Authoring AVCHD from DV AVI (am I insane?!) or stick with DVD?

[dvd3500]

Dear All,
After being away from digital video for nearly 10 years I am trying to get back up-to-speed.

I have spent several days (OK hours) searching this forum and others and I haven't found my exact use-case so I figure I would ask...

Situation:
I dug out my old Hi8 (PAL) camcorder and decided to start archiving since those tapes are now pushing 15 years old and might deteriorate.

I used an ADVC 300 to create (PAL) DV AVIs.

As you all know, DV AVI is Bottom (lower) field first.

I did a quick test with AVStoDVD and the quality of the resulting DVD was noticeably worse than the DV AVIs.

Since I had made multiple AVCHDs from Hi Def TV shows I thought to myself: These clips are on average only 20-40 minutes long, why not make an AVCHD out of them!

My goal it convert/edit/process the files as little as possible. The DV AVI output from the ADVC 300 is so good that post-processing is counter productive IMHO:

According the specification 720×576 interlaced video is supported:

https://en.wikipedia.org/wiki/AVCHD#Specifications

The only thing bugging my is the field order. From what I read AVCHD is usually Top (upper) Field First.

Questions:
If the field order IS a problem, how do I go about getting it changed? AVISynth? VirtualDub?

If field order is not an issue, what is a good workflow to get AVCHDs? (In the past I had to use about 5 different apps to get it a working AVCHD from Hi Def TV shows.) I was thinking of using AVCHDCoder since MultiAVCHD is so out of date... Very few of them specifically state that handle AVIs.

OR(!) Am I just crazy and I need to tweak the DVD settings cranking up bit rate and setting it VBR?

Again, I think I have been reading to too much about it. ...

Thank you all in advance!

[pippas]

Like you, I digitise my analogue footage with a Canopus converter (ADVC50 in my case). That gives me PAL DV.avi files which, as you say, are interlaced with lower field first.
I keep my final files in the .mp4 format. Same coding (H.264) as used in AVCHD files, but I find the mp4 format more compatible with my various replay options (computer - WD Live media player - PVR USB input)
I have no interest in keeping my files as interlaced, or in 'burning' any discs and use Virtualdub to modify and convert my DV files in one procedure...(apart from the final quick conversion to mp4 at the end)

- I de-interlace uisng the Yadif filter set to 'double frame rate - bottom field first'.
That allows the files to retain the motion 'smoothness' that comes with interlaced footage, which, in the case of PAL, creates 50 images per second (fields rather than frames, but 50 images nevertheless)

- I then resize to allow the final file to have square pixels, and not be reliant on aspect ratio 'flags' in the file headers to display correctly in all replay situations.
For 4:3 footage I resize to 768x576, and for 16:9 I use 1024x576 (The PAL DV files from the Canopus converter will always be 720x576)

- I then crop the borders which are created when capturing VHS tapes. (That may not be necessary with Hi8 files??)

- I then select the x.264vfw codec and export from Virtualdub as a H.264 .avi file (Virtualdub doesn't export as mp4 without involving quite a lot of extra 'tricks')

- You can then convert to mp4 quite simply (without recoding the video) by using the free MKVtoMP4 program

Quite small final files (compared to DV!) and no apparent quality loss.

Just one way of approaching things of course.....

[usually_quiet]

Dear All,
After being away from digital video for nearly 10 years I am trying to get back up-to-speed.

I have spent several days (OK hours) searching this forum and others and I haven't found my exact use-case so I figure I would ask...

Situation:
I dug out my old Hi8 (PAL) camcorder and decided to start archiving since those tapes are now pushing 15 years old and might deteriorate.

I used an ADVC 300 to create (PAL) DV AVIs.

As you all know, DV AVI is Bottom (lower) field first.

I did a quick test with AVStoDVD and the quality of the resulting DVD was noticeably worse than the DV AVIs.

Since I had made multiple AVCHDs from Hi Def TV shows I thought to myself: These clips are on average only 20-40 minutes long, why not make an AVCHD out of them!

My goal it convert/edit/process the files as little as possible. The DV AVI output from the ADVC 300 is so good that post-processing is counter productive IMHO:

According the specification 720×576 interlaced video is supported:

https://en.wikipedia.org/wiki/AVCHD#Specifications

The only thing bugging my is the field order. From what I read AVCHD is usually Top (upper) Field First.

Questions:
If the field order IS a problem, how do I go about getting it changed? AVISynth? VirtualDub?

