HOWTO: Install and use the latest FFmpeg and x264

[popper]

PDA1.

why are you asking the same questions yesterday about getting a working ffmpeg line in
http://ffmpeg.org/pipermail/ffmpeg-u...er/003431.html

i thought you said the iPod 640x480 line i worked out for your device was working perfectly fine.... days ago?

and why are you still using the old -vpre syntax there when i gave you the current
newer internal -preset -vprofile etc syntax to use

[rulet]

Hello, I have this error with missing preset:

r@ngf:~/Рабочий стол/1$ ffmpeg -i 1.mp4 -acodec libfaac -ab 128k -ac 2 -vcodec libx264 -vpre slow -crf 22 -threads 0 2.mp4
ffmpeg version git-2011-12-04-f88d5df, Copyright (c) 2000-2011 the FFmpeg developers
built on Dec 4 2011 11:21:47 with gcc 4.6.1
configuration: --enable-gpl --enable-libfaac --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libtheora --enable-libvorbis --enable-libx264 --enable-nonfree --enable-postproc --enable-version3 --enable-x11grab --enable-libvpx
libavutil 51. 30. 0 / 51. 30. 0
libavcodec 53. 40. 0 / 53. 40. 0
libavformat 53. 24. 0 / 53. 24. 0
libavdevice 53. 4. 0 / 53. 4. 0
libavfilter 2. 50. 0 / 2. 50. 0
libswscale 2. 1. 0 / 2. 1. 0
libpostproc 51. 2. 0 / 51. 2. 0
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '1.mp4':
Metadata:
major_brand : isom
minor_version : 512
compatible_brands: isomiso2avc1mp41
creation_time : 1970-01-01 00:00:00
encoder : Lavf53.24.0
Duration: 06:35:03.80, start: 0.000000, bitrate: 198 kb/s
Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 800x596, 66 kb/s, 15 fps, 15 tbr, 15 tbn, 30 tbc
Metadata:
creation_time : 1970-01-01 00:00:00
handler_name : VideoHandler
Stream #0:1(und): Audio: aac (mp4a / 0x6134706D), 48000 Hz, stereo, s16, 127 kb/s
Metadata:
creation_time : 1970-01-01 00:00:00
handler_name :
File for preset 'slow' not found

[popper]

Hello, I have this error with missing preset:
built on Dec 4 2011 11:21:47 with gcc 4.6.1

File for preset 'slow' not found

" -vpre slow" is the old syntax.... its replaced with the internal " -preset slow" now , you might also want to add a "-vprofile high" to the line if you care about better visual quality too, or even add some filters.

-vf gradfun,hqdn3d,unsharp=5:5:0.5,setdar=16:9
works nicely on just about all DAR (display aspect ratio) 16:9 ration videos, replace setdar=16:9 with =4:3 were required or remove the ,setdar= filter if you have other ratio input files to re-code, hqdn3d and especially unsharp=5:5:0.5 having the most notable visually improved quality effect IMO, try them and compare.

[rulet]

Thanks for the answer, -preset slow works. I just need recoding for compession for now, in other words, the most smallest file with the best quality. Maybe you have any ideas how to compress it better then in that command which I've wrote.

[popper]

Thanks for the answer, -preset slow works. I just need recoding for compession for now, in other words, the most smallest file with the best quality. Maybe you have any ideas how to compress it better then in that command which I've wrote.

some people don't seem to like my answer to that one as they prefer unrestricted bit-rate within the confines of the CRF setting and wasting drive space needlessly , BUT if you want to give it a try i do this all the time and its fine for everyday use (not for archiving or a first backup OC) remove the minrate., maxrate etc for these one off keeps

ffmpeg.exe -threads 4 -i "inputfile.whatever" -f mp4 -vcodec libx264 -crf 18 -minrate 800k -maxrate 900k -bufsize 900k -refs 3 -preset fast -vprofile high -flags2 -bpyramid -wpredp 0 -strict experimental -acodec aac -ac 2 -ab 128k -vf gradfun,hqdn3d,unsharp=5:5:0.5 "outfile.MP4"

ffmpeg.exe -threads 4 -i "inputfile.whatever" -f mp4 -vcodec libx264 -crf 18 -minrate 800k -maxrate 1500k -bufsize 1500k -refs 3 -preset fast -vprofile high -flags2 -bpyramid -wpredp 0 -strict experimental -acodec aac -ac 2 -ab 128k -vf gradfun,hqdn3d,unsharp=5:5:0.5 "outfile.MP4"

ffmpeg.exe -threads 4 -i "inputfile.whatever" -f mp4 -vcodec libx264 -crf 16 -minrate 800k -maxrate 2000k -bufsize 2000k -refs 3 -preset fast -vprofile high -flags2 -bpyramid -wpredp 0 -strict experimental -acodec aac -ac 2 -ab 128k -vf gradfun,hqdn3d,unsharp=5:5:0.5 "outfile.MP4"

as you can see only the maxrate and bufsize change and match each other for simplicity of a seconds worth of buffering.

and you get a larger file with each change, i find 900k , 1100k and 1300k to be fine for most web streaming and every day use ,and thats also why the combined
" -flags2 -bpyramid -wpredp 0" are there for the older VLC browser plugin to be able to play the -vprofile high produced output files....

