Encoding DVD Movie - Does the DAR Really matter? and 710x480 vs 648x480 ?

[newpball]

...various attachments....

Huh?

The first image is 412×288 you resize that to 384×288
The third image is 412×309

We were talking about 720x480 right?

(really 864x648 to keep it to scale with the original):

You lost me completely here.

720x540 goes straight to 1440x1080 by scaling exactly 2x

640x480 to 1440x1080 is messy!

So seriously if you would have to go from 720x480 to 1080 you prefer to decimate the horizontal resolution first to 640 and then do a messy upscale to 1440x1080 over a straight forward resize of 2x horizontal and 2.25 vertical?

Stunning!

Needless to say that edge "enhancement" (a darling some engineers even in the HD era can't still not get enough of) with standard bicubic upsampling gets increasingly worse as well:

Original - From 720 - From 640:

[hello_hello]

We all know the theory.....

I sure you all do, but I recon that "we've always done it like that", "it's more standard"and "trust us we know better" take preference to some.

I guess you can only speak for yourself......

[hello_hello]

The first image is 412×288 you resize that to 384×288
The third image is 412×309

We were talking about 720x480 right?

(really 864x648 to keep it to scale with the original):

You lost me completely here.

FFS!

720x480 is 1.5:1.

Pic #1 is 412x288 which is 1.43:1. Not quite 1.5: 1 but close enough for the point I was making. I just used it because it was there.

Pic #2 is 384×288. That's 4:3 by reducing the width. Just like resizing the 720x480 DVD to 640x480.

Pic #3 is 412x309, once again just like resizing a 4:3 DVD but this time by increasing the height instead of reducing the width.

We then discussed resizing the 4:3 DVD to 1080p. ie 640x480 to 1440x1080 or 720x540 to 1440x1080.

I resized pics 2 and 3 by a similar percentage but they're not 640x480 or 720x540, so the closest aspect ratio while maintaining the same relative upscaling to 1080p would be around 864x648.

It's not a perfect example because pic #1 isn't exactly 1.5:1 so maybe I should have resized to a slightly narrower aspect ratio than 4:3 to compensate, but can you see any difference in the upscaled images?

720x540 goes straight to 1440x1080 by scaling exactly 2x

640x480 to 1440x1080 is messy!

So seriously if you would have to go from 720x480 to 1080 you prefer to decimate the horizontal resolution first to 640 and then do a messy upscale to 1440x1080 over a straight forward resize of 2x horizontal and 2.25 vertical?

Stunning!

I don't know what you're on about.

(640x480) x 2.25 = (1440x1080)

(720x540) x 2 = (1440x1080)

The only thing that seems messier to me is stretching the height first and then upscaling it just because you like the look of the numbers better.
The only "straight forward resize of 2x horizontal and 2.25 vertical" taking place is in your head, as stunning as it may seem to you in there. Either way, the video is being resized twice. Once when it's encoded, and again when it's upscaled. Anamorphic encoding is not an option. The player ignores the aspect ratio.

Once again, show me an example of a real world DVD where it makes a visible difference or once again you're ignoring the examples that have been offered because they don't suit you. Debating a subject with an ostrich is fairly pointless. I've done you the courtesy of looking at what you've posted and commenting, how about for once you do the same?

Needless to say that edge "enhancement" (a darling some engineers even in the HD era can't still not get enough of) with standard bicubic upsampling gets increasingly worse as well

I'll leave you to argue about that one with yourself given it's not really anything to do with the discussion.

[manofsteel31]

So I just tried encoding a few Deleted Scenes off of a live action bluray disc (the Santa Clause 2). Unlike the movie itself, the scenes were not enhanced and were presented in SD 720x480 4:3 just like they would have been on a DVD.

The original Movie was 1:85:1 and it looked like the deleted scenes were too, but they had giant black bars at the top and bottom (I believe the correct term is Letterboxed) so they could be presented in 4:3 instead. So after I cropped them it completely changed the aspect ratio to 1:85:1 because they were resized to 710x382. What I find strange is that the the aspect ratio error was still something like 0.00198% (or something like that). Why would it say its that close to the original aspect ratio when it completely different (1.36 to 1.85)?

I am also curious why the FPS sometimes changes after encoding a video and other times it stays the same. Does it have something to do with Interlacing?

For example, most of the scenes changed from 29.970fps to 23.976fps after I encoded them. I am fine with that since i would prefer if all my videos were 23.976, but I still dont understand it. Especially since there was a "gag reel" i also encoded off the disc and it stayed at 29.970fps even after encoding (using the One Click Encoder and Automatic Deinterlacing checked).

I thought maybe it had to do with interlacing, so I tried to encode the gag reel again, and then time I used the "Analyze" button in the Script Creator which told me the source was "M-in-5 Decimation Required", "M=2" and "Tritical Decimate". Strangely when encoding this way, the fps resulted in 17.986fps (which is a speed I have never seen before).

I think the Gag Reel should also have been 1.85:1 after cropping and resizing, but it ended up as 718x396 and 16:9. Blu-ray.com lists it as being 1:85:1 and since it obviously came off the same reel as the actual movie and deleted scenes then I dont know why it wouldnt be.

[hello_hello]

I haven't got time for a long reply at the moment, but even though they both usually full under the de-interlacing category for encoder GUIs, de-interlacing and inverse telecine aren't the same thing. 29.970 to 23.976 would be IVTC. You can read all about that one here.

http://en.wikipedia.org/wiki/Three-two_pull_down
http://www.zerocut.com/tech/pulldown.html

When you apply IVTC you're recovering the original progressive frames at 23.976fps. When you de-interlace, you're errr..... de-interlacing.

http://en.wikipedia.org/wiki/Interlaced_video

It's worth reading the info and understanding the differences.

[hello_hello]

The original Movie was 1:85:1 and it looked like the deleted scenes were too, but they had giant black bars at the top and bottom (I believe the correct term is Letterboxed) so they could be presented in 4:3 instead. So after I cropped them it completely changed the aspect ratio to 1:85:1 because they were resized to 710x382. What I find strange is that the the aspect ratio error was still something like 0.00198% (or something like that). Why would it say its that close to the original aspect ratio when it completely different (1.36 to 1.85)?

