Restoring ambitions using a video processor and Cinema Craft Encoder

[Jeroen1000]

First time poster but have been reading for ... (I lost track of time ). I'll probably have a lot to explain and ask, so do get a cup of coffee
Yesterday until late at night I was reading this thread https://forum.videohelp.com/topic289311.html?sid=ba0cc55731d2a9efea05ddb1d4ded64d

As to date I have got 2 plans:

Plan 1:
First, I am to have my good old VCR cleaned (heads and stuff) connect it to a DV camera and use firewire to get 'it' to my computer. Perhaps apply a few enhancement filters. Finally, CCE would encode the lot to MPEG2. A poster called davideck got (in my opinion) the best results using a DV camera to handle the analogue to digital conversion. I just so happen to be able to borrow one from a friend

Please forgive the perhaps silly questions I've got regarding this:

1. Is there something better to capture in except DV? (Yes, I'm still clueless about what DV entails)*
2. Is WinDV still a good choice of software (as it is no longer being maintained)?
3. I have 2 VCR's (I don't know the make and model, I'll have to check) a DVD recorder (philips VR 175/13) and some S-VHS tapes. Does someone still have the TBC test images mentioned in the thread above? I know how to get those on a DVD but I am unsure how to create the test setup needed. I need see with my own eyes the DV cam's TBC is working. In short, I need to record such an image on a S-VHS tape using VCR 1, and capture the playback of that S-VHS tape using VCR 2. Then I need to be able to inspect the images for TBC errors.
4. In a follow up thread to the one I mentioned, colour images were used ( https://forum.videohelp.com/viewtopic.php?t=291582 ). Again I would like to have them. I'm sure someone can help with the interpretation as they are a lot harder to understand.

*I know you can capture DV with winDV. Does the capture program actually matter? Or are they all equal in quality?


Plan 2:

1. Capture the VCR's output with an DVDO EDGE (which will accompany my Plasma panel)
2. Capture the EDGE's output via a HDMI capture card. The output being.... DV?MPEG2?
3. Use CCE to convert to MPEG2 (I am still checking out what exactly would come out of the EDGE)

The major downside here is that the output will always be progressive. I would chose either 576p or 480p. Any ideas what this deinterlacing will do to quality?


Any help will be much appreciated!

Cheers,

Jeroen

[guns1inger]

Plan 1

If you use a DV camera with pass-through then you will get DV output. From the camera to the PC, using WinDV or anything else, is basically a data transfer operation, albeit without built-in error checking. Therefore, if you choose to go down this path, DV is really the only sensible option at the PC end.

However, after filtering, subsequent encoding needs to be considered. If you are looking at 2-pass (or more) encoding to mpeg-2, then you might also consider encoding your filtered footage using a lossless codec first. This is because if you encode directly to mpeg-2, you will have to run through the filters every pass. A long or complex filter chain could take hours to execute, multiplied by the number of passes. It is usually quicker in these cases to encode to a lossless codec first, then do your mpeg-2 encoding. WinDV is a small, reliable and with low overheads. There is no reason not to use it.

Plan 2

480p encoded for DVD will be fine assuming 23.976 fps. You can then apply 2:3 pulldown and author. If it is not 23.976 fps then you may have to spend a lot more time processing it first.

With 576p you will enc up encoding to 576i, however the field pairs will be from a singe frame image instead of being 1/50th of a second apart.

[Jeroen1000]

Plan 1

If you use a DV camera with pass-through then you will get DV output. From the camera to the PC, using WinDV or anything else, is basically a data transfer operation, albeit without built-in error checking. Therefore, if you choose to go down this path, DV is really the only sensible option at the PC end.

I'm with you so far.

However, after filtering, subsequent encoding needs to be considered. If you are looking at 2-pass (or more) encoding to mpeg-2, then you might also consider encoding your filtered footage using a lossless codec first. This is because if you encode directly to mpeg-2, you will have to run through the filters every pass. A long or complex filter chain could take hours to execute, multiplied by the number of passes. It is usually quicker in these cases to encode to a lossless codec first, then do your mpeg-2 encoding. WinDV is a small, reliable and with low overheads. There is no reason not to use it.

With subsequent encoding you mean encoding the avi-DV file with a lossless video codec?

So: step 1: apply filters to the DV-avi file, for things like noise reduction, sharpness...
step 2: encode the DV-avi file with a lossless codec, because encoding a smaller sized file to MPEG2 will go faster?
step 3: encode to MPEG2 and author the DVD

Plan 2

480p encoded for DVD will be fine assuming 23.976 fps. You can then apply 2:3 pulldown and author. If it is not 23.976 fps then you may have to spend a lot more time processing it first.

it will be PAL VHS. That is what you mean with it not being 23.976fps?

With 576p you will enc up encoding to 576i, however the field pairs will be from a singe frame image instead of being 1/50th of a second apart.

