FerretBomb
Active Member
OBS appears to have a latency issue with its Video Capture Device module, or the handling of video from device sources.
I've done a couple of demos, showing the differences between capture methods using three capture cards; a Datapath VisionE2, an SC512, and an Elgato HD60 Pro. All video captured 1080p@60fps on Ultrafast x264, with CPU load below 60% for the duration.
Apples to Apples - https://www.youtube.com/watch?v=sO3CDpZcM0M
All of these are using OBS' Video Capture Device source method. DPV wins. SC512 super-slow.
Best-Case - https://www.youtube.com/watch?v=znANnjC1QEM
HD60 Pro using OBS VidCapDev (baseline). DPV is using the DP Native OBS plugin. SC512 using Amarec. SC512 goes from last to first.
Rigor - https://www.youtube.com/watch?v=aM0_7zYhIfg
HD60 Pro using Elgato Game Capture HD. DPV using DP Native. DP edges out a win... barely.
(I was unable to test the HD60 Pro via Amarec as it does not present a video device for some reason; Elgato's own software should be a best-case though.)
Screenshot attached from the 'Best-Case' video demonstrating the capture delay offsets (tiles sliding upward; higher is faster capture. HD60P, DPV E2, SC512). How is it possible that a Game-Monitor Capture in OBS of a different program capturing from a cap card using DirectShow is faster than OBS getting video directly from the hardware?
I've done a couple of demos, showing the differences between capture methods using three capture cards; a Datapath VisionE2, an SC512, and an Elgato HD60 Pro. All video captured 1080p@60fps on Ultrafast x264, with CPU load below 60% for the duration.
Apples to Apples - https://www.youtube.com/watch?v=sO3CDpZcM0M
All of these are using OBS' Video Capture Device source method. DPV wins. SC512 super-slow.
Best-Case - https://www.youtube.com/watch?v=znANnjC1QEM
HD60 Pro using OBS VidCapDev (baseline). DPV is using the DP Native OBS plugin. SC512 using Amarec. SC512 goes from last to first.
Rigor - https://www.youtube.com/watch?v=aM0_7zYhIfg
HD60 Pro using Elgato Game Capture HD. DPV using DP Native. DP edges out a win... barely.
(I was unable to test the HD60 Pro via Amarec as it does not present a video device for some reason; Elgato's own software should be a best-case though.)
Screenshot attached from the 'Best-Case' video demonstrating the capture delay offsets (tiles sliding upward; higher is faster capture. HD60P, DPV E2, SC512). How is it possible that a Game-Monitor Capture in OBS of a different program capturing from a cap card using DirectShow is faster than OBS getting video directly from the hardware?
