Beiträge von dream

Hey,

Please excuse my somewhat limited knowledge, but I have a few thoughts and suggestions regarding AV1_NVENC and potential improvements in FFMPEG. I might be overlooking something, so feel free to correct me or expand on these points!

1. ConstQP Rate Control in AV1_NVENC
   When using the constqp rate control mode in AV1_NVENC, I’ve noticed that the QP value can only be set between 0 and 51. However, according to the FFMPEG 7.0 documentation, this value should theoretically go up to 255.
   Is there a reason why this range is limited, or am I missing something here?
2. VBR Rate Control and CQ Values
   Similarly, in the VBR rate control mode, the CQ values are also restricted to a range of 0 to 51, whereas they should actually be configurable between 0 and 63. Is this a bug or an intentional limitation?
3. Setting B-Frames
   I might have overlooked this, but is it possible to configure certain general FFMPEG parameters like B-Frames in AV1_NVENC or other encoders? From what I understand, in FFMPEG, you would typically set B-Frames using the -bf <int> parameter, but I haven’t been able to figure out how to adjust this in Voukoder Pro.

4. Split-Frame Encoding
   Since the release of NVIDIA SDK 12.1 (and I believe FFMPEG 7.0), there’s now support for split-frame encoding, which allows GPUs with multiple NVIDIA chips to encode in parallel by manually forcing this feature for any resolutions and presets instead of only the automatically selectet (I think it was above 4k and p1,p2 but not sure on that).

   
   In my tests, this significantly improved performance, almost doubling the encoding frame rate! For instance, with split-frame encoding enabled, I achieved around 250 FPS at 2560x1440 resolution using the P7 preset (Very Slow), compared to only 120 FPS without it.

   FFMPEG now supports this feature for both AV1_NVENC and HEVC_NVENC, and the command looks like this:

   Code

   -split_encode_mode <int>

   With the following options:

   • disabled: 15 – Disabled for all configurations
   • auto: 0 – Enabled or disabled based on the preset and tuning
   • forced: 1 – Enabled with horizontal strip selection by the driver
   • 2: 2 – Forces two horizontal strips when NVENCs > 1
   • 3: 3 – Forces three horizontal strips when NVENCs > 2

   For split-frame encoding to work properly and fully utilize the NVENC chip, it was crucial to set the following parameters in FFMPEG:

   Code

   -hwaccel cuda -hwaccel_output_format cuda

5. Command Line Output in Voukoder Pro
   In the older versions of Voukoder, there was a feature that displayed the FFMPEG command line at the bottom, based on the selected options in the GUI. This was really useful for summarizing the settings in one simple line, especially given how many menu options there are for certain encoders.
   Would it be possible to reintroduce this as an optional feature? It helped a lot when working through complex configurations without needing to scroll through a ton of settings to see at one glance what you have configured.

These are just some observations and suggestions based on my experience. I think Voukoder might already consider some of those things, but im not sure.

Just wanted to ask if the split encoding thing is in one of the newest Voukoder versions?

I see that the Pro Version use SDK 12.2

Is this topic about the dual encoding aka split encoding where two of the 40 series chips getting used to encode av1?

In the video codec SDK 12.1 nvidia allowed developer to implement a flag to make use of the feature on any resoultion or preset:

New Video Creation and Streaming Features Accelerated by the NVIDIA Video Codec SDK | NVIDIA Technical Blog

For over a decade, NVIDIA GPUs have been built with dedicated encoders and decoders called NVENC and NVDEC. They have a highly parallelized architecture,…

developer.nvidia.com

I don't know if this is already implemented but would be awesome to have.

Zitat von Vouk

Can you try updating DVR to 18.5.1 build 6 (Latest version) and see if this is fixed? I don't have control over that setting. DVR decides if it shows it or not.

Still not showing.

Mainly in DaVince Resolve Studio but sometimes im verry happy that it has support for Virtualdub.

Zitat von Abd

I found a bug: the latest 13 or 12.1, in Davinci studio 18.5beta or 18.5, the data level (Auto, Video, Full) is missing. However, in Studio 18.1.4, it is ok.

Same thing for me. Im using 13.1 Voukoder with the latest connector.

