Hi,
I recently got some .webm-overlays for my stream. I'm getting quite some "Skipped frames due to encoding lag" since using them. According to my hardware specs, I would totally assume, that my streaming PC should be powerful enough to stream with x264 / medium / 6000kbit/s. I have no such issues without the "facecam" overlay. I would love to get some second opinion, if that PC should do fine (and maybe some settings or the video files need optimization), or if I should throw more computing power at my stream.
OBS log file (medium preset, while using .webm media): https://obsproject.com/logs/8RXydvHxzsL6qB8a -> encoding lag
OBS log file (medium preset, without using .webm-media for facecam frame): https://obsproject.com/logs/zIRrRaocfdklC8lU -> basically no encoding lag
So I also tried "fast" preset. I get a almost no encoding lags, but occasionally I still get some encoding lag.
OBS log file (x264 "fast"-preset, while uising .webm media): https://obsproject.com/logs/S1pGulWsTQGqY6m4
More details about the configuration:
PCs: I'm streaming to Twitch with OBS using a 2-pc-setup (Elgato 4k60 capture card). The streaming PC has a AMD Ryzen 2600 CPU, 8GB RAM and a Nvidia GTX 1050ti GPU.
Gameplay: I'm playing Playerunknown's Battlegrounds at the moment - so fast paced first-person shooter - at 2560x1440. Capturing with the capture card at 60 fps.
Resolution: The base canvas is 2560x1440, but I output to twitch with 1280x720.
Overlay: The overlay is a 1280x720 .webm file (~1mb filesize) with alpha channel. Codec is VP8. I also tried .mov files (Quicktime RLE) with various resolutions, e.g. 1280x720 (50mb file), but that's even worse. Way more encoding lag with uncompressed .mov-files.
Background: I also do have another .webm (~1mb filesize) video file, which I sometimes use on a facecam-scene. No alpha channel. Also 1280x720, VP8. I don't notice any encoding lags with this background and my camera source.
Scenes: I do use nested scenes in the following way. I have one "[FACECAM]" scene which just contains the "video capture device" source for capturing the camera. The camera has the original size of 1920x1080, but I resize it to the full 2560x1440 (to match the scene size). I have another scene "[FACECAM /w Frame] 16:9 blackwhite" which contains the "[FACECAM]" scene. It also adds the "Overlay" .webm-Media source as a some sort of frame around the scene. The "[FACECAM]" scene is a bit resized to fit into that overlay/cam frame.
I also have "color correction" filter applied to the Media source to make it black-white (saturation= -1.00). I also tested without that filter, but no change. On all scenes where I want my camera to appear, I'm adding the "[FACECAM /w Frame] 16:9 blackwhite" scene as a source and resize it to my needs. Now while writing this down, I feel like my resizing might be a bottleneck. So I tried to re-do the scenes roughly but with less resizing of the sources/nested scenes. Seems to work better, but still getting encoding lags. Log with x264 preset "fast" and less stupid resizing: https://obsproject.com/logs/ZSlyxeskDRy3eCUU)
Putting the video-capture device and the media source directly into the scene leads to roughly the same result.
TL;DR: Is my PC (Ryzen 2600) not powerful enough to handle .webm media sources? Do nested scenes need more computing power? Is resizing nested scenes multiple times dumb?
I appreciate any input to this, thanks :)
I recently got some .webm-overlays for my stream. I'm getting quite some "Skipped frames due to encoding lag" since using them. According to my hardware specs, I would totally assume, that my streaming PC should be powerful enough to stream with x264 / medium / 6000kbit/s. I have no such issues without the "facecam" overlay. I would love to get some second opinion, if that PC should do fine (and maybe some settings or the video files need optimization), or if I should throw more computing power at my stream.
OBS log file (medium preset, while using .webm media): https://obsproject.com/logs/8RXydvHxzsL6qB8a -> encoding lag
OBS log file (medium preset, without using .webm-media for facecam frame): https://obsproject.com/logs/zIRrRaocfdklC8lU -> basically no encoding lag
So I also tried "fast" preset. I get a almost no encoding lags, but occasionally I still get some encoding lag.
OBS log file (x264 "fast"-preset, while uising .webm media): https://obsproject.com/logs/S1pGulWsTQGqY6m4
More details about the configuration:
PCs: I'm streaming to Twitch with OBS using a 2-pc-setup (Elgato 4k60 capture card). The streaming PC has a AMD Ryzen 2600 CPU, 8GB RAM and a Nvidia GTX 1050ti GPU.
Gameplay: I'm playing Playerunknown's Battlegrounds at the moment - so fast paced first-person shooter - at 2560x1440. Capturing with the capture card at 60 fps.
Resolution: The base canvas is 2560x1440, but I output to twitch with 1280x720.
Overlay: The overlay is a 1280x720 .webm file (~1mb filesize) with alpha channel. Codec is VP8. I also tried .mov files (Quicktime RLE) with various resolutions, e.g. 1280x720 (50mb file), but that's even worse. Way more encoding lag with uncompressed .mov-files.
Background: I also do have another .webm (~1mb filesize) video file, which I sometimes use on a facecam-scene. No alpha channel. Also 1280x720, VP8. I don't notice any encoding lags with this background and my camera source.
Scenes: I do use nested scenes in the following way. I have one "[FACECAM]" scene which just contains the "video capture device" source for capturing the camera. The camera has the original size of 1920x1080, but I resize it to the full 2560x1440 (to match the scene size). I have another scene "[FACECAM /w Frame] 16:9 blackwhite" which contains the "[FACECAM]" scene. It also adds the "Overlay" .webm-Media source as a some sort of frame around the scene. The "[FACECAM]" scene is a bit resized to fit into that overlay/cam frame.
I also have "color correction" filter applied to the Media source to make it black-white (saturation= -1.00). I also tested without that filter, but no change. On all scenes where I want my camera to appear, I'm adding the "[FACECAM /w Frame] 16:9 blackwhite" scene as a source and resize it to my needs. Now while writing this down, I feel like my resizing might be a bottleneck. So I tried to re-do the scenes roughly but with less resizing of the sources/nested scenes. Seems to work better, but still getting encoding lags. Log with x264 preset "fast" and less stupid resizing: https://obsproject.com/logs/ZSlyxeskDRy3eCUU)
Putting the video-capture device and the media source directly into the scene leads to roughly the same result.
TL;DR: Is my PC (Ryzen 2600) not powerful enough to handle .webm media sources? Do nested scenes need more computing power? Is resizing nested scenes multiple times dumb?
I appreciate any input to this, thanks :)