alatteri
commented
2 years ago Thinking about this some more, and doing additional testing, I believe that the issues is not the decode specifically, but the speed of R10k conversion itself. Which would explain why I am unable to capture UHD realtime using a R10K native codec no matter what powerful CPU resources I have tested with.
For instance on Ryzen 9 3950X:
UltraGrid.AppImage --tool uv -m 1316 -t decklink -c libavcodec:encoder=libx265:x265-params=slices=4:bitrate=50M --param use-hw-accel -P 5004
UltraGrid 1.7+ (tags/continuous rev 3af2d021 built Nov 9 2021 13:14:27)
Display device : none
Capture device : decklink
Audio capture : none
Audio playback : none
MTU : 1316 B
Video compression: libavcodec:encoder=libx265:x265-params=slices=4:bitrate=50M
Audio codec : PCM
Network protocol : UltraGrid RTP
Audio FEC : none
Video FEC : none
Display initialized-none
[DeckLink] Auto-choosen device 0.
Using device DeckLink 4K Pro
[DeckLink capture] Setting single link by default.
The desired display mode is supported: 525i59.94 NTSC
Enable video input: 525i59.94 NTSC
[DeckLink] Trying to autodetect format.
Start capture
DeckLink capture device enabled
Video capture initialized-decklink
Created new RTP session with SSRC 0x688212e3.
Frame received (#0) - No input signal detected
[Decklink capture] Format change detected (display mode, color space).
[Decklink capture] Capturing detected 10-bit signal, use ":codec=UYVY" to enforce 8-bit capture (old behavior).
[Decklink capture] Using codec: R10k
Enable video input: 2160p24
[lavc] Using codec: H.265, encoder: libx265
[lavc] Setting bitrate to 50000000 bps.
[lavc] Warning: Codec doesn't support slice-based multithreading.
[lavc] Setting preset to ultrafast.
x265 [info]: HEVC encoder version 3.5+1-f0c1022b6
x265 [info]: build info [Linux][GCC 7.5.0][64 bit] 10bit
x265 [info]: using cpu capabilities: MMX2 SSE2Fast LZCNT SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2
x265 [warning]: halving the quality when psy-rd is enabled for 444 input. Setting cbQpOffset = 6 and crQpOffset = 6
x265 [info]: Main 4:4:4 10 profile, Level-5 (High tier)
x265 [info]: Thread pool created using 32 threads
x265 [info]: Slices : 4
x265 [info]: frame threads / pool features : 1 / wpp(68 rows)
x265 [info]: Coding QT: max CU size, min CU size : 32 / 16
x265 [info]: Residual QT: max TU size, max depth : 32 / 1 inter / 1 intra
x265 [info]: ME / range / subpel / merge : dia / 57 / 0 / 2
x265 [info]: Cb/Cr QP Offset : 6 / 6
x265 [info]: Lookahead / bframes / badapt : 0 / 0 / 0
x265 [info]: b-pyramid / weightp / weightb : 0 / 0 / 0
x265 [info]: References / ref-limit cu / depth : 1 / off / off
x265 [info]: Rate Control / qCompress : ABR-50000 kbps / 0.50
x265 [info]: VBV/HRD buffer / max-rate / init : 16666 / 50000 / 0.750
x265 [info]: tools: rd=2 psy-rd=2.00 early-skip rskip mode=1 tmvp fast-intra
x265 [info]: tools: strong-intra-smoothing lslices=8 slices=4 deblock
[lavc] Selected pixfmt: gbrp10le
[Decklink capture] 73 frames in 5.03387 seconds = 14.5018 FPS
[Decklink capture] 77 frames in 5.06164 seconds = 15.2124 FPS
[Decklink capture] 77 frames in 5.03387 seconds = 15.2964 FPS
[Decklink capture] 77 frames in 5.01489 seconds = 15.3543 FPSAnd with VP9 this basically hits only about 6 cores:
UltraGrid.AppImage --tool uv -m 1316 -t decklink -c libavcodec:encoder=libvpx-vp9:bitrate=50M --param use-hw-accel -P 5004
UltraGrid 1.7+ (tags/continuous rev 3af2d021 built Nov 9 2021 13:14:27)
Display device : none
Capture device : decklink
Audio capture : none
Audio playback : none
MTU : 1316 B
Video compression: libavcodec:encoder=libvpx-vp9:bitrate=50M
Audio codec : PCM
Network protocol : UltraGrid RTP
Audio FEC : none
Video FEC : none
Display initialized-none
[DeckLink] Auto-choosen device 0.
Using device DeckLink 4K Pro
[DeckLink capture] Setting single link by default.
The desired display mode is supported: 525i59.94 NTSC
Enable video input: 525i59.94 NTSC
[DeckLink] Trying to autodetect format.
Start capture
DeckLink capture device enabled
Video capture initialized-decklink
Created new RTP session with SSRC 0x4fcc00ca.
Frame received (#0) - No input signal detected
[Decklink capture] Format change detected (display mode, color space).
[Decklink capture] Capturing detected 10-bit signal, use ":codec=UYVY" to enforce 8-bit capture (old behavior).
[Decklink capture] Using codec: R10k
Enable video input: 2160p24
[lavc] Using codec: VP9, encoder: libvpx-vp9
[lavc] Setting bitrate to 50000000 bps.
[lavc] Warning: Codec doesn't support slice-based multithreading.
[lavc] Warning: Unable to find suitable preset for encoder libvpx-vp9.
[libvpx-vp9 @ 0x7f5c84001a80] Application has requested 32 threads. Using a thread count greater than 16 is not recommended.
[libvpx-vp9 @ 0x7f5c84001a80] v1.10.0
[lavc] Selected pixfmt: gbrp10le
[Decklink capture] 61 frames in 5.02243 seconds = 12.1455 FPS
[Decklink capture] 54 frames in 5.02745 seconds = 10.741 FPS
[Decklink capture] 53 frames in 5.01333 seconds = 10.5718 FPS
[Decklink capture] 54 frames in 5.08645 seconds = 10.6164 FPS
[Decklink capture] 54 frames in 5.09119 seconds = 10.6066 FPS
[Decklink capture] 54 frames in 5.0686 seconds = 10.6538 FPS
[Decklink capture] 53 frames in 5.03724 seconds = 10.5216 FPS
MartinPulec
Hello,
I am outputting via decklink (UltraStudio 4K Mini ). I am using the parameter
decoder-use-codec=R10kto have the Decklink out full range 10bit (0-1023). Without that parameter, it seems to output "legal" (16-235) and everything is lifted. When usingdecoder-use-codec=R10kthough, UG drops lots of frames and complains the computer is too slow. If I do not use this parameter, the stream is decoded and displayed just fine. So I am wondering, is there a different parameter or switch I should use to have Decklink output full range, or maybe there is an optimization that could be made to that?With
decoder-use-codec=R10kSame stream WITHOUT
decoder-use-codec=R10kbut lifted output:The only difference above, is the use of the
decoder-use-codec=R10kI've tested bitrates up to 120M, and without the parameter it plays without issue.The input stream is simple
-t decklink:codec=R10k -c libavcodec:encoder=libx264:subsampling=444:bitrate=40MThe encoder is fine, no performance issues there, and consistently hits 24fps.