You can put a wide aspect ratio picture on a 4:3 DVD frame just as you can put one on a 16:9 DVD frame, but because they're resized differently (or have different aspect ratios), a 1.85:1 picture on a 16:9 frame wouldn't require much in the black bars department (1.85:1 is relatively close to 1.77:1) but to put a 1.85:1 picture on a 4:3 frame you need large black bars top and bottom, so what you're saying makes sense. Once you crop away the black bars though, you're still left with a 1.85:1 picture.
It's better to put a 1.85:1 picture on a 16:9 frame because the smaller black bars mean more of the 720x480 resolution contains actual picture. Plus the player only knows the video is 4:3 and resizes it to 4:3, so to display it on a 16:9 TV it must add black bars down the sides. There's already large black bars top and bottom, so when that poor 1.85:1 video is finally displayed on a 16:9 TV, it ends up looking something like this:



After you've cropped away the black top and bottom and encoded just the green with a 1.85:1 aspect ratio, the player can zoom the green until the width fills the screen (no black bars down the sides) but the resolution won't be very exciting.

I thought maybe it had to do with interlacing, so I tried to encode the gag reel again, and then time I used the "Analyze" button in the Script Creator which told me the source was "M-in-5 Decimation Required", "M=2" and "Tritical Decimate". Strangely when encoding this way, the fps resulted in 17.986fps (which is a speed I have never seen before).

Sounds like MeGUI got it wrong. If it's animation, which tends to have lots of repeated frames anyway, it's hard to analyse.
If you've read the links I posted earlier you'll hopefully understand the difference between pulldown and interlacing.
If an NTSC DVD is interlaced, you'd normally de-interlace and the output would be 29.970fps progressive or 54.940fps progressive. It depends how you de-interlace. The second method is labelled "Yadif with Bob" in MeGUI, as opposed to plain "Yadif" and should provide smoother motion.
If an NTSC video is "film" it'll be 23.976fps "film" converted to 29.970fps using pulldown. You wouldn't de-interlace. You'd apply IVTC to reverse the pulldown process and output 23.976fps progressive. "Film" can also be 23.976fps progressive on a DVD and the DVD player is supposed to add pulldown on playback and output 29.970fps, but if a DVD is 23.976fps progressive "film" you can just encode it that way. Likewise I think NTSC "video" can also be 29.970fps progressive, so you'd just encode it that way.

The problem with NTSC is it can be a combination of the above, which is why MeGUI has "Hybrid" de-interlacing options, but the automatic analysis doesn't always get it right. Use it as a guide, but if in doubt, post a sample.

I live in PAL land and as a rule it tends to be easier. It's either interlaced or it's progressive. "Film" is 24fps progressive sped up to 25fps progressive and "PAL" is 25fps interlaced. You'd either de-interlace the latter to 25fps progressive (ie Yadif) or 50fps progressive (ie Yadif with Bob).

I think the Gag Reel should also have been 1.85:1 after cropping and resizing, but it ended up as 718x396 and 16:9. Blu-ray.com lists it as being 1:85:1 and since it obviously came off the same reel as the actual movie and deleted scenes then I dont know why it wouldn't be.

If MeGUI cropped and resized to 16:9 with minimum aspect error it'll probably be correct. There's no guarantee the gag reel video wasn't cropped to a slightly different aspect ratio when it was transferred to Bluray. Although 718x396 is 1.81, so where did the 16:9 aspect ratio come from?

[manofsteel31]

I think the problem with the video using DirectShowSource was just a fluke, however here is the original and shakey video you asked for:

ORIGINAL:



SHAKEY:




I have spent the last few days encoding and testing out some videos using the information you have me and its mostly going well. I have been trying to open my videos in "File Indexer" but some work and some dont. I have tried .MKV, .VOB, .mpls and .m2ts. Sometimes it gives me error when trying to create the .ffindex or .d2v file (and i am not sure why), so in those cases I open the videos in "DirectShowSource" instead.

I have been opening the videos, letting it analyze the video, letting it AutoCrop and then doing a little extra if needed and then resizing the video as close to the original as I can (while upscaling the height) while maintaining an Aspect Ratio Error at 0.10000% or Under.

I am ending up with frame sizes like 708x526, 706x524 and 702x524.

Again, I know that you prefer resizing yours to 640x480.

However, I am curious about something. While I have switched my TV to "mode1" where only the Height fills the screen for 4:3 videos and the Width fills the screen for 16:9 videos, if I were to watch one of 4:3 encoded videos in "mode2" just because I wanted it to completely fill the screen again, would the "708x526" video look better than the "640x480" since the width would not have to stretch as much? I realize you don't recommend watching videos in mode2 anyway, but if I were to then would those larger dimensions actually end up being more beneficial? Just wondering.


In the last few days I sat down and took a long hard look at my encoding style in terms of CRF values and Bitrate. While I still am not willing to start encoding everything at a set CRF value (simply because I want my file sizes to be manageable and my tests with a constant CRF value have shown otherwise), I have made some changes. To begin with I questioned myself as to why I am aiming for the same bitrate range for videos that are so different in Frame Size? Videos that are 708x526, 1920x800 and 1920x1080 should NOT be encoded to the same Bitrate. The bigger the picture, the more bitrate it should need. So I wanted to find a way to maintain the current bitrate I have been using for the last year (2,200-2,399Kbps) as the MINIMUM for my encodes and then just increase the value depending on how big the Frame Size is. Here is what I came up with:

All 4:3 SD videos will be encoded to a set bitrate of 2,300Kbps

16:9 HD Videos (1920x????) will have their bitrate determined by a calculation. I decieded to alot "2.875Kbps" per line of horizontal resolution (meaning determined by height of video). For example a 2.40:1 video resized to 1920x800 would get 2,300Kbps (2.875 x 800 = 2,300). The highest bitrate would be for a 1.77 1920x1080 Video which would get 3,105Kbps (2.875 x 1080 = 3,105).

I have not yet determined what to do with HD videos that are 1.37:1 (1480x1080) or 1.67:1 (1800x1080) yet, but I guess I will decide when I come across the next one.

I feel that I am adding just the right amount of bitrate to my HD videos (for a picture quality I am happy with). I am most likely giving too much bitrate to some of my 4:3 SD videos, but I wanted to maintain a minimum bitrate within the range that I had been using for so long, so 2,300Kbps sounded like a good way to do that.

Given the fact that I am not calculating exact birtates, I switched to "2-Pass" encoding which (since I only have to encode each video 1 time now, lol) is much faster in the long run.