If field order is not an issue, what is a good workflow to get AVCHDs? (In the past I had to use about 5 different apps to get it a working AVCHD from Hi Def TV shows.) I was thinking of using AVCHDCoder since MultiAVCHD is so out of date... Very few of them specifically state that handle AVIs.

OR(!) Am I just crazy and I need to tweak the DVD settings cranking up bit rate and setting it VBR?

Again, I think I have been reading to too much about it. ...

Thank you all in advance!

If you are using AVStoDVD, two-pass VBR encoding with HCEnc would allocate the bitrate more efficiently and might look a little better. Cranking up the bitrate a little would also help, but as you know DVD sets limits the highest bitrate allowed. Putting at most one hour of video on a DVD 5 would be best if you are trying to maximize quality. You could try it and see what you think.

[_Al_]

Putting at most one hour of video on a DVD 5 would be best if you are trying to maximize quality.

In that case you can encode with Constant Quantization, setting max to 8000 for example, and it is going to be just 1pass. 1hour 10 minutes should be safe to encode like that.
I use Quant. value 3 (CQ_MAXBITRATE), 8000 max bitrate, rest is default.

[usually_quiet]

Putting at most one hour of video on a DVD 5 would be best if you are trying to maximize quality.

In that case you can encode with Constant Quantization, setting max to 8000 for example, and it is going to be just 1pass. 1hour 10 minutes should be safe to encode like that.
I use Quant. value 3 (CQ_MAXBITRATE), 8000 max bitrate, rest is default.

AVStoDVD does not allow access to all HCEnc's settings. It only has a very simplified advanced settings page for HCEnc, which includes an AQ (Adaptive quatization) setting but not constant quantization. To follow this advice, the OP will need to manage the encoding himself.

[dvd3500]

Thanks all!
I am not afraid of using multiple tools and I have mucked around with quantization in the past. I did want to try to avoid de/multiplexing since I am afraid of audio/video desynch (one of the reasons I got the ADVC 300 is because its synch is dead on all the time).

@Pippas: I may look into de-interlacing as you noted. For years people were screaming "DON'T DEINTERLACE" but maybe those days are past... I still prefer discs partially because I am old-fashioned, but also because I share them with my mother who is not technical savvy and has a tendency to delete things/get viruses.

[bulgom]

That allows the files to retain the motion 'smoothness' that comes with interlaced footage,

Hi,

interesting (above).

But below, I wonder if I understand right:

- I de-interlace using the Yadif filter set to 'double frame rate - bottom field first'. (...) which, in the case of PAL, creates 50 images per second (fields rather than frames, but 50 images nevertheless)

— So your video is converted to 50 images/second. Is your FINAL result going to STAY at that frame rate? I guess yes, but still wonder. Because, unless it will never have unexpected effects, for instance with some players, may be complaining or refusing to play (?), in general, I stick to the original framerate (PAL 25, NTSC 29.97).

If, according to you, there no real or "bad" need to stick to the most standard framerates, could you confirm?


— (I know what fields are) If I understand what I had read already + remember right, "YADIF" creates or extracts two images, out of each original full frame (though I also wonder if there is ever a state of full frame, with interlaced video...), :

one is the upper field, the other one is the lower field. But of what size (if PAL std): 720 × 288 pixels? Or 720 × 576 px.?

In the 1st case, what will the actual image size of the final result be?

In the 2nd case, are the missing lines of each image replaced by or filled with something (such as interpolation, may be)?


Reason why I'm asking: I would like to deinterlace quite a few videos but, with other deinterlacers than "YADIF", I'm never satisfied: I always find the result more blurry than the original film... I liked "the "YADIF" result but, with that one, if set to double framerate, I of course wonder about the compatibility of the "final product" with all players...


.

[_Al_]

25i transfered to 50p doubles the frame rate, those fields (half frames) are just transformed into full frames.

I found Yadif not good enough though. Edges are not smooth etc., remember today we watch those images on full HD screens, where everything gets blown up. The only double frame rate conversion quite usable is done by QTGMC (there might be some other softwares around, perhaps not free, not sure).

QTGMC is used together with Avisynth so one might resize to square pixel right away making 768x576 50p for example, together with cropping etc. getting some other size. But that is not necessary, Video could have 720x576 and 8:9 sar flag could be given to it while encoding H264 stream. This whole thing could be done in many ways, and could be topic on its own, as it already was discussed on this forum already.

I'd say if not using QTGMC better to keep it interlaced and make a DVD, just an opinion.