Really need help encoding this anime (Gintama)

[jagabo]

Can you upload an original VOB/MPG/m2v of the logo?

[Taizenshū]

Here.

[jagabo]

Here.

Thanks for the sample. Yes, the flickering edges are in the source. There was no problem with your processing of the sequence.

[Taizenshū]

Okay now I'm having an issue with the audio, it seems that the Audio Cutter isn't working as how it should. I would set the frames to cut, but there seems to a 2 min delay every time. Any solution to this?

[ndjamena]

I'd like to know the answer to this question too. I hope you mean 2 seconds, that's been an issue for as long as I can remember. I'm wondering if it's ac3 thing or if it affects all codecs?

[Taizenshū]

I hope you mean 2 seconds,

Nope, I actually mean 2 min (1:58 on VLC). It's really starting to piss me off.

[ndjamena]

Are you using DGIndex to cut it or something else? Are you re-encoding the audio? You could try MKVMerge, AVISynth Trim(), or PCM, or you could process the audio separately.

-Edit- By 'delay' so you mean 'out-of-sync' or that it just takes too long to start up?

[poisondeathray]

Or you can specify audio in the script . This way you can preview to see if it's in sync before wasting time encoding errors

[Taizenshū]

This is how it is:
-It takes 1:59 to start
-It doesn't end at the frame where I want it to (instead ends a few frames earlier).

So how do I use trim()? And how do I specify it in the script?

[coldstatic]

I'm not that good with the programs your using but my understanding is that when you detelecine/interlace you lose frames but it does not change the audio. For something like that i would normally use avidemux and in the audio resample it along with changing it from film to pal or pal to film which ever way you're going. Take that thought with a grain of salt though as i am no expert like some others here.

Not trying to hijack but this seems close to what i am dealing with. Cowboy Bebop. YEs i know there are bunches of threads out there about this one b/c its a tough one. But i believe the newest one with any relevant info on it was circa 2006. My thoughts are that the filters have been greatly upgraded and refined from those old days.

I started with deinterlace but after looking at it i believe its telecine. So i switched to that and no effect. Tried a the decomb with diff options. Tried combo's of detelecine with decomb diff options. Nothing seems to work. I have set video frame to constant same/23.97 and variable same as source.

Strangely most of my results are the vid looks the same as dvd. Any thing else actually seems to make everything worse in some way. one weird event with detelecine and deinterlace the scene i am examining looked great but the rest looked like crap. most of the time the out of sync/interlaced parts get scrambled a bit more but extended further over the scene. scene goes like this clean, interlaced, aliased, clean.
media info says progressive with 2:3 pulldown.
The zip is the screenshots from vlc not modified in any way. Also a gif of them one again not modified. At there native resolution you really see the first interlaced but the frame after that needs to be anti-aliased is not very noticeable. When i go full screen in vlc or if you use irfanview to full screen the img/gif you will see the AA as clear as day. Also attached is mediainfo export as html. you have to remove .txt to see it formatted though.

[jagabo]

Your GIF files are useless. The frames have been resized, corrupting the field structure. Post a sample of the original video (VOB?).

[Guest34343]

Post a sample of the original video (VOB?).

+1

We can surely help you if you provide an appropriate sample. I've done a lot of work with Cowboy Bebop when developing Decomb, and I'm sure jagabo has too. If you can tell me the episode and time I can post a sample for you also. manono too has a lot of experience with this.

[coldstatic]

https://doc-14-30-docs.googleusercontent.com/docs/securesc/ha0ro937gcuc7l7deffksulhg5h...14193614665626

the image size should be correct....pulled screenshot from dl'ed zip file. disk 1 episode/session 2 5:40. just after he makes the guy drink the ****roach.

its been a long time from when i had to work with dvd's (circa 2003-4). Do you just want me to upload the whole vob? I can put it on a cloud drive somewhere. Or is there an easy way to just snag part of the vob without re-encoding?

Thanks in advance btw. This is one of my all timer favorites and would die if it got damaged.


ps....have tmpgenc just looking for guide on vob's...should only be a few min

[Guest34343]

You can do either of these:

1. Cut the VOB with DGSplit and upload the resulting piece somewhere.

2. Open the VOB in DGIndex, set a range with the [ and ] keys, then do Save Project and Demux Video. Upload the resulting M2V file somewhere.

Meanwhile I will try to find that on my copy.

Don't bother with screenshots; they are useless.

[coldstatic]

glad you posted that.....i have mpg2cut2 as well but it was not working right. but for that matter neither did this. I got an error while chopping the clip. A field order transition was detected correct it? i made one with the correction and one without. i will update this post with folder link....Its about 19% uploaded.


