Help - How encoding with Variable Frame Rate using x264 command line.

[Kuronoe]

Hello.
I would like to know how I can encode using Variable Frame Rate (same as media-info below) using x264 and avisynth + avspmod. I have some knowledge of the use of these, I have been using them for a few years for encoding, but I found an anime in which the passage of panoramic scenes are jumping/skipping when I use 23.976 or 29.970 FPS. I found an encoding where they used the FPS media-info below and this seems to have eliminated the problem, unfortunately the old encoder has some problems with filter abuse (denoise). I would like to do an encoding directly from the DVD I have using the same FPS variation, but I have not found ways to do this, everything I tested and found did not work. My last test was using d2v and TIVTC to create timescodes, but the encoding did not go as it should, FPS was accelerated.
Is there a way to do this using x264?

Thank you for your time.

Video
ID                          : 1
Format                      : AVC
Format/Info                 : Advanced Video Codec
Format profile              : High@L4
Format settings             : CABAC / 6 Ref Frames
Format settings, CABAC      : Yes
Format settings, Reference  : 6 frames
Codec ID                    : avc1
Codec ID/Info               : Advanced Video Coding
Duration                    : 29 min 16 s
Bit rate                    : 1 665 kb/s
Maximum bit rate            : 7 430 kb/s
Width                       : 960 pixels
Height                      : 720 pixels
Display aspect ratio        : 4:3
Frame rate mode             : Variable
Frame rate                  : 25.162 FPS
Minimum frame rate          : 14.985 FPS
Maximum frame rate          : 119.880 FPS
Original frame rate         : 29.970 (30000/1001) FPS
Color space                 : YUV
Chroma subsampling          : 4:2:0
Bit depth                   : 8 bits
Scan type                   : Progressive
Bits/(Pixel*Frame)          : 0.096
Stream size                 : 349 MiB (89%)
Writing library             : x264 core 60 r920 8c8acae

[poisondeathray]

My last test was using d2v and TIVTC to create timescodes, but the encoding did not go as it should, FPS was accelerated.
Is there a way to do this using x264?

You have to tell x264 where the timecodes file is with --tcfile-in

 --tcfile-in "timecodes.txt"

Or instead of having to encode again, just mux in the timecodes with mkvmerge /mkvtoolnix or (mp4fpsmod for mp4)

But you will get slightly different results encoding wise, because x264 is "vfr aware" . It adjusts the crf quantizers based on the fps

[Kuronoe]

poisondeathray

I used this command when I did the first encoding, but it didn't help. The FPS came out with almost 300 fixed, it didn't have that variation that appears in the media-info.

--tcfile-in "g:\encoding\mytimecodesanime.txt"

I didn't try to mux the timescodes with mkvtools, I didn't know it was an option.

[ProWo]

Read this

https://slhck.info/video/2017/02/24/vbr-settings.html

[Kuronoe]

Read this

https://slhck.info/video/2017/02/24/vbr-settings.html

Thanks for your tip.

I had already read this and other articles, I had looked at about 80% of the posts about it here on VH and doom9 too. The problem was understanding the methodology for creating the VFR in avsp.

I got results by testing the TFM + Tdecimate models, as well as AnimeIVTC(mode = 4), but I was feeling not 100% correct, despite having solved my DVD problem after a few attempts.

The problem that came up afterwards (and I don't know how to solve it) was that some DVD's don't have a frame change even after the analysis process or when TFM and AnimeIVTC don't remove the interlacing, it is only removed with other filters, for example, qtgmc or bruteIVTC.

[poisondeathray]

Was it a problem with your timecodes or analysis? You might have to adjust the settings or thresholds

Did you get the same results with muxing? if so , that suggests problem with method or timecodes

Another method is duplicate decimation such as dedup, and that generates timecodes too; but if you use TFM preceding, that's only good for 23.976, 29.97 sections

(And some DVD's are terrible and have many issues, no matter what you do, sometimes you need to specify manual overrides for sections)

[Kuronoe]

I redid the procedure a few times using TFM + Tdecimate, as well as AnimeIVTC mode=4 (step 1) and mode=5 (step 2), this was the best result, it removed the interlaces and jumps in the image. I think that TFM was not very precise in interlacing.

The timecodes.txt I generated using virtualdub analysis and in another test using avsp analysis pass, were done correctly in theory. The dvd simply had no change in fps, it continued 23,976 FPS. It could be the bad DVD, as you mentioned.

What bothers me is that the encoding I see using VFR has other information about the frame (for example: original, minimum and maximum fps). Mine does not, only 1 information appears, which gives me the impression that it is not correct. I can add the information by passing the video through mkvtools and adding timecodes, but the question remains.

ID                          : 1
Format                      : AVC
Format/Info                 : Advanced Video Codec
Format profile              : High@L4.2
Format settings             : CABAC / 4 Ref Frames
Format settings, CABAC      : Yes
Format settings, Reference  : 4 frames
Codec ID                    : V_MPEG4/ISO/AVC
Duration                    : 26 min 49 s
Bit rate                    : 537 kb/s
Width                       : 760 pixels
Height                      : 576 pixels
Display aspect ratio        : 4:3
Frame rate mode             : Variable
Frame rate                  : 24.603 FPS
Color space                 : YUV
Chroma subsampling          : 4:2:0
Bit depth                   : 8 bits
Scan type                   : Progressive
Bits/(Pixel*Frame)          : 0.050
Stream size                 : 103 MiB (73%)
Title                       : Encoder by Kuronoe

(ignore the low quality, it is a fast encoder for testing)

Another point is that I can only encode using virtualdub, in x264 it causes an error that I don't understand.

whether or not using the timecodes.txt on the line.


[Attachment 52159 - Click to enlarge]


Thank you for your patience.


ps: sorry for any paranoids

[poisondeathray]

Examine the timecodes

Maybe there is error in procedure somewhere