Strange ffmpeg error when ProRes encoding

[SameSelf]

Hello, I am getting the strangest error when I try to encode an m2ts video to prores for editing in PP:

ffmpeg -i input.m2ts -vcodec prores -profile:v 2 -q:v 1 -an output.mov

The following message appears prior to encoding:

WARNING codec timebase is very high. If duration is too long,
the file may not be playable in quicktime. Specify a shorter timebase
or choose a different container.

Well, I don't really care if the file is unplayable in quicktime because I want to edit it in PP. But after ffmpeg finishes encoding, ffmpeg spits out the following:

FATAL error, the file duration is too long for timebase, this file will not be
playable with quicktime. Choose a different timebase or a different
container format

Sure enough, the encoded mov kicks off an error when I try to import into PP. However, if I use this same script on a short 20 second shot, the same warning appears but no FATAL error and the file is easily brought into PP.

So I am guessing it has something to do with resulting file size? The only thing is, I haven't a clue how to choose a different timebase or otherwise solve this problem. Any ideas on how to get this to work with the standard prores profile?

Thanks!

[smrpix]

First, most MPEG 2 transport streams play and edit fine in Premiere. Why do you need to transcode?
Second, ProRes is a lousy choice for PCs, even though it is now technically possible. The reverse engineering is imperfect and compatibility is less than ideal. Cineform or DNxHD is better.
Third, what is your source time base?

[poisondeathray]

This is a decoding error, has nothing to do with ffmpeg/libav prores implementation

Post the whole console log, both of the long test, and the 20 second test

You can try setting the timebase with settb
https://www.ffmpeg.org/ffmpeg-filters.html#settb_002c-asettb

Or maybe try ffmbc instead

[SameSelf]

^^Thank you for your response. I will try to respond to each point. First, while you are correct in that most m2ts play fine in PP, the reason I am trying to transcode is that of all the video I had, this one was choking in PP. Don't ask me why. So I thought I would try a different codec. Second, I am looking into prores because I am thinking of getting an Atomos Ninja for my cam. Third, if I understand the term correctly, my source's timebase is 1440x1080i30 or 29.97 fps.

^A decoding error. Hmm, very interesting. I would not have guessed that. I will be happy to post the whole console log once I am back at my computer this evening. I would love to use ffmbc but it doesn't support avisynth scripts

[poisondeathray]

I would love to use ffmbc but it doesn't support avisynth scripts

You can compile with avs support, or there are windows binaries with avs support

0.7rc8 has avs support here
http://www.mediafire.com/?8xynxe10rxb88

[smrpix]

The atomos ninja records DNxHD natively also.

[SameSelf]

You can compile with avs support, or there are windows binaries with avs support

0.7rc8 has avs support here
http://www.mediafire.com/?8xynxe10rxb88

poisondeathray to the rescue! Thanks so much. I was able to encode after downloading the windows binary with avs support using the following:

ffmbc -i input.avs -vcodec prores -profile std -b 145000k -pix_fmt yuv422p10le -an output.mov

The only thing I noticed is sometimes the q value during the encoding was around 7 or 8. Should I be concerned with that? When I set -qscale to 1, instead of limiting the bit rate to 145M, the file was too big imo, ~350M bitrate.

I was able to bring into PP and now PP can handle the video just fine whereas PP was choking on the m2ts version. Thanks!

[poisondeathray]

Or you can set qscale to different , higher value like 6 or 7. That's more like VBR encoding. Complex frames get allocated more proportional bitrate. Simple frames (e.g. blank wall, black frame etc...) get less

Specifying a -b for bitrate is more like CBR encoding. It might decode very slightly faster due to CBR, but complex frames will be lower in quality, simple frames overallocated

[SameSelf]

Or you can set qscale to different , higher value like 6 or 7. That's more like VBR encoding. Complex frames get allocated more proportional bitrate. Simple frames (e.g. blank wall, black frame etc...) get less

Specifying a -b for bitrate is more like CBR encoding. It might decode very slightly faster due to CBR, but complex frames will be lower in quality, simple frames overallocated

Thanks for the tutorial. I really don't understand what -qscale is all about and what sort of values to use. So your insight is greatly appreciated!