On a side note (b/c i'm curious) why are screen shots bad? my thoughts were that as long as you didn't convert the file to jpg from its original format (or just saved it to jpg but at 100%) it would be a direct image of the video. and for advancing the video i used vlc's frame by frame. If its a big long explanation don't worry about it..... anyway vids done uploading.

https://drive.google.com/folderview?id=0B0VDO3rn0zlgSnR1NHBLOUhMWjg&usp=sha ring
first vid with correction
first vid direct dl
second vid no correction (g on end of file)
second vid dl

[Guest34343]

This simple Avisynth script works fine on that (save the project with Honor Pulldown if using DGIndex):

Telecide()
Decimate()

The postprocessor doesn't even kick in on the scene contained in your GIF.

Don't worry about the field order correction. It only affects the D2V file. We only need the resulting M2V.

[coldstatic]

This simple script works fine on that:

Telecide()
Decimate()

The postprocessor doesn't even kick in on the scene contained in your GIF.

Don't worry about the field order correction. It only affects the D2V file. We only need the resulting M2V.

so is it just handbrake and avidemux are just bad at this? Not afraid of cli megui or whatever it is....just havent learned it yet but if now's the time then lets bring on the noise......lets bring on the pain.

in case you needed it here is my handbrake cli output. its not set right as far as most settings go. i was just testing for the telecine.


CLI Query: -i "G:\\" -t 7 --angle 1 -c 1-6 -o "Q:\Cowboy Bebop-7.m4v" -f m4v -4 -O --detelecine --crop 0:0:0:0 --strict-anamorphic --modulus 16 -e x264 -q 16 --cfr -a 3,1 -E copy:ac3,copy:ac3 -6 auto,auto -R Auto,Auto -B 0,0 -D 0,0 --gain 0,0 --audio-fallback ffaac --markers="C:\Users\COLD\AppData\Local\Temp\Cowboy Bebop-7-7-chapters.csv" -x threads=5:ref=10:bframes=6:b-adapt=2:direct=auto:analyse=all:trellis=2:aq-strength=0.6sy-rd=0.4,0.15:deblock=1,1 --verbose=1 --no-dvdnav

[Guest34343]

Sorry, I don't use Handbrake. I'm a pedal to the metal guy.

You need to do the equivalent of IVTC plus adaptive postprocessing for Cowboy Bebop.

[coldstatic]

"HandBrake uses a detelecine filter called pullup (because it's the opposite of pulldown). It's a very advanced inverse telecine filter from the MPlayer project." so basically just turn it on....which i did. as far ias i know HB dosent have a post processor. But avidemux can use avisynth Only now thats a whole new set of detelecine filters i have to test

Thank you for the help....Hopefuly in a day or two i will post some positive results. I know i could just do small test sections but i prefer doing one whole section focusing on one point but still having the rest to look at. It only takes 9min2sec to encode so meh.

[Guest34343]

"HandBrake uses a detelecine filter called pullup (because it's the opposite of pulldown). It's a very advanced inverse telecine filter from the MPlayer project."

Given the terrible result you showed, it's either not so advanced or you are using it wrong.

That section field-matches just fine without needing postprocessing. You don't have to test a whole set of filters; just make an Avisynth script as I showed.

[jagabo]

I can't get clean decoding of the samples. The picture starts out OK but after a dozen frames or so it breaks up:



It gets worse and worse until the next keyframe where it is cleaned up for a while and breaks up again. I see the same thing in several video players including VLC and SMPlayer which use their own decoders. Why are the GOPs so long (~60 frames)?

[coldstatic]

"HandBrake uses a detelecine filter called pullup (because it's the opposite of pulldown). It's a very advanced inverse telecine filter from the MPlayer project."

Given the terrible result you showed, it's either not so advanced or you are using it wrong.

That section field-matches just fine without needing postprocessing.

HAHAHAHAHAHAH lol . well there are only a few settings for it. default custom and off. i was going to try custome but was unable to locate what the switch settings are.


"-9, --detelecine Detelecine (ivtc) video with pullup filter Note: this filter drops duplicate frames to restore the pre-telecine framerate, unless you specify a constant framerate (--rate 29.97) <L:R:T:B:SB:MP:FD> (default 1:1:4:4:0:0:-1)"

from the HB guide "To enable detelecine, mark the detelecine checkbox in the Picture Settings.