I realize none of this is anything you would recommend since every video is different and should really be treated as individuals that could need dramatically different amounts of bitrate to maintain a certain level of quality (which is why you use a constant CRF value), but I dont like my file sizes being all over the place, and I dont have the space at the moment to manage file sizes that that 4+ GB each, or even 8 minute videos that are over 1GB. Maybe some day I will, but right now I don't. So this is the best I have come up with at the moment. I still think it is an improvement.


Moving on, I did end up cropping those 1.85:1 Deleted Scenes out of their 4:3 letterboxing. Yes, the videos ended up looking like CRAP, but then again the deleted scenes look like CRAP on the Bluray itself already, so I didnt really notice a big change at all. At least now they are encoded to their original Aspect Ratio.

So, I encoded a video the other day that was originally 29.970fps. MeGUI first determined it was "M-in-5 Decimate" and encoded it to 17.xxxfps. Then I tried it again and it saw it as "Hybrid Film mostly interlaced" with resulted in the fps staying at 29.970fps. Then I went and encoded it a third time, chosing my own options. Based on MediaInfo which specified the video as "Interlaced - Top Field First" and Selected "Hybrid mostly film - Top Field First - TIVTC" and the result was a video at 23.976fps. While I should probably not be picking my own options based on a change in fps that i desire, I didnt notice a difference in the 3 videos, which I think I should have. Since i prefer an output of 23.976fps Progressive video, I have a feeling I will select the same settings in other interlaced videos (that are originally 29.970fps) if the resulting video does not look different than the one encoded with the recommended settings. I am sure that is All Kinds of Wrong, lol, so I figured I would mention it.

Oh yeah, Yandif with Bob resulted in 50fps which is much too high for me. I realize that people used to PAL video as apposed to NTSC might also be used to higher speeds, but when I watch video at a high fps it looks like it is being "fast forwarded" to me and it not relaxing and enjoyable to watch. Of course that might just be my opinion.

Sorry about saying the Gag Reel was 16:9 when it was really 1.81. I was really just quoting what MediaInfo listed it as. I think because 1.81 is so close to 1.77 that MediaInfo just claimed it was 16:9 even though it isnt exactly. Sorry for the confusion.

[hello_hello]

I think the problem with the video using DirectShowSource was just a fluke, however here is the original and shakey video you asked for:

The original isn't perfectly smooth but maybe the right pulldown removal wasn't used or maybe MeGUI got the frame rate wrong and DirectShow is trying to convert it. The only way to know for sure is to look at an untouched sample of the original.

I have spent the last few days encoding and testing out some videos using the information you have me and its mostly going well. I have been trying to open my videos in "File Indexer" but some work and some dont. I have tried .MKV, .VOB, .mpls and .m2ts. Sometimes it gives me error when trying to create the .ffindex or .d2v file (and i am not sure why), so in those cases I open the videos in "DirectShowSource" instead.

If you have ffms2 problems try selecting L-Smash as the indexer (if MeGUI selects ffms2 automatically). It'd be preferable to DirectShow and it's probably more reliable than ffms2 these days as it's not updated all that frequently. There's been a little discussion in the MeGUI thread at doom9 regarding changing the default indexer to L-Smash in preference to ffms2. DGIndex etc would still be used for the same types of video they're indexing now.

I have been opening the videos, letting it analyze the video, letting it AutoCrop and then doing a little extra if needed and then resizing the video as close to the original as I can (while upscaling the height) while maintaining an Aspect Ratio Error at 0.10000% or Under.

I am ending up with frame sizes like 708x526, 706x524 and 702x524.

Again, I know that you prefer resizing yours to 640x480.

I'm not married to 640x480 as such, but I like to keep the aspect ratio at 4:3 if it's really close and I like to keep related videos the same resolution. Each to their own......

However, I am curious about something. While I have switched my TV to "mode1" where only the Height fills the screen for 4:3 videos and the Width fills the screen for 16:9 videos, if I were to watch one of 4:3 encoded videos in "mode2" just because I wanted it to completely fill the screen again, would the "708x526" video look better than the "640x480" since the width would not have to stretch as much? I realize you don't recommend watching videos in mode2 anyway, but if I were to then would those larger dimensions actually end up being more beneficial? Just wondering.

The higher the resolution the more detail should be retained in theory, but I doubt you'd notice the difference as 4:3 DVD video isn't exactly packed full of fine picture detail to begin with, and then there's the bitrate/resolution/quality balancing act. In theory the less video there is to encode the higher the encoding quality should be for a given bitrate, and the more you resize the video the more you'll see any compression artefacts. There's probably no "best way" answer as a result. It'd depend.....

If you're really keen you can crop a lot of 4:3 video to 16:9 and it'll still be quite watchable, but you'd probably want to still keep a good quality 4:3 copy as well. Because I use MPC-HC as my media player it's easy for me to zoom in till 4:3 video fills the screen, then I move the picture down a little as the 4:3 action tends to be mostly centred a little above the middle of the frame, then I watch it and mostly forget it's 4:3. It's a pity the TV doesn't have a non-stretch zoom option. Mine doesn't seem to.
I've been working on re-encoding Twin Peaks recently. It's HD which makes it easier to get away with cropping a lot of picture, but I started re-encoding because the over-saturated red look really annoys me (artistic or not, I don't like it) and while I was playing with that I started to crop a bit and got carried away. So I started out with the top pic and my encoded version is the bottom one.





Some episodes were nice and easy. I set the cropping and it barely changed, while for a few I was moving the cropping up and down quite a bit (effectively panning and scanning) and that can take a bit of time when you're doing it with a text editor and Avisynth.
Normally I'd use MPC-HC to zoom in the same way on playback and just encode at 4:3, but as I said, I got carried away.....

......simply because I want my file sizes to be manageable and my tests with a constant CRF value have shown otherwise......

I don't know what your definition of manageable is but don't forget the file size is also dependant on the duration, not just the bitrate or CRF value.

All 4:3 SD videos will be encoded to a set bitrate of 2,300Kbps

I checked a few SD encodes sitting on my hard drive. All different resolutions, but as a few examples:
784x576, CRF18, 1751kbps
960x540, CRF18, 1952kbps
768x564, CRF18, 2862kbps
854x358, CRF17, 1337kbps
640x480, CRF18, 1205kbps
656x480, CRF18, 1599kbps
Season one of "The Vikings", PAL DVD source, 960x540, CRF18, Lowest 2051kbps, highest 2623kbps.

