VIDEO_TS Conversion - Poor quality

[WG_Grace]

QTGMC is one of the worst filters with which to start leaning AviSynth. Start with installing AviSynth and learning how to create a basic script.

http://avisynth.nl/index.php/Main_Page#New_to_AviSynth_-_start_here

Then get and install QTGMC and all the other filters it requires (put them in AviSynth's plugins folder):

http://avisynth.org/mediawiki/QTGMC

For MPEG 2 videos you should use DgMpgDec:

http://neuron2.net/dgmpgdec/dgmpgdec.html

Build and Index file (*.d2v) with DgIndex, then create an AviSynth script to import your video:

Mpeg2Source("video.d2v")
AssumeTFF()
QTGMC()

Finally, open that AviSynth script with an editor or encoder that supports AviSynth scripts. I use VirtualDub to preview scripts. You can make an AVI file with VirtualDub too.

Tried to give all of this a go but it's all a bit too much for me to understand.

[WG_Grace]

I've got a smooth conversion using the Bob deinterlace filter and using 50 fps output in Handbrake, but when I tried uploading to Youtube the video doesn't play smoothly and the quality is only 480p. Could someone help me on this issue?

[jagabo]

I've got a smooth conversion using the Bob deinterlace filter and using 50 fps output in Handbrake, but when I tried uploading to Youtube the video doesn't play smoothly and the quality is only 480p. Could someone help me on this issue?

Of course. Youtube reencodes to 25p. There is no 50p content at at Youtube.

[manono]

...and the quality is only 480p.

What was the resolution you uploaded? 720x576? If so, 480p is the highest you'll get. The next highest is 720p and you'll have to upload a video of 720p or higher to get that.

And if you did upload a 720x576 video, didn't you notice the people were taller and thinner than they should be? And is this the first time you mentioned this whole exercise was for YouTube?

[WG_Grace]

Of course. Youtube reencodes to 25p. There is 50p content at at Youtube.

Is there a solution to this?

[jagabo]

Of course. Youtube reencodes to 25p. There is 50p content at at Youtube.

Is there a solution to this?

Oops, I left out a word. I meant to say "There is NO 50p content at Youtube." There is no way around them reencoding to 25p.

[WG_Grace]

And is this the first time you mentioned this whole exercise was for YouTube?

Wasn't planning to initially, this is just something additional. I'm very grateful for the responses and solutions people posted earlier, they helped a lot.

[manono]

And is this the first time you mentioned this whole exercise was for YouTube?

Wasn't planning to initially, this is just something additional.

Yeah, but you could have saved yourself and others the time and effort of trying to walk you through making a 50fps video, if it was to be all for nothing. The output formats and intended destination (internet, computer playback, a hardware device attached to a TV set, etc.) are very important bits of information.

I'm very grateful for the responses and solutions people posted earlier, they helped a lot.

Yep, quite a learning experience, isn't it?

[WishMe]

maybe you have a imac, if you want to play/edit your video, you'd better remux you video into mp4, but if your codec not support,you need convert your video into prores, i know ffmpeg can also output prores video stream, maybe you can try it

[codemaster]

sanlyn procedure is excellent, however I would like to add an additional step to it, that I think is necessary because of the type of camera the OP uses.

I have a similar camera, but from Canon, that also writes .vob, .bup and .ifo files on mini dvd disc. I inform the OP that footage from these dvd camcorders will be very hard to edit if he doesn't know what I explained below.

Camera generates one or two VTS (video title sets). Many individual and separate clips, are stored in one VTS, each inside what is probably called a "vob cell". When you open the vob or vobs from VTS 1 or VTS 2, in DGIndex, and save the .d2v project, DGIndex will demux the .ac3 audio from that VTS. That .ac3 audio file will probably be properly synchronized with the first 2 or 3 individual clips, but if there are more than 3 individual clips in that VTS, that .ac3 audio file will no longer be properly synchronized with the rest of the individual clips. You will get audio delay, which is extremly difficult and time consuming to fix, because you have to fix it for every clip except the first 2 or 3. One way to fix it is with avisynth (using DGDecode.dll and NicAudio.dll plugins, the DelayAudio() function, and repeatedly previewing with virtualdub).