576p would be the most logic choice for PAL would it not? The EDGE will output 576p50. As for your answer to this, could you please elaborate? I know NORMAL interlaced video fields are 1/50 of a second apart. Whilst 2 fields of a progressive frame belong to the same moment in time. But I don't yet understand what you mean with your reply

If I had to guess, you are talking about reinterlacing the 576p video to 576i?

[lordsmurf]

Your best image transfer comes from a S-VHS VCR with built-in filters and TBC and/or frame sync embedded. Oftent his is a single "TBC/DNR" type unit, such as JVC and/or Panasonic equipment. The normal VCR will look like a generic consumer machine, not all that great, not getting everything it can off the tape.

Filtering in the analog domain is best. Once it's made digital, errors often stick, or can only be reduced and not removed.

Your method looks fine. I prefer UNCOMPRESSED AVI over the DV compression If I want to compress, I use 15-20 MB/s MPEG-2 with I-frame only. Compression is comparable. DV capture has it's own downside, such as excessive colorspace compression to 4:1:1 in NTSC (PAL has no such issues).

Look for older posts from GSHELLEY, myself, DAVIDECK, EDDV, etc. We probably all participated and/or comment in the thread you mention.

[Jeroen1000]

Your best image transfer comes from a S-VHS VCR with built-in filters and TBC and/or frame sync embedded. Oftent his is a single "TBC/DNR" type unit, such as JVC and/or Panasonic equipment. The normal VCR will look like a generic consumer machine, not all that great, not getting everything it can off the tape.

Filtering in the analog domain is best. Once it's made digital, errors often stick, or can only be reduced and not removed.

I can certainly agree with that, it all starts at the source. If it produces crap, the end result won't be as good as it can get. However, one could argue where exactly to do the TBC and noise supression (I'm referring to which device in the analogue domain):
If you could take a look at post https://forum.videohelp.com/topic289311-60.html#1450793
Davideck got the best image (it's nice and bright, and sharper than all the others) with his DV cam also acting as a TBC + frame sync device. Question is, does that DV cam apply its filers in the analogue domain (I do suspect it does some filtering and such) or does it apply its filters after the A/D conversion? Either way, it looks like it did a better job than the VCR.

Your method looks fine. I prefer UNCOMPRESSED AVI over the DV compression If I want to compress, I use 15-20 MB/s MPEG-2 with I-frame only. Compression is comparable. DV capture has it's own downside, such as excessive colorspace compression to 4:1:1 in NTSC (PAL has no such issues).

Look for older posts from GSHELLEY, myself, DAVIDECK, EDDV, etc. We probably all participated and/or comment in the thread you mention.

I thought DV was a lossless compression? Judging from your anwser it probably isn't (I'm reading up on it right now so this comment may look stupid afterwards )
One small thing though, how does one get uncompressed avi from a VCR to a PC?
I'm not going to go into the colorspace thing yet as I barely know what it is for the moment. Even if it does not matter for PAL (which is good actually) I am too curious to pass on such interesting topics.
Being I'm converting PAL the DV downside is basically a non issue for me.

So I gather plan 2 is likely to yield a lesser quality than plan 1. I'm trying to find out what the EDGE does with its analogue feed. IF it applies filters in both the analogue and digital domain, it may very well produce the best image as it also acts like a TBC and frame sync. Albeit, it will be the most expensive, read advanced, (tv excluded) piece of equipment I own

[guns1inger]

DV compresses at a ratio of around 5 : 1 and is a constant bitrate compression. You get uncompressed AVI (or better yet, losslessly compressed) by using a capture card that takes an analogue input, and a machine with fast enough CPU and HDDs to handle the data as it comes in, gets compressed, and gets stored.

[2Bdecided]

So: step 1: apply filters to the DV-avi file, for things like noise reduction, sharpness...
step 2: encode the DV-avi file with a lossless codec, because encoding a smaller sized file to MPEG2 will go faster?
step 3: encode to MPEG2 and author the DVD

No, a single step is DV > filter > lossless.

There's no point doing DV > filter > DV and then DV > lossless! You just get a larger file, but no gain in quality.


NR and sharpness processing is better quality in the digital domain than the analogue domain (IMO lordsmurf is wrong in this respect - and his favourite VCRs are doing NR in the digital domain anyway!), but it can be much slower than real time to do this on a PC.

However, you must get the best signal from tape, and get it properly digitised. A TBC somewhere in the chain is near-essential - either in the VCR, or the capturing device.

As others will tell you, there are some decent DVD recorders available that make messing around with PC software unnecessary for most.

Cheers,
David.

[Jeroen1000]

Hi David,

Thanks for the clear DV explanation, seems like a relatively doable solution . The more I learn the more attractive plan 2 starts to sound also.

We can all agree one this point: the VCR has to be A++. The