16:9 HD Videos (1920x????) will have their bitrate determined by a calculation. I decieded to alot "2.875Kbps" per line of horizontal resolution (meaning determined by height of video). For example a 2.40:1 video resized to 1920x800 would get 2,300Kbps (2.875 x 800 = 2,300). The highest bitrate would be for a 1.77 1920x1080 Video which would get 3,105Kbps (2.875 x 1080 = 3,105).

I encoded a documentary series the other day. 6 episodes, 1280x720, CRF18. Lowest bitrate 3711kbps, highest 5048kbps.
720p encoding of a police drama series. Not a huge amount of action despite the police drama description. Lowest 2463kbps, highest 3201kbps.
Every Sopranos episode at 720p (I spent a fair bit of time cleaning those up and removing noise so that'd help keep the file sizes down). CRF18. Lowest 2088kbps, highest, 3755kbps

I'd need to go digging for 1080p encodes as mostly I use a lower resolution, but I can tell you iTunes use around 4000kbps for 1280x720 and 5200kbps for 1920x1080. From what I've seen it doesn't vary much and the quality is quite high. They use more efficient encoder settings for 1080p which helps explain why the 720p and 1080p bitrates don't differ by as much as you might expect.

Anyway, that might give you an idea of the quality ballpark you'll be in. I'd possibly consider stealing some bitrate from the SD stuff and giving it to the HD stuff, or errrr...... letting CRF do it for you.

I realize none of this is anything you would recommend since every video is different and should really be treated as individuals that could need dramatically different amounts of bitrate to maintain a certain level of quality (which is why you use a constant CRF value), but I dont like my file sizes being all over the place, and I dont have the space at the moment to manage file sizes that that 4+ GB each, or even 8 minute videos that are over 1GB. Maybe some day I will, but right now I don't. So this is the best I have come up with at the moment. I still think it is an improvement.

The problem is the odd video that requires a bitrate way outside of average. Generally it'd be something really noisy with lots of action that's hard to compress, but I'd be willing to bet every so often you'll find an encode with more compression artefacts than usual.
I don't always stick to an exact CRF value, although I don't vary it a lot, but while around CRF18 to CRF 20 will give you quite good quality most of the time there's always exceptions. Lately I seemed to have ramped up my war against noise, and I'm either trying to remove it completely (or close to) or if not I try to encode it "accurately" without the encoder turning really fine, inoffensive noise into "dancing blocks" that annoy me, and sometimes I go down as low as CRF16 to fix that, but mostly I'm encoding in the CRF18 to CRF20 range.

So, I encoded a video the other day that was originally 29.970fps. MeGUI first determined it was "M-in-5 Decimate" and encoded it to 17.xxxfps. Then I tried it again and it saw it as "Hybrid Film mostly interlaced" with resulted in the fps staying at 29.970fps. Then I went and encoded it a third time, chosing my own options. Based on MediaInfo which specified the video as "Interlaced - Top Field First" and Selected "Hybrid mostly film - Top Field First - TIVTC" and the result was a video at 23.976fps. While I should probably not be picking my own options based on a change in fps that i desire, I didnt notice a difference in the 3 videos, which I think I should have. Since i prefer an output of 23.976fps Progressive video, I have a feeling I will select the same settings in other interlaced videos (that are originally 29.970fps) if the resulting video does not look different than the one encoded with the recommended settings. I am sure that is All Kinds of Wrong, lol, so I figured I would mention it.

If you leave MeGUI's de-interlacing disabled and step through the frames in the preview (and if you're decoding via DirectShow make sure the decoder isn't having a go at de-interlacing) and you don't see any combing anywhere, it's probably progressive. If you see combing in every frame it's probably interlaced. If there's a 3:2 pattern of clean and dirty frames it probably requires IVTC. You'll need to check frames where there's movement but the combing looks like this: http://www.videonet.webspace.virginmedia.com/images/i_in1.jpg

If MeGUI doesn't agree with your eyes or it's a Hybrid and you're not sure if MeGUI is getting the de-interlacing right etc, post a sample if you need to.

Oh yeah, Yandif with Bob resulted in 50fps which is much too high for me. I realize that people used to PAL video as apposed to NTSC might also be used to higher speeds, but when I watch video at a high fps it looks like it is being "fast forwarded" to me and it not relaxing and enjoyable to watch. Of course that might just be my opinion.

It's sometimes referred to as the "Soap Opera" effect as daytime soapies are often interlaced video (or they were), but interlaced NTSC is 29.970fps so it'd be de-interlaced to 59.940 (a higher frame rate than PAL ). Most TVs/players de-interlace the same way and at similar quality to Yadif. Stick an interlaced NTSC DVD in a Bluray player and watch it on your TV and it'll no doubt be de-interlaced to 59.940fps. If you de-interlace with Yadif the same way and motion looks a lot different, something's probably wrong.

[jagabo]

I think the problem with the video using DirectShowSource was just a fluke, however here is the original and shakey video you asked for:

The original isn't perfectly smooth but maybe the right pulldown removal wasn't used or maybe MeGUI got the frame rate wrong and DirectShow is trying to convert it. The only way to know for sure is to look at an untouched sample of the original.

I downloaded the videos from the site. The "original" as 29.97 fps with a duplicate frame every 5th frame (ie, 23.976 fps progressive converted to 29.97 fps by duplicating every 4th frame). The processed video was decimated to 23.976 fps but every pair of frames was identical. So it had only 11.99 different frames every second. Hence the jerkiness.

[hello_hello]

Is there a download link on that site I'm too stupid to find or did you use a video downloading program?

[jagabo]

Is there a download link on that site I'm too stupid to find or did you use a video downloading program?

I right clicked on the video while it was playing and selected "Save video as". I wasn't expecting that. Just stumbled across it.

[manofsteel31]

Is there a download link on that site I'm too stupid to find or did you use a video downloading program?

If you want to download them and have any problems then I could upload them to a hosting site like Hugefiles or something. I didnt know where to upload them so they could be streamed and I just picked that site (never used it before) because it didnt make me create an account to upload them, which was easier. I obviously trimmed it as its only 11 seconds of video so its only a few MB in size. So its not a big problem to upload somewhere else, however as was mentioned, it looks like you can just right click the video and save it.

**And Yes, the Original is not perfectly smooth either, but its much better than that encode. As the motion scrolls, it seems to stop and restart the whole way giving it that Shakey or Choppy look.

