[FFmpeg-user] Poor quality HEVC from nvenc

I am experimenting with hevc_nvenc and I am having trouble getting a high quality image with VBR mode. I tried giving it a super high bitrate of 100Mbps but this only encodes 29Mbps, which makes me think I am missing some important setting. How can I make it use more bits, and are there any

[FFmpeg-user] Measuring individual image components

Does ffmpeg have any capability to separate image components for analysis? For example, given a JPEG2000 image encoded as xyz12le, I want to find out the size of each of the three components in Bytes. I could not find anything about this in the documentation, so I'm guessing this is not

[FFmpeg-user] PSNR for lossless JPEG2000 doesn't match

Hello. I’m trying to encode a v210 AVI to lossless JPEG2000. I believe libopenjpeg defaults to the lossless profile. When I use -psnr while encoding, it shows inf, that’s good. But when I run -filter_complex psnr on the encoded file it shows different values. From the documentation, I know

[FFmpeg-user] map_channel without re-encoding

Is this possible? I have a WAV file with a single track containing 8 channels. I want to convert it to an MOV with a 5.1 track and a stereo track. The below command works, but if I use -c:a copy then it makes both output tracks 7.1. ffmpeg -I input.wav -vn -c:a pcm_s24le -map 0:0 -map 0:0

[FFmpeg-user] libx264 supported pixel format

Hi all. I have ffmpeg installed from MacPorts and apparently x264 does not support yuv420p! I can probably fix this by recompiling myself, but I’m curious how did they remove support for this common pixel format? I don’t see anything relevant in the configuration listed. $ ffmpeg -h

Re: [FFmpeg-user] FFplay network stream with low latency

> This is just duplicating the latency. Ok, understood, so I will avoid piping. > > As said, the first thing to check is the actual framerate: > If it really is 30fps, this has to be fixed first. I did change the stream, so now ffmpeg detects it correctly at 30 tbr, did you see my last

Re: [FFmpeg-user] FFplay network stream with low latency

I discovered an option in picamera python module to include the framerate in the SPS headers, so now I will use that instead of raspivid. Plus it’s more flexible, allowing for more advanced text overlays. Now ffplay detects it correctly as 30 tbr, although it shows 60 tbc and it shows

Re: [FFmpeg-user] FFplay network stream with low latency

> Are you sure about the framerate? > If it really is 30fps and FFmpeg reports it as 25fps, then you > need a filter chain to speed playback up, "-r" only works with > ffmpeg, see the documentation for setpts: > I’m pretty sure. The stream comes from a raspberry pi like this: raspivid -n -ih -t

[FFmpeg-user] FFplay network stream with low latency

I’m trying to use ffplay with a network stream from netcat. It’s raw h.264, 1280x720, 30fps. When I play it with ffplay, there is about 10 seconds latency over the LAN. Is there any way to reduce this? I’ve tried -probesize 32 and —fflags nobuffer. I notice it plays at 25fps, so maybe

Re: [FFmpeg-user] Use multiple lavfi input generators

:sample_rate=48000 -c:a pcm_s24be ffmpeg.mov > On Jul 23, 2018, at 2:02 PM, Elliott Balsley wrote: > > I’m having trouble figuring out the syntax for this. I want to generate a > test file using video testsrc and audio sine wave. I can do both separately, > but how can I ge

Re: [FFmpeg-user] ProRes XQ 12-bit support

> On Apr 27, 2018, at 4:22 PM, Carl Eugen Hoyos wrote: > > We did and we stay corrected: > https://trac.ffmpeg.org/ticket/7163 > Thank you, Carl. ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org

Re: [FFmpeg-user] ProRes XQ 12-bit support

> On Apr 25, 2018, at 1:14 AM, Paul B Mahol wrote: > > If you want definite proof, provide export of 12 bit prores into some > other 12 bit format > or 16bit one and give it to compare with ffmpeg decoded output. Thanks for that suggestion. I have just done a thorough test

Re: [FFmpeg-user] ProRes XQ 12-bit support

On Apr 25, 2018, at 01:11, Paul B Mahol <one...@gmail.com> wrote: > > On 4/25/18, Gonzalo Garramuno <ggarr...@gmail.com> wrote: >> >> >> El 24/04/18 a las 18:32, Elliott Balsley escribio: >>> Here are a few sources stating there is a 12-bit code

Re: [FFmpeg-user] ProRes XQ 12-bit support

>> Looking at the source code here, I see that PRORES_BITS_PER_SAMPLE is >> defined as 10. So it sounds like ffmpeg simply decodes all ProRes as >> 10-bit. I would like to request the option to decode ProRes with 12-bit >> precision, but unfortunately I do not have any programming knowledge to

Re: [FFmpeg-user] ProRes XQ 12-bit support

>> Can it be possible for ffmpeg to misinterpret the image as 10-bit instead of >> 12, and still make the image look correct? > > No. I think this actually could be possible. If you ignore the two least significant bits, then the image would still look correct, just lacking precision.