and notice how i use CRF 18 or even lower... as the min/max resticts that anyway so trying to keep as good a visual quality as it can.

also notice how i don't even try and re-scale the input on purpose, as i don't like the jaggys that always produces, just lowering the max bit-rate is perfectly fine in my book to whatever limit you prefer for a given input file.

for things like David Attenborough Frozen Planet 1080P TS you might want to use the -maxrate1600k to 2500k range OC if you care to reduce that down for good every day visual playback on less powerful old single core laptops etc, works for me

[rulet]

Well, I tried your first command

ffmpeg -threads 4 -i "inputfile.whatever" -f mp4 -vcodec libx264 -crf 18 -minrate 800k -maxrate 900k -bufsize 900k -refs 3 -preset fast -vprofile high -flags2 -bpyramid -wpredp 0 -strict experimental -acodec aac -ac 2 -ab 128k -vf gradfun,hqdn3d,unsharp=5:5:0.5 "outfile.MP4"

but that's even makes larger file then with that command that I used before. Seems like the video-file is already compessed as it possible with normal quality.

[qyot27]

Faster presets = lower quality at same filesize or larger filesize for same quality, considering the use of CRF. Use a slower preset if you want to maintain the quality with a smaller filesize.

There's also the option of using 10-bit H.264 (requires a 10-bit build of x264), and/or using Wine in order to have access to AviSynth and thereby get better filtering possibilities.

[popper]

Faster presets = lower quality at same filesize or larger filesize for same quality, considering the use of CRF. Use a slower preset if you want to maintain the quality with a smaller filesize.

There's also the option of using 10-bit H.264 (requires a 10-bit build of x264), and/or using Wine in order to have access to AviSynth and thereby get better filtering possibilities.

ohh yeah forgot that and was looking at an older script of mine, so -preset slow or -preset slower will be better but take more time to finish so try those too OC


actually talking about AviSynth , do you/anyone have or can make a working self contained rar of the needed files and how-to text using Wine ,as using that with QTGMC de-interlace script will produce far greater quality But far slower even on a i5 sandy bridge , than the internal ffmpeg jaggy "yadif" de-interlace filter, assuming OC that the QTGMC etc even work's on Linux/Wine.

and some better de-ringing advice for working avisynth scripts rather than using the internal ffmpeg/avconv gradfun filter as its not very good but all we have really

[popper]

Well, I tried your first command


but that's even makes larger file then with that command that I used before. Seems like the video-file is already compessed as it possible with normal quality.

that doesn't really surprise me given the smaller oddball re-scaled
Video: h264 (High) (avc1 / 0x31637661), yuv420p, 800x596, 66 kb/s, 15 fps, 15 tbr, 15 tbn, 30 tbc

i don't really like re-scaled anything as it usually looks crap when played back on generic full screen monitors etc , unless it's to make it actually play on a given device that cant take generic dvd/BR sizes of 16:9/4:3 ratio files such as the mac device a few pages back. but hows the -preset slower looking, still larger or the same size now as your original line, what about the quality at full screen size, the same, better,worse ?

[qyot27]

actually talking about AviSynth , do you/anyone have or can make a working self contained rar of the needed files and how-to text using Wine ,as using that with QTGMC de-interlace script will produce far greater quality But far slower even on a i5 sandy bridge , than the internal ffmpeg jaggy "yadif" de-interlace filter, assuming OC that the QTGMC etc even work's.

I rarely ever purely deinterlace things and my computer is far too slow (P3 Coppermine-based Celeron) to even attempt to use QTGMC - I still rely on TDeint or TomsMoComp. Heck, it takes 45-50 minutes for me to just build FFmpeg on here (the entire basic FFmpeg->FFMS2->x264 8/9/10bit->full FFmpeg process takes something on the order of 3 hours).

What I would recommend is - if one has access to Windows - to get all their needed AviSynth plugins worked out there, and then just copy them over to the same folder under Wine. If something doesn't work, then refer to the AviSynth MediaWiki to find alternatives that do. Of course, if one's first experiences with AviSynth are under Linux it would be virtually the same - just that you have to keep in mind Wine's organization hierarchy. It helps to add a bookmark to drive_c in Nautilus' or PCManFM's sidebar (Konqueror and Dolphin probably have a similar feature but I've not used KDE in a couple years) so you can just jump right there if needed.

The single biggest thing (to my knowledge) that AviSynth-under-Wine lacks is the use of DirectShowSource, since Wine can't use DirectShow. AVISource will work so long as ffdshow's VFW interface is installed and configured*, and MPEG2Source (an SSE-optimized version of DGDecode.dll is available on the main HCenc release page), FFmpegSource 1.x, and FFMS2 will all work since they're self-contained.

*I would suggest doing this on Windows and telling ffdshow to export the settings to a .reg file. The .reg can then be imported into Wine's regedit. This is because ffdshow's decoders can't be configured under Wine, or at least they couldn't last time I checked.

and some better de-ringing advice for working avisynth scripts rather than using the internal ffmpeg/avconv gradfun filter as its not very good but all we have really

For debanding, I used to use gradfun2db, but have since moved to flash3kyuu_deband (especially since it can output in higher bit depths, and if x264 has been patched to accept high bit depths coming from AviSynth, then it's a pretty straightforward process to maintain a high bit depth workflow - note that that only works with Windows builds of x264 since Linux builds can't take AVS input at all).