I downloaded the videos from the site. The "original" as 29.97 fps with a duplicate frame every 5th frame (ie, 23.976 fps progressive converted to 29.97 fps by duplicating every 4th frame). The processed video was decimated to 23.976 fps but every pair of frames was identical. So it had only 11.99 different frames every second. Hence the jerkiness.

I re-encoded the video and I think its much better now. Just the crud was removed and it was resized to 706x524, and I think its as smooth in the opening sequence as the original. I believe the video was analyzed as Progressive so no deinterlacing was needed. MediaInfo lists the Original Video as "Progressive 2:3 Pulldown". Unless someone sees a problem with it, I think its good. Take a look:

[jagabo]

I re-encoded the video and I think its much better now.

Yes, that one is properly decimated back to each film frame. No missing frames, no duplicates.

[hello_hello]

If you want to download them and have any problems then I could upload them to a hosting site like Hugefiles or something. I didnt know where to upload them so they could be streamed and I just picked that site (never used it before) because it didnt make me create an account to upload them, which was easier.

I'd just attach samples to a post. It's easier and they'll always be there. Some sites delete video/pics after a certain time. It appears there's a 500mb file size limit for attachments. When you're uploading samples, especially if they're small, I don't think there's a need to make them stream-able.

[manofsteel31]

OK, so I encoded this 1920x1080 bluray extra. Megui Analyzed it as "Hybrid - Mostly Film / Top Field First / TIVTC". I think the encode looks fairly good, but there are several scenes where the video becomes Choppy and I am not sure why. I THINK the original video is actually changing FPS during a few scenes which is causing problems after I encode it, although I am not sure. One example is this scene where the camera circles around a Chandelier on the ceiling. In the original it seems to change speeds and in my encode it looks very Choppy and not at all like smooth motion. If that is indeed the problem, then how would i fix something like that?

Here is a approx 30 sec video:

Original - Supposed to be 29.970fps as original .m2ts file, but strangely after muxing into an .MKV it changes to 30fps:



Encoded - 23.976fps (Hybrid - Mostly Film / Top Field First / TIVTC):




Also, in a completely Unrelated question .... Blu-ray movies often time comes with DTS or TrueHD audio which I seem to have a problem encoding with MeGUI. For example, when i encode DTS audio the Audio is in sync, but Ends before the Video does in the movie, so the last 1min or so of audio is missing in the movie. I fix the problem by using MeGUI to Extract the Audio as .FLAC first and then encode it as usual. Is there a better way to do it?

Even worse is the rare "TrueHD+AC3" audio which I have a hard time extracting at all. I think that only applied to "Dolby Atmos", but I am not positive. That one is a REAL pain.

[manono]

All I could download was 640x360 MP4s, 29.97fps for what was supposed to be from the source, and 23.976fps from what was supposedly the reencode.

Those were by right-clicking the video and 'Save Video As'. When I hit the downpointing arrow it asked me to sign in or register. No thanks. No sign of an M2TS file.

[manofsteel31]

OK, so I figured out how to upload the files to this site instead, so they should be attached.

The video was originally in a .m2ts file on the bluray disc and was listed as 29.970fps. In order to trim 30 seconds of video to upload here, I muxed it into a .mkv which strangely changed it to 30fps (no idea why). That is what i was saying. I was NOT uploading the .m2ts file, just a muxed version. So both the ORIGINAL and ENCODE are both .mkv files and should be 1920x1080, not 640x360 (the streaming site must have resized it on its own).


EDIT: The Original MKV Below is actually 29.940fps now (no idea why since the original .m2ts is 29.970fps. It was simply muxed and split based on timecodes in MKVMerge).

[hello_hello]

I'll defer to manono's expertise on the de-intrlacing, but until he comes along I think MeGUI got the de-interlacing correct and the frame rate wrong. I'm in PAL-Land though so I'm not exactly an NTSC expert.

I don't think the frame rate changes at all. when I opened you MKV sample with ts muxer it displays 29.970 as the frame rate. I think MediaInfo is getting it wrong and causing MeGUI to get it wrong. The version of MediaInfo I'm using shows this for your MKV:

Frame rate : 29.940 fps
Original frame rate : 29.970 fps

"Frame Rate" is the frame rate being passed along to MeGUI which in turn tells ffms2 to use that frame rate. It only changes the frame count by four but that's still four less frame. Without MeGUI telling ffms2 the frame rate it decodes it as 59.940fps progressive which is not unusual for ffms2 and VC1. The version of ffms2 MeGUI uses is the latest stable version but it's not been updates in quite a while. The moral of the story would be don't index VC1 with ffms2.

Here's the ffms2 line from the script MeGUI creates (yours might be different depending on the index file location but that doesn't matter)

FFVideoSource("E:\Original.mkv", cachefile="D:\Original.mkv.ffindex", fpsnum=1497, fpsden=50, threads=1)

The frame rate part is in blue. Change it to this and ffms2 will output the correct 29.970fps frame rate (if it'll decode it properly).

FFVideoSource("E:\Original.mkv", cachefile="D:\Original.mkv.ffindex", fpsnum=30000, fpsden=1001, threads=1)

In theory, you don't need it at all, if ffms2 always got it right.

MeGUI doesn't automatically add frame rate conversion to the script for L-Smash. You can add it yourself but for me L-Smash is getting it right without it.

LWLibavVideoSource("D:\Original.lwi")
or
LWLibavVideoSource("D:\Original.lwi" fpsnum=30000, fpsden=1001)

For L-Smash, it seems to be bottom field first. Or I didn't set a field order and it didn't seem to matter I guess ffms2 was getting that wrong too? Might explain some of the jerky motion.

So this works:
tfm(order=-1).tdecimate(hybrid=1)

Although there seems to have been some sort of frame blending horror inflicted on that in line with the generally &%^%-up state extras seem to often be in by the time they make it to disc, and it seems to be different for the interlaced and TVTC sections. I think the first part is interlaced and would be frame blended to 23.976fps when probably makes it look worse. I don't know if that's worth tackling. Probably not, and de-blending at 1080p would be painfully slow, but manono might have a clever idea. The sample attached to this post is as far as I got.

Also, in a completely Unrelated question .... Blu-ray movies often time comes with DTS or TrueHD audio which I seem to have a problem encoding with MeGUI. For example, when i encode DTS audio the Audio is in sync, but Ends before the Video does in the movie, so the last 1min or so of audio is missing in the movie. I fix the problem by using MeGUI to Extract the Audio as .FLAC first and then encode it as usual. Is there a better way to do it?