Also happens when i don't try to use an preset i saved before.

As soon as i select vokouder as format the data level option is disappearing

Since the topic is over 1 year old and i was reminded about it because the new OBS beta version has
udated to FFMPEG 4.4.1.

I wanted to ask carefully if their are any news about this feature for Voukoder

Ahh okay.

Are there any plans to bring it back in the future?

Hello Vouk,

in the current ffmpeg build (i think it's 4.3.1) we have new presets for "hevc_nvenc" and i think some old ones got renamed.

Some of those presets give you even better quality than slow like the new "p7" preset (but of course they're slower).

Would be nice if you could add them, when you do the next ffmpeg update for voukoder.

ffmpeg -h encoder=hevc_nvenc:

Sehr gut! Mit 0.5.1 funktioniert es.

Vielen Dank!

Mit der neuen Version, besteht das Problem bei mir nach wie vor.

Format                                   : Matroska
Format version                           : Version 4
File size                                : 285 MiB
Duration                                 : 1 min 0 s
Overall bit rate mode                    : Variable
Overall bit rate                         : 39.6 Mb/s
Writing application                      : Voukoder (VEGAS)
Writing library                          : Lavf58.35.101
ErrorDetectionType                       : Per level 1

Video
ID                                       : 1
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : Baseline@L4.2
Format settings                          : 1 Ref Frames
Format settings, CABAC                   : No
Format settings, Reference frames        : 1 frame
Codec ID                                 : V_MPEG4/ISO/AVC
Duration                                 : 1 min 0 s
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Constant
Frame rate                               : 60.000 FPS
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Writing library                          : x264 core 157 r2981 b5bc5d6
Encoding settings                        : cabac=0 / ref=1 / deblock=0:0:0 / analyse=0:0 / me=dia / subme=0 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=0 / me_range=16 / chroma_me=1 / trellis=0 / 8x8dct=0 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=0 / threads=24 / lookahead_threads=4 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=0 / weightp=0 / keyint=250 / keyint_min=25 / scenecut=0 / intra_refresh=0 / rc=crf / mbtree=0 / crf=23.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40 / aq=0
Default                                  : Yes
Forced                                   : No

Audio
ID                                       : 2
Format                                   : FLAC
Format/Info                              : Free Lossless Audio Codec
Codec ID                                 : A_FLAC
Duration                                 : 1 min 0 s
Bit rate mode                            : Variable
Channel(s)                               : 2 channels
Channel layout                           : L R
Sampling rate                            : 44.1 kHz
Bit depth                                : 16 bits
Compression mode                         : Lossless
Default                                  : Yes
Forced                                   : No

Moin,

hast du hier schon mehr untersuchen können.

Thank you!

With Voukoder 2.3 my lossless avi files get the correct colors, because i can use this code from ffmpeg in voukoder:

ffmpeg -i "G:\test.mkv" -vf "scale=in_range=pc:out_range=tv" -pix_fmt yuv420p -c:v libx264 -preset ultrafast -crf 21 Output.mkv

Now i get the correct colors in the output with

1. .mp4 videos that i record with nvenc (There were always correct, so no need to use a filter)
2. Lossless .avi (magicyuv, utvideo) files recored with dxtory, obs etc. when i use the Voukoder filter colorspace.

colorspace filter setting for lossless .avi files recored in limited range:

Output

• Range: Limited
• Matrix: BT.709
• Primaries: BT.709
• Transfer: BT.709

Input:

• Range: Full (The important one)
• Matrix: BT.709
• Prismaries: BT.709
• Transfer: BT.709

Thank you for your excellent work and fast responses.

2.3rc1 is working. Thank you!

Anbei der Log.

2019-11-29.txt

Still the same problem in the 2.3beta7

Thank you, its working!

When using the Scale (Z-Lib) filter the problem still exists, with 2.3beta6 and vegas-connector-0.4.5

Ohh can you please try that with lower values like:

-qmin 0 -qmax 16?

Instead of adding -b:v 0 a simpler "fix" would be to allow these values to be set to 0, what give me the correct result, when i'm testing in FFMPEG:

FFMPEG Example: