24p broadcast in 60i, convert back to 24p with H.265?

[Nerva]

So, I've got something that started out as 24p film, that was converted years ago to 60i in MPEG2 format, that I would like to convert back to 24p, and I'd like to store it in H.265.

Is there conversion software that recognizes the 60i file is actually showing 24p media and would somehow know how to restore the frames back to the 24p as it looked in the first place, or would it screw up the frames even more?

I'm not doing this to save disk space -- my objective is to improve the playback on my plasma HDTV, which can sense 24p and switch refresh modes.

I'd like to use H.265, since I'm already switching formats, so I might as well re-encode it using the most modern technology.

[kolak]

Your plasma will most likely detect pulldown and remove it, which means your exercise is almost pointless. It just depends what model you have- decent one or some cheap.

[Nerva]

This is obviously some strange usage of the word "cheap" that I wasn't previously aware of.

I have a Panasonic TC-P65V10 -- it switches refresh rates between 60Hz (for 60i/60p) and 96Hz (for 24p) based on what mode my HTPC is outputting. MPC-HC switches display modes based on the format of the video -- it outputs either 60p or 24p. Yes, the first thing I did was to check what mode the TV uses with this video, and it is 60Hz -- I can see that when I pull up the TV's menu.

[benwaggoner]

This is obviously some strange usage of the word "cheap" that I wasn't previously aware of.

I have a Panasonic TC-P65V10 -- it switches refresh rates between 60Hz (for 60i/60p) and 96Hz (for 24p) based on what mode my HTPC is outputting. MPC-HC switches display modes based on the format of the video -- it outputs either 60p or 24p. Yes, the first thing I did was to check what mode the TV uses with this video, and it is 60Hz -- I can see that when I pull up the TV's menu.
Yeah, a high quality offline inverse telecine and then encoding to native 24p will always be at least as good, and typically significantly better than relying on real-time processing in a display. Particularly if you have a 96Hz mode. HEVC doesn't have MBAFF, so the efficiency gap for progressive v. interlaced will be bigger than for H.264.

That said, x265 doesn't have any preprocessing as all. You'd need to use AVISynth, ffmpeg or other processing tool to perform the inverse telecine prior to processing with x265.

I generally pipe the output of ffmpeg to x265 for this kind of thing.

[kolak]

This is obviously some strange usage of the word "cheap" that I wasn't previously aware of.

I have a Panasonic TC-P65V10 -- it switches refresh rates between 60Hz (for 60i/60p) and 96Hz (for 24p) based on what mode my HTPC is outputting. MPC-HC switches display modes based on the format of the video -- it outputs either 60p or 24p. Yes, the first thing I did was to check what mode the TV uses with this video, and it is 60Hz -- I can see that when I pull up the TV's menu.


Definition of "cheap":

https://en.wiktionary.org/wiki/cheap

eg. cheap TV, cheap plasma model....

TV has to "see" signal with pulldown to be able to analyse it and than do its magic. Good TVs should easily recognise and remove 3:2 pulldown. Your TV is good, so I'm not that convinced about real advantage of your task. Any re-encoding will also affect quality.
You can also force your output to be 24Hz, so TV definitely sees it this way and than does magic to remove pulldown. I would at least try before doing some re-encoding.

If you have time than avisynth is your friend- just search for IVTC.

[Nerva]

Well, I tried running it through DVDFab 9, and was surprised to come back the next morning and found it actually had finished without crashing. I haven't watched the entire movie, but skipping around to various parts, it looks good on my computer's CRT.

I'm guessing DVDFab isn't considered the highest-quality solution. I've never used Avisynth or x265 -- would they be considered the "best" method?

[foxyshadis]

If you don't want to go down a rabbit hole of learning the nitty-gritty behind conversion and command-lines, DVDFab is a solid simple choice, along with Handbrake, StaxRip, and more.

[Motenai Yoda]

Imho you can try with ffmpeg/hybrid or handbrake, they use pullup filter which seems to be better than TIVTC out of the box.

[Nerva]

What does x265 bring to the table compared to something like DVDFab? In the past I got the impression that x264 and x265 did a better job of compressing (better use of the standard) in their respective formats compared to the consumer products. And the fact everyone seems to use x264/x265 speaks volumes.

With DVDFab, I set it to use a custom 0.15 bits per pixel with H.265, which is 50% more than the program's highest-quality preconfigured setting for H.265. I chose that because in the past I ran a series of tests with DVDFab's H.264, and found that at anything below 0.25 BPP I could tell the difference from the original DVD when comparing identical frames, so I picked 0.30 BPP "for insurance" -- I forget what the default rates were for DVDFab with H.264, but they were probably something like 0.15 or 0.20 for the highest quality preconfigured setting. Using 0.30 BPP with H.264 tends to use about half the space of the original DVD MPEG2 file. Since H.265 is supposed to be twice as good as H.264, I went with 0.15 BPP for H.265. But, back to x265 -- how much BPP would it use to get the exact same quality as DVDFab with 0.15 BPP?