Whether there's a better way might depend on how you're doing it when you're not extracting the audio as flac first. Actually how do you define "extract the audio as flac? Just so I know we don't end up talking about different things. There shouldn't be a need for that though. Are you loading the extracted dts audio into the audio section and encoding it that way?
You can use the HD Streams Extractor under the Tools menu for extracting audio. I use it to rip Bluray discs (AnyDVD running in the background, and if you have NeroAacEnc.exe in the "MeGUI\tools\eac3to" folder it'll convert to AAC as it extracts. There's several ways to extract and convert, which is why I ask.

To be honest I rarely use MeGUI for audio encoding as I have foobar2000 set up for that. MeGUI's big limitation when it comes to audio is the "one audio job at a time" limit. Apparently, that was to stop some "popping' issues when encoding more than one audio job simultaneously via Avisynth. I don't know if it's still a problem but the limit remains.
With foobar2000 I can open a bunch of files via the right click menu, highlight the ones to convert, right click and select a pre-configured conversion preset, and it'll convert as many simultaneously as you have CPU cores until it's done. You can even convert the same files to different formats at the same time. It doesn't have the same normalising function as MeGUI (which I rarely use) but it'll open Avisynth scripts (via a plugin), so you can open audio via a script and use Normalise() if you need to, you can add DSPs to the conversion chain (which I don't do much aside from a matrix DSP now and then to fiddle with how a track is downmixed or adjust the channel volumes/order), and it'll convert using pretty much any command line encoder.

I'm not aware of any problem that'd cause the last minute not to be re-encoded for DTS all the time. If it only happens for a particular file I'd guess it's a decoding issue (the decoder may simply stop prematurely) but "all the time" would be odd.

Even worse is the rare "TrueHD+AC3" audio which I have a hard time extracting at all. I think that only applied to "Dolby Atmos", but I am not positive. That one is a REAL pain.

I'm not sure I've ever worked with any. How are you extracting it?

[manofsteel31]

I did not notice that the ORIGINAL 30-second sample I posted was actually 29.940fps. Thats very strange ...... let me try this again:

The original video is a .m2ts file and is 29.970fps. When I muxed the entire movie into a MKV using MKVMerge it changed to 30.000fps. Then when I muxed that same original untouched m2ts file again, but this time I split the video by timescodes (because splitting by frames was giving me problems) it resulted in the 30 second sample I wanted but this time its listed as 29.940fps. What the Heck is going on here? Muxing and splitting should not affect the frame rate.

BTW ... I am using v7.70 on MediaInfo. I just tried installing the newest v7.73 but it kept giving me an annoying popup "external exception error 80000003" that I did not feel like troubleshooting at the moment, so I re-checked the videos and the FPS was the same, so I just re-installed 7.70 for now. I dont think my version is a problem. I think it was my eyes, lol.


I was actually worried that the muxed version might be a problem since it stated 30fps, so I used the "HD Streams Extractor" in MeGUI to extract the video as a ".h264" file to try and maintain the 29.970fps and used that as my source, which resulted in the 23.976 encode I uploaded for you. So i guess it didnt make a difference. That seemed to be my only choice since when I tried loading the .m2ts file directly it would not Analyze it. It would just freeze during the Analyze process and never finish.

--------------
When trying to encode from the muxed .mkv I get:

FFVideoSource("D:\original.mkv", fpsnum=30, fpsden=1, threads=1


When encoding from the .m2ts file I get:

LWLibavVideoSource("D:\original.m2ts")

OR

FFVideoSource("D:\original.m2ts", fpsnum=30000, fpsden=1001, threads=1)


When using the extracted .h264 file I get:

AVCSource("D:\original.dga")
--------------

Whether there's a better way might depend on how you're doing it when you're not extracting the audio as flac first. Actually how do you define "extract the audio as flac? Just so I know we don't end up talking about different things. There shouldn't be a need for that though. Are you loading the extracted dts audio into the audio section and encoding it that way?
You can use the HD Streams Extractor under the Tools menu for extracting audio. I use it to rip Bluray discs (AnyDVD running in the background, and if you have NeroAacEnc.exe in the "MeGUI\tools\eac3to" folder it'll convert to AAC as it extracts. There's several ways to extract and convert, which is why I ask.

To be honest I rarely use MeGUI for audio encoding as I have foobar2000 set up for that. MeGUI's big limitation when it comes to audio is the "one audio job at a time" limit. Apparently, that was to stop some "popping' issues when encoding more than one audio job simultaneously via Avisynth. I don't know if it's still a problem but the limit remains.
With foobar2000 I can open a bunch of files via the right click menu, highlight the ones to convert, right click and select a pre-configured conversion preset, and it'll convert as many simultaneously as you have CPU cores until it's done. You can even convert the same files to different formats at the same time. It doesn't have the same normalising function as MeGUI (which I rarely use) but it'll open Avisynth scripts (via a plugin), so you can open audio via a script and use Normalise() if you need to, you can add DSPs to the conversion chain (which I don't do much aside from a matrix DSP now and then to fiddle with how a track is downmixed or adjust the channel volumes/order), and it'll convert using pretty much any command line encoder.

I'm not aware of any problem that'd cause the last minute not to be re-encoded for DTS all the time. If it only happens for a particular file I'd guess it's a decoding issue (the decoder may simply stop prematurely) but "all the time" would be odd.

What I meant was I use MeGUI's "HD Streams Extractor" to extract the DTS or TruHD audio as .FLAC. Which I would presume is essentially converting it to FLAC. I then use that FLAC file as my source Audio and encode it as usual with the video.

While I do Have NEROAAC installed and setup through the Megui settings, I truely HATE AAC audio with a passion. So I encode all my audio to "2-channel 192Kbps AC3" using the DirectShow Decoder. Yes I always normalize peaks to "100" and Apply Dynamic Range Compression.

I have to use the "NicAudio" decoder sometimes when I am trying to encode AAC audio to AC3, but its never turns out any better anyway because AAC really ruins the audio to the point of no return. Just My Opinion of course.

Since I watch my videos on my tv directly, AC3 works MUCH better for me that AAC, so I avoid aac like the plague, lol.

Never used foobar2000 before ... or even heard of it.

Dolby Atmos is in several movies. I think the first was the last Transformers movie. I dont remember how I ended up finally encoding it. I remember trying to use TS Muxer, then trying to extract it as a RAW audio file and several other ways and I dont remember which one ultimately worked. I have come across it a few times now and its a real pain in the @ss each time.

