all the time), without gsync (the same), the 144 monitor to fixed 60 Hz without gsync (same slight stutter as on the 60 Hz monitor).
A video contains timestamps for each frame, that means it is a command to the player how long each frame must be shown. For a constant frame rate video, the timestamps must be increasing at exactly the same rate - at 1/60 s for a 60 fps video.
I extracted the timestamps from the video and built diffs between them. They are constant - almost. The frame intervals diff in 1 ms, but the average over 3 frames is 0. The first frame is in time, the next frame is 1 ms late, the next frame is 1 ms advanced, and the 4th frame is in time again. This sequence is repeated for 5 seconds, then there are 2 consecutive frames in time instead of 1, then the pattern is repeated. Might be this results in one stutter every 5 seconds. These by 1 ms variant intervals seem the result of rounding: 0.333 ms would be the correct timestamp diff, but in the video the precision is milliseconds, not microseconds. (0.333 ms are 333 us).
If this is expected for every 60 fps video, I don't know. Usually, I don't observe such lags in other material. May be the absolutely constant scrolling is making such lags so visible.
Unfortunately, I cannot really help if there is something to tune to make it go away.
callphoenix
New Member
You can try to use a live bootable USB key with a GNU/Linux distribution on it... Like Ubuntu, so you don't mess with your actual setup... From there you can install OBS (to USB key or in RAM) and then see if the stuttering is still occurring... Then you can blame your motherboard, your GC or even your CPU (but don't, I kinda like AMD ️)
D
Deleted member 121471
Tested it, only stutters/ghosts if captured FPS isn't a divisor of my refresh rate. Recording it had the same results.
For reference, using 2x 144Hz monitors, freesync setting made no difference.
For reference, using 2x 144Hz monitors, freesync setting made no difference.
D
Deleted member 121471
After reading the first post, I was wondering why are you capturing PC gameplay with a capture card over just using NVENC? Did I misread it?
Adega
Member
I had a similar problem with jerking. I cured it like this: 1.Fix exactly 60 fps in the nvidia panel and it is forbidden to use the low latency mode!
2. In the game, enable vertical sync.
I understand that the obs works at 60.00000 fps, and my V-sync monitor works in the 59-61 range. More precisely 60.004 and caused video stuttering. And the low latency mode sets the minimum frame size and the obs immediately sees the micro-freezes. Hope it helps you) Good luck!
2. In the game, enable vertical sync.
I understand that the obs works at 60.00000 fps, and my V-sync monitor works in the 59-61 range. More precisely 60.004 and caused video stuttering. And the low latency mode sets the minimum frame size and the obs immediately sees the micro-freezes. Hope it helps you) Good luck!
idreaminhd
New Member
Have you tried replacing any cables(psu cables, hdmi cables, etc)? Can you include a photo of your setup?
energizerfellow
New Member
This may also be a lack of NUMA awareness in OBS (along with many other applications I'd imagine...), per the description here:
ideas.obsproject.com
Any AMD Zen CPU with >8 cores are multi-CCD, so will likely need NUMA memory code in OBS to truly fix.
That said, however, AMD's AGESA BIOS should have the CCD Control and Core Control options to limit the CPU to a single CCD for testing.
Improve NUMA awareness in OBS Studio · OBS Studio Ideas and Suggestions
OBS Studio is currently completely NUMA unaware, resulting in sub-par performance on NUMA (or CCD) based systems. For these systems, it is extremely
ideas.obsproject.com
Any AMD Zen CPU with >8 cores are multi-CCD, so will likely need NUMA memory code in OBS to truly fix.
That said, however, AMD's AGESA BIOS should have the CCD Control and Core Control options to limit the CPU to a single CCD for testing.
Last edited:
energizerfellow
New Member
Additional information on the NPS / CCD Control / Core Control options on AMD and the (lack of) inter-core performance:
www.anandtech.com
Interestingly, Intel CPUs of 6+ cores are also affected since they're built of two separate 3+ core sides.
AMD Zen 3 Ryzen Deep Dive Review: 5950X, 5900X, 5800X and 5600X Tested
Interestingly, Intel CPUs of 6+ cores are also affected since they're built of two separate 3+ core sides.