[foxyshadis]

DVDFab uses x265 (and x264) via ffmpeg internally.

I wouldn't take the "twice as efficient" as gospel, that's only for teleconferencing and streaming. Anything with grain and lots of detail will be much closer to H.264.

[benwaggoner]

With DVDFab, I set it to use a custom 0.15 bits per pixel with H.265, which is 50% more than the program's highest-quality preconfigured setting for H.265. I chose that because in the past I ran a series of tests with DVDFab's H.264, and found that at anything below 0.25 BPP I could tell the difference from the original DVD when comparing identical frames, so I picked 0.30 BPP "for insurance" -- I forget what the default rates were for DVDFab with H.264, but they were probably something like 0.15 or 0.20 for the highest quality preconfigured setting. Using 0.30 BPP with H.264 tends to use about half the space of the original DVD MPEG2 file. Since H.265 is supposed to be twice as good as H.264, I went with 0.15 BPP for H.265. But, back to x265 -- how much BPP would it use to get the exact same quality as DVDFab with 0.15 BPP?
BPP has a pretty weak correlation to visual quality. What gets to a given perceptual quality varies substantially by all of frame size, frame rate, content type, and other encoder settings.

If you are trying to hit a particular quality target, CRF (while far from perfect) is certainly going to get you a lot closer to the ballpark.

Also "HEVC half the bitrate as H.264" is also a very coarse estimate. It varies enormously depending on frame size (HEVC has a bigger advantage as frames get bigger, and is >50% more efficient at UHD resolutions). And it varies a lot between different H.264 and HEVC encoders, versions, and implementations.

At the start of the year, many felt that x264 could provide transparent quality at lower bitrates than x265, although x265 would look better than x264 at much lower bitrates. Improvements in x265 over the last few months have yielded big improvements in detail retention.

[Nerva]

Oh, I should have mentioned I'm using 2-pass encoding with DVDFab, if that matters. I don't think there's a CRF setting in DVDFab.

[Nerva]

OK, a couple more things:

1) The film I want to convert has Japanese subtitles hard-coded in the black area below the film itself, so I figure I might as well crop the black bars and only have the move. But it turns out DVDFab can't crop, so I need to use something else if I want to do that.

2) I've read about how using 10bit color improves video playback quality even if the original is only 8bit, and you get smaller file sizes for the same quality. I think x265 supports 10bit for encoding, but do I need to do anything special on the IVTC side of things in order to the get maximum benefit?

[Asmodian]

2) I've read about how using 10bit color improves video playback quality even if the original is only 8bit, and you get smaller file sizes for the same quality. I think x265 supports 10bit for encoding, but do I need to do anything special on the IVTC side of things in order to the get maximum benefit?

No, nothing special needed before x265.

[Motenai Yoda]

2) I've read about how using 10bit color improves video playback quality even if the original is only 8bit, and you get smaller file sizes for the same quality. I think x265 supports 10bit for encoding, but do I need to do anything special on the IVTC side of things in order to the get maximum benefit?

It's all about what kind of artifacts you're expecting, 10b bitdepth will avoid almost all colorbanding caused by x264/x265, but not much else.
quality/bitrate improvement will be about 5%

[Nerva]

BPP has a pretty weak correlation to visual quality. What gets to a given perceptual quality varies substantially by all of frame size, frame rate, content type, and other encoder settings.

If you are trying to hit a particular quality target, CRF (while far from perfect) is certainly going to get you a lot closer to the ballpark.

...


I had been under the impression that the "best" quality/size comes from using 2-pass encoding -- or is that just better at giving the best quality for a specified file size? Is CRF better at providing certain level of quality, while letting the file size end up at whatever the source material dictates?

[benwaggoner]

I had been under the impression that the "best" quality/size comes from using 2-pass encoding -- or is that just better at giving the best quality for a specified file size? Is CRF better at providing certain level of quality, while letting the file size end up at whatever the source material dictates?
Two-pass basically gives you a file that is at the CRF that would have given you the requested VBR bitrate. You can view the first pass as picking the optimum CRF value if you squint.

That said, if you are hitting the VBV limit significantly, a 2nd pass, CRF or otherwise, in x265 will help smooth out the bitrate and thus QP spikes around where the VBV limit is hit. If you have a high peak bitrate that might never matter, but if your ABR is close to --vbv-maxrate, you can get some quality improvement where your video would look worst by doing a second pass.

Also, don't use DVDFab, since it seems pretty ill-suited towards your task :). StaxRip seems to be the most popular GUI with good x265 support. Personally, I generally just do my processing in ffmpeg and pipe the output to x265. I can't say I recommend that for most people who don't have to deal with my terribly particular use cases.

[Nerva]

In the setup for StaxRip, do I choose Auto AVISynth?