Anyway, I do have problems every single time I have tried to encode 6/7/8 Channel DTS or TrueHD audio down to 2-channel AC3. I always lose the end of the audio. Its not a HUGE deal for newer movies (unless they have an extra scene after the credits), but OLD movies used to put the credits at the beginning of the movie, so you actually lose part of the movies audio if the last minute is missing. Anyway, thats why I have been extracting it as FLAC then encoding it to AC3 to fix the problem. I just figured there must be a better way.

[newpball]

Yes I always normalize peaks to "100" and Apply Dynamic Range Compression.

Why?

[manofsteel31]

Yes I always normalize peaks to "100" and Apply Dynamic Range Compression.

Why?

When I first started encoding I followed a lot of online tutorials that told me to use Nero AAC as my audio. Well, I ended up Hating it because it did not work well when played on my TV, or other TV's I tested. It played fine on my computer, but on my tv I would have to turn the volume all the way up to near MAX VOLUME just to hear the AAC audio. It was terrible. So I tried to figure out Why it could be so terrible yet still be so popular with people who encode.

I started running test encodes using every single possible audio encoding option that Handbrake and MeGUI had available. I was not really even sure what all the different options did or why i should choose one decoder over another, so I just tried them all. Different decoders, different bitrates, different encoding types, different normalize values etc. I finally settled on those same settings for my AAC audio, but was still not happy with how low the volume was (which appears to be the fault of AAC audio in general). So when I switched to AC3 audio, I just kept all my settings the same because I did not want to mess with anything and screw up the good results I was getting with my AC3 audio. The audio I have been encoding has been excellent as far as I am concerned. I can play my encoded videos on my tv with the same volume setting I use to watch normal TV and I can hear the audio great. No problems at all.

Again, I dont even remember what either setting really does, although I THINK they make the audio set to a constant level throughout the movie as apposed to one scene being too loud and the another too low......... but not really sure.

I can tell you that I really hate watching a movie when one scene has an extrememly loud explosion that makes me want to turn the volume DOWN and the next scene has the characters Whispering which makes me want to turn the volume UP. I want it constant throughout the movie.

[hello_hello]

I did not notice that the ORIGINAL 30-second sample I posted was actually 29.940fps. Thats very strange ...... let me try this again:

The original video is a .m2ts file and is 29.970fps. When I muxed the entire movie into a MKV using MKVMerge it changed to 30.000fps. Then when I muxed that same original untouched m2ts file again, but this time I split the video by timescodes (because splitting by frames was giving me problems) it resulted in the 30 second sample I wanted but this time its listed as 29.970fps. What the Heck is going on here? Muxing and splitting should not affect the frame rate.

It's probably partly due to the way different containers store timecodes. I think for MKV it's rounded to 1ms. If you work it out, for 29.97fps the frame duration is 1000ms/29.97fps = 33.3667ms. So the first frame duration might be 33ms, then 34ms, then 33ms and 33ms and 34ms..... something like that...... but it's probably one reason why MediaInfo doesn't always get it exactly right or reports it slightly differently after remuxing. Then MeGUI uses the frame rate reported by MediaInfo if it needs to add it to the script.

When trying to encode from the muxed .mkv I get:
FFVideoSource("D:\original.mkv", fpsnum=30, fpsden=1, threads=1
When encoding from the .m2ts file I get:
LWLibavVideoSource("D:\original.m2ts")
OR
FFVideoSource("D:\original.m2ts", fpsnum=30000, fpsden=1001, threads=1)
When using the extracted .h264 file I get:
AVCSource("D:\original.dga")

MeGUI only adds the frame rate conversion for FFVideoSource, DirectShowSource and AVISource. Hopefully soon also for L-Smash, but when it's not added by MeGUI it's up to the decoder to determine the frame rate. If you want to know what that is, put Info() at the end of the script and refresh MeGUI's preview.

Your video isn't going to be anything other than 29.97fps. Well 29.97fps is itself a slight rounding of the frame rate. It's more accurately expressed as the fraction 30000/1001 (= 29.97002997002997002997002997003). The other frame rate it might pay to keep in your head is 23.976, which is 24000/1001.

So for your video "fpsnum=30, fpsden=1" would be wrong and "fpsnum=30000, fpsden=1001" would be correct. You should fix it manually if MeGUI adds it and if not use Info() to check it's correct.

While I do Have NEROAAC installed and setup through the Megui settings, I truely HATE AAC audio with a passion. So I encode all my audio to "2-channel 192Kbps AC3" using the DirectShow Decoder. Yes I always normalize peaks to "100" and Apply Dynamic Range Compression.

I have to use the "NicAudio" decoder sometimes when I am trying to encode AAC audio to AC3, but its never turns out any better anyway because AAC really ruins the audio to the point of no return. Just My Opinion of course.

Since I watch my videos on my tv directly, AC3 works MUCH better for me that AAC, so I avoid aac like the plague, lol.

When you downmix to stereo a formula is applied that reduces the individual channel volumes by quite a bit before they're combined. It's to prevent the peaks of the combined audio being louder than maximum level, or "clipped". It's "worse case scenario" so often the volume is reduced until it's quite low. Normalising adjusts it again until the peaks are at maximum, then it's encoded. You probably can't avoid normalising when downmixing to stereo, but the same applies no matter what the format you use for encoding. I've been encoding using AAC almost exclusively for years and I can promise you there's nothing wrong it. Neither of the Samsung TVs in this house have an issue decoding it and both the USB Bluray players do likewise.

If you're using DirectShow for decoding though all bets are off to a certain extent because a Directshow decoder could be doing anything to the audio as it decodes. If ffdshow's decoding, for example, it can adjust the volume, down-mix, swap channels around....... anything if any of it's filters are enabled.

I think the "Apply Dynamic Range Compression" option only applies to AC3 audio that contains compression information. If it does the decoder uses it and the audio is compressed a little as it's decoded and encoded that way. For AC3 without dynamic range compression information or other audio formats the setting won't make any difference.

Having said all that..... I use my PC as a media player with ffdshow decoding the audio and compressing it on the fly with a Winamp plugin. There's no reason why it couldn't compress as you encode and it does a very good job, but you can't "uncompress" once it's encoded. Have a look in your TVs speaker settings under Sound and the "Auto Volume" option. "Off" is...... well it goes without saying. "Normal" compresses a bit and "Night Mode" compresses more. See if that helps.