If you use DGIndex or VOB2MPG, you'll get audio delay. Not always. It depends on how many individual clips are in a VTS, and in what vob cell they are.

If you use Premiere Elements (which has support for dvd camcorders and for importing footage from them), it will import the vob from a finalized mini dvd, but the audio will have delay.

If you use VideoReDo TVSuite to open the dvd to cut clips from it, VideoReDo will not be able to navigate in certain points in it's timeline. You click or go to a certain time, or skip forward 2 min or 10 min, and it will not be able to go there, so you won't be able to cut.

If you use DVD Decrypter to extract the dvd chapters into separate vob files, because each chapter is an individual clip in a VTS (the camera separates clips by dvd chapters, not by VTS), you will get audio delay.

Sony Creative Software develops a non-linear video editor called Vegas Pro. In Vegas, they added native support for DVD camcorders, like the sony handycam that OP uses, and the Canon DC100 like I use. In Vegas, you can go to "File->Import->DVD Camcorder Disc...", select the drive in which you inserted the finalized mini dvd disc, or select the folder where you copied the VIDEO_TS folder, and Vegas will copy all clips from the dvd to your hdd. It will extract each individual clip from a certain VTS, and put each individual clip in it's individual .mpg file. It copies the clips, it does not transcode them. It extracts them from a VTS, it wraps them in mpg container, and most importantly, none of the mpg clips have audio delay. I found that Vegas is the only software that was able to maintain proper audio synchronization for all individual clips. None of those mpg clips imported by Vegas have audio delay, just like none of them have audio delay when you watch the finalized mini dvd on a dvd player or a software player like mpc-hc.

If you edit many of these mpg clips in Vegas, you will have problems such as too much ram usage, and inability to render, which means it can't save, to a file, what you edited, so that you can upload that file to youtube. It can also cause Vegas to take too much time to load the project, or to crash. If you import and edit many of these mpg clips in Premiere Pro, Premiere will not be able to open your project. These mpg clips are problematic for Vegas and Premiere, perhaps because they are coded with an inter-frame codec, and perhaps because on some of the clips, the first keyframe is missing. The solution is to use, for editing in NLEs, clips coded with an intra-frame codec, such as Lagarith, DV, MJPEG. For example, if you want to edit the clips in iMovie, you can use avisynth and ffmpeg to process and convert the mpg clips to dv or mjpeg, and audio to pcm, in .mov container, and after editing, perhaps you can render to PNG lossless in .mov container.

So, the simplest way to edit footage from that camera, is to use Vegas, but you could have problems if there are many clips on the timeline (with only 5 of those mpg clips on the timeline, perhaps it will render, but with 50 clips, it will probably not).

This is the solution that worked for me: if you want to make sure you will not have any problems with editing, then the solution I recommend is to use Vegas to import the clips from a finalized mini dvd, so that you will get those properly synchronized mpg clips, and then open each mpg clip in DGIndex in turn (a .d2v project for every mpg clip, not one .d2v project for all mpg clips), and then use sanlyn's procedure. With this solution, you can avoid rendering problems and audio delay problems.

After editing, and after rendering to something lossless, such as Lagarith or PNG, if you want the video to look good on youtube (good quality), then:
- deinterlace it with QTGMC with preset "slower" and SelectEven() for original frame rate (if input has little blur, you can try QTGMC(Preset="Slower", FPSDivisor=2, ShutterBlur=1, ShutterAngleSrc=30, ShutterAngleOut=180));
- resize to 640x360 with Lanczos;
- encode it with x264 with preset "slow" or higher, and 2-pass at 5000 kbit/s (you can use avisynth and x264 CLI for this, or MeGUI if you don't prefer command line interface);
- use neroAacEnc or MeGUI to encode the audio to AAC-LC at 640 kbit/s CBR;
- mux to .mp4 container and upload mp4 file.