Pullup is stateless. What this means is that it looks at every frame of the source video. It visually analyzes the video and looks for fields that can be woven together in ways that form progressive frames. For example, with NTSC hard telecined material, pullup sees that pattern of progressive-progressive-interlaced-interlaced-progressive and that the interlaced frames' fields can be swapped to create frames that show less combing. Each time it encounters that situation, it deinterlaces the frames that need it and makes a guess as to where the "breaks" in the pattern are. This lets it know which frames are the duplicates that should be dropped from a source.

When it does this, you're left with a pattern like this:

Frame 1: A1A2
Frame 2: B1B2
Frame 3: B1B2 (candidate for dropping)
Frame 4: C1C2
Frame 5: D1D2

Now, this will play fine. However, notice that Frames 2 and 3 are exactly the same. This means you're wasting bitrate and time encoding frames that don't matter, and in the end the video still has around 30 frames per second.

If you've specified that HandBrake should use a framerate of 29.97fps, this is exactly what HandBrake will encode for the output. But it is wasteful, because of the duplicate frames.

If you've specified that HandBrake should use a framerate of 23.976fps, it will drop those duplicate frames. However, if there are so many duplicate frames that the framerate drops lower than that, HandBrake will have to leave them in place to keep the framerate steady. Similarly, if there are so few duplicates that the framerate stays above the specified 23.976, it will be forced to drop some frames at random, with no guidance from the detelecining filter.

If you just leave the framerate as "Same as source," though, HandBrake will drop those duplicate frames and only those duplicate frames, with no extra dropping or copying of frames. We call this process Variable Frame Rate detelecine."


L:R:T:B:SB:MP:FD is what i need to know the settings for. if i knew those i bet i could get it right.

[coldstatic]

I can't get clean decoding of the samples. The picture starts out OK but after a dozen frames or so it breaks up:


[Attachment 25552 - Click to enlarge]

It gets worse and worse until the next keyframe where it is cleaned up for a while and breaks up again. I see the same thing in several video players including VLC and SMPlayer which use their own decoders. Why are the GOPs so long (~60 frames)?

I was getting the same results......GOPs ?? i goggled it and seeing things like should be 12 and N=3 N/M=5. But if its a setting i needed to adjust in dgindex i did not. I opened the program set my edits and hit the save and demux.

[Guest34343]

I can't get clean decoding of the samples. The picture starts out OK but after a dozen frames or so it breaks up:

I'll upload a nice clean sample for you later tonight. Anyway, I doubt you'll find this very interesting. The solution is simple as I have stated. Still, second opinions are always welcome. And maybe you can help with the Handbrake setup.

[Guest34343]

Here you go:

http://neuron2.net/misc/cb_sample.demuxed.m2v

[jagabo]

I was getting the same results......GOPs ?? i goggled it and seeing things like should be 12 and N=3 N/M=5. But if its a setting i needed to adjust in dgindex i did not. I opened the program set my edits and hit the save and demux.

It's a function of the source. DgIndex can't change it. It's the distance between key frames. DVDs are usually from about 12 to 18 (the max allowed for DVD). Neuron2's sample has normal 12 frame GOPs -- and a simple TFM().TDeciamte() returns it to 23.976 fps film frames.

Here you go:

http://neuron2.net/misc/cb_sample.demuxed.m2v

Thanks for that.

[Guest34343]

Just as I explained.

Telecide().Decimate()

and

TFM().TDecimate()

are pretty much equivalent and equally effective for this material. It's a matter of taste and/or experience with them that determines which one you use.

[jagabo]

Yes, I was just confirming your observation. Handbrake should work pretty well with that too. But there's something wrong with the OP's source. 60 frame GOPs is definitely not from a commercial DVD.

[coldstatic]

Pushed drive button and took this pic. I have all the disks. When purchased from a store afaik its commercial.

my media info said this as well.

Video
ID : 224 (0xE0)
Format : MPEG Video
Format version : Version 2
Format profile : Main@Main
Format settings, BVOP : Yes
Format settings, Matrix : Custom
Format settings, GOP : M=3, N=12
Duration : 3s 36ms
Bit rate mode : Variable
Bit rate : 423 Kbps
Width : 720 pixels
Height : 480 pixels
Display aspect ratio : 4:3
Frame rate : 29.970 fps
Standard : NTSC
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Interlaced
Scan order : Top Field First
Compression mode : Lossy
Bits/(Pixel*Frame) : 0.041
Time code of first frame : 00:59:51:00
Time code source : Group of pictures header
Stream size : 157 KiB (98%)

[jagabo]

There's no way a commercial DVD would be produced with 60 frame GOPs so something is going wrong. How did you rip (decrypt) the DVD?