Re: [FFmpeg-user] ProRes XQ 12-bit support

> > Two or more wrongs do not make one good. Sorry, I’m not understanding your point. The question I’m trying to understand is, how are pixels actually encoded? Is each pixel represented by a triplet of RGB / YUV values? If so, is there some way to look at it in a text editor and see

Re: [FFmpeg-user] ProRes XQ 12-bit support

> > Or more likely a marketing issue. I think maybe what is happening is that since Resolve processes images in a 32-bit space, and it’s not convenient to fit 12-bit words, it pads each word to 16-bits so then it fits better. Anyway, on this topic, is there a way to definitely read the bit

Re: [FFmpeg-user] ProRes XQ 12-bit support

> You sure this displays alpha when reopened in exporter? > > It can just mark that source format had alpha. > __ No I'm not, in fact I don't expect there to be an alpha channel here. I only think that because Resolve detects it as 16-bit. It could be a

Re: [FFmpeg-user] ProRes XQ 12-bit support

Sample clip here, this is 6MB. https://drive.google.com/file/d/1hdBmoCgn_M-W-8siZGlEpI1oPNTCgvO_/view?usp=sharing ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or

[FFmpeg-user] ProRes XQ 12-bit support

According to the white paper, ProRes files can contain 12-bit RGB video with an optional 16-bit alpha channel. Is there some way to tell whether this alpha channel exists? I believe this video clip is 12-bit RGB with a 16-bit alpha channel, but I am trying to verify that. Ffprobe calls it

Re: [FFmpeg-user] Lavfi video generator slow performance

> Why do you convert to rgb48? > Do you understand that this cannot improve quality? Yes of course. I don’t care about the quality, but I need sample files in this format to test my other program. > The mandelbrot filter is slower than the color filter, yes. > (asm optimization patch likely

Re: [FFmpeg-user] Lavfi video generator slow performance

> If you add " -f null -" to the above, you can test if the disk > is the bottleneck (see "fps" in the status line). > (The command line is supposed to occupy >= 8 cores) Thanks for the advice everyone. I was wrong about the RAID speeds, it does seem like disk write is the bottleneck in this

Re: [FFmpeg-user] Lavfi video generator slow performance

Those are good questions, thank you. I have 8 SSDs in RAID5 on a LSI 9361 SAS controller, formatted as NTFS. I got the 3000 number from AJA System Test. That test was writing a single file, so probably performance would drop when writing lots of small files. I will benchmark again with

Re: [FFmpeg-user] Lavfi video generator slow performance

Yes, this RAID can write at 3000MBps. > On Apr 5, 2018, at 17:22, Kieran O Leary <kieran.o.le...@gmail.com> wrote: > > Hi > >> On Fri, 6 Apr 2018, 01:11 Elliott Balsley, <elliottbals...@gmail.com> wrote: >> >> Thank you, with that suggesti

Re: [FFmpeg-user] Lavfi video generator slow performance

erhead: unknown Exiting normally, received signal 2. (I canceled it with Ctrl-C) On Wed, Apr 4, 2018 at 4:08 PM, Carl Eugen Hoyos <ceffm...@gmail.com> wrote: > 2018-04-05 0:56 GMT+02:00, Elliott Balsley <elliottbals...@gmail.com>: > > > ffmpeg -f lavfi -i color=color=red -t

[FFmpeg-user] Lavfi video generator slow performance

Hello. I'm trying to use ffmpeg to generate test images, and I'm not getting very fast performance. Is there any way to improve speed? If Lavfi is the problem, then is there any other high-performance virtual input I can use instead? I would actually prefer to use something more complex, like

Re: [FFmpeg-user] Scaling legal video levels

. > On Feb 17, 2017, at 01:08, Carl Eugen Hoyos <ceffm...@gmail.com> wrote: > > 2017-02-17 2:39 GMT+01:00 Elliott Balsley <elliottbals...@gmail.com>: >> I am trying to convert a single-frame H.264 MKV to a JPEG image. >> Unfortunately the output is always clipp