Anyway, I do have problems every single time I have tried to encode 6/7/8 Channel DTS or TrueHD audio down to 2-channel AC3. I always lose the end of the audio. Its not a HUGE deal for newer movies (unless they have an extra scene after the credits), but OLD movies used to put the credits at the beginning of the movie, so you actually lose part of the movies audio if the last minute is missing. Anyway, thats why I have been extracting it as FLAC then encoding it to AC3 to fix the problem. I just figured there must be a better way.

Decoding with something that's not DirectShow would be the first place I'd start testing.

[Cornucopia]

@hello_hello, IIRC the MKV timebase is 90000Hz (aka 1/90000sec), so a 30fps file would have 90000/30=3000 clicks of its clock. And yes, I believe you are right in the fudging being done to keep it even overall.

Scott

[hello_hello]

This zip file contains some short audio samples I was playing with a while back comparing compression. Or it's attached to the bottom of this post if that link doesn't work. It's fairly dynamic with quiet speech in the first part, gunshots and sirens in the second. I wanted to see if it could be compressed without the compression being too audible.

1. Is the original (no compression)
2. Is the compression I normally use on playback (RockSteady plugin running in ffdshow).
3. Is a compressor DSP based on ReplayGain for foobar2000
4. Is ffdshow running a different Winamp compressor (LoudMax)
5. Is Levelator. Designed for speech but I thought I'd give it a spin.

See what you think, although sorry....... they're AAC.

[hello_hello]

@hello_hello, IIRC the MKV timebase is 90000Hz (aka 1/90000sec), so a 30fps file would have 90000/30=3000 clicks of its clock. And yes, I believe you are right in the fudging being done to keep it even overall.

Scott

Ahhh......

Cheers.

[manofsteel31]

OK, so this particular video is becoming a real pain for some reason. Luckily when I use the "Auto Encode" button and select the checkbox for "Add Additional Content" before hitting Queue, I get the Adaptive Muxer Screen which shows me what the resulting fps will be before I actually encode it. So here are the results it shows me:

FFMSIndex and L-Smash Works Analyze the video as "Hybrid - Mostly Film / Top Field First / TIVTC", which both result in 23.976fps (even when I manually specify "fpsnum=30000, fpsden=1001")

DirectShowSource Analyze the video as "M-in-5 Decimation / M=5 / Tritical Decimate", which results in 5.99fps.

Just because, I tried a few other options as well just to see how they would turn out:

"Interlaced / Top Field First / Field Deinterlace" resulted in 29.970fps and made the chandelier motion scene look better, but ruined some other motion scenes.

"Interlaced / Top Field First / Yandif with BOB" and " Interlaced / Top Field First /TDeint with Bob" both resulted in 59.94fps.

I even tried treating the video as Progressive just so no deinterlacing would be done, and the results were still not ideal.

I would consider uploading the whole .m2ts file for you (since its only 7-8 minites) but its also 1.66GB.

BTW ... L-Smash took roughly 4 Hours just to Analyze this 7-8min video. Normally I would never have waited that long but I left it running before going out for a few hours and came home and it was still not done. Then I finally saw it finish. Usually videos only take a few short minutes to analyze, if that, so not sure what the problem with L-Smash is.

Oh, I did check my TV and "Auto Volume" is set to "Off". Are you saying that changing it to Normal would improve AAC audio on my TV? I have not had a chance to check the AAC files you upload yet.

So what is wrong with using the "DirectShow" Decoder? Which one do you recommend? In MeGUI I appear to have the following options (that I am aware of):

FFmpeg AC3 or Aften AC3. I have been using FFmpeg AC3 (though not sure of the difference).

Then for Decoders I have a choice of:

-NicAudio
-FFAudioSource
-DirectShow
-BassAudio
-LWLibavAudioSource

If I should not be using DirectShow, then which would you choose? Or is there a way to add another one which is better?


Another Question. I have always encoded by Bluray movies by selecting the ".mpls" file for the movie. OneClickEncoder and DirectShowSource can both handle this type of file, but it appears that when I try to open it in "FileIndexer" that neither FFMSIndex or L-Smash can. FFMSIndex errors out when trying to create the .ffindex file, and L-Smash completely crashes MeGUI when it tries to create its .lwi file. So if you are content that I always use the FileIndexer, then how do you recommend that I do it? Should I be using the .m2ts file instead or should I be using MKVMerge or MeGUI's Stream Extractor to mux the video into a MKV first? OR should I just use DirectShowSource for them since they are usually Progressive anyway and no Analyzing for Deinterlacing is typically needed?

[jagabo]

If you're still talking about the video in post #75: That video is a field blended frame rate conversion where the change varied over time. This was done to speed up or slow down the motion at different points in the clip -- ie, an artistic choice. So you will not be able to reduce it to a single frame rate and have no duplicate, missing, or blended frames. The "problem" may be limited to the intro so you should check the body show.

[manofsteel31]

If you're still talking about the video in post #75: That video is a field blended frame rate conversion where the change varied over time. This was done to speed up or slow down the motion at different points in the clip -- ie, an artistic choice. So you will not be able to reduce it to a single frame rate and have no duplicate, missing, or blended frames. The "problem" may be limited to the intro so you should check the body show.

Thats kinda what I thought. The deliberate slowing down and speeding up of the motion is what was causing the problem. There are 3 or 4 points in the video where this happens and they are unfortunately all spread out through the video (I just posted the first occurance which was the Chandelier scene).

So considering the changes, how should I go about encoding it?

[jagabo]

If you smart bob to 59.94p the motion won't be any worse than it is now. But the resulting video may not play on some devices and slow computers.

[manofsteel31]

Not sure what "Smart" BOB is.

I have to select "Source Type" as either "Interlaced" or "Partially Interlaced" to get any "with BOB" options.

Then my "Deinterlace" choices are:

-Yandif with BOB
-TDeint with BOB
-TDeint with EDI and BOB

I just tried to encode the muxed MKV version of the video (which MediaInfo claims is 30.000fps) using ffindex and manually changing "fpsnum=30, fpsden=1" to "fpsnum=30000, fpsden=1001" using "Interlaced / Top Field First / Yandif with BOB" and that resulted in a 60fps video with random full screen pixelation which makes it unwatchable. I dont know if it would make any difference if I used the m2ts file @ 29.970fps instead. That should have resulted in 59.94fps instead, but I have a feeling the encode would not be much different.