[FFmpeg-user] Scaling legal video levels

I am trying to convert a single-frame H.264 MKV to a JPEG image. Unfortunately the output is always clipped at 235. The source video is full-range, and I want to keep it full-range. Sample attached to this email. The clipping can be seen quite obviously with this test pattern. If I instead

[FFmpeg-user] Can't stop recording

I'm trying to capture a stream from a webcam, and it says "press [q] to stop, [?] for help". But neither of those keys has any effect. The only way I can get it to stop is by pressing ctrl-C repeatedly, and then it gives an error and the output file is not playable. I'm running ffmpeg in

Re: [FFmpeg-user] speed display during encode

Nope. -loglevel warning disables progress display, so I re-enable it with -stats. ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-user

Re: [FFmpeg-user] speed display during encode

I realize that; I was merely asking if the display can be disabled according to user preference. When using a narrow Terminal window, the display is constantly overwritten, so this new change means a wider Terminal window of 100 chars is required. ___

[FFmpeg-user] side data type

Since upgrading ffmpeg to a December build, I am getting a warning "unknown side data type 10 (24 bytes)". What does this mean? I'm using the MacPorts builds, and this appeared sometime between July and December of 2015. It would be great to identify this so it doesn't display warnings. $

[FFmpeg-user] speed display during encode

Sometime between July and December 2015, it seems ffmpeg has added a "speed" display measured as a multiple of the framerate. Is it possible to either hide this or hide the fps display? It seems redundant to show both. I run ffmpeg in a script that shows progress with "-loglevel warning -stats"

[FFmpeg-user] x264 level option

I expected the x264 -level option to limit all parameters within the Level specification, but that is not happening. I'm encoding 1080p video using -preset veryslow and -level41. The veryslow preset uses ref=16, but Level 4.1 only allows 4. Is the level supposed to take precedence here? FFmpeg

Re: [FFmpeg-user] MKV to MP4

Hi Jack, You can download some free codecs which allow Windows Media Player to play 10-bit H.264. Just google that. Alternatively, change your command from “-vcodec copy” to “-pix_fmt yuv420p -vcodec libx264”, but this will result in some loss of quality and a much slower encode (because you

[FFmpeg-user] Slit-Scan Technique

Does anybody know how to achieve a slit-scan effect with ffmpeg?  It can be done in After Effects with the time displacement filter.  The way it works, is you take one source video, and one displacement map (typically a black-white linear gradient).  Each pixel of the source video is offset by

Re: [FFmpeg-user] Change frame rate without dropping/adding frames

After a long break from this issue, I have a new problem. For example, I want to take a 24fps video and make it play slow-mo at 12fps. The input option -r is dropping frames, rather than slowing down the playback speed. I don't have my main computer right now, which uses the latest git master;

Re: [FFmpeg-user] Change frame rate without dropping/adding frames

=0.0 q=-1.0 Lsize=2154kB time=00:00:16.08 bitrate=1097.0kbits/s video:2143kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.475101% On 1/15/15, Nicholas Robbins nickrobbins-at-yahoo@ffmpeg.org wrote: On Thursday, January 15, 2015 7:30 PM, Elliott Balsley

[FFmpeg-user] crop detect not working

I’m trying to use the cropdetect filter on some JPEG images, but it’s not showing anything useful in the console output.  I checked to see if it’s available with “ffplay -filters” and that shows the following line:  T.. cropdetect       V-V       Auto-detect crop size. Example command: $ ffplay

[FFmpeg-user] Audio input device on OS X

I’m trying to stream audio from my Macbook’s internal microphone using ffmpeg, but can’t figure out how to get the input device.  From what I’ve read online, it sounds like avfoundation doesn’t yet support audio.  I try listing devices for qtkit, but it only mentions video devices too. $

[FFmpeg-user] Encode DNxHD in MXF OP-Atom

As the title says, is this possible?  When I create an MXF with “-vcodec dnxhd”, it’s always OP1a, and therefore can’t be read by Avid Media Composer. ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-user

[FFmpeg-user] Change frame rate without dropping/adding frames

Hello, I want to take a video at 60fps and convert it to 24fps, without adding or dropping frames.  So the duration would be 2.5 times longer and the video would play in slow motion.  I realize the setpts filter can do this, but I’d like to avoid re-encoding.  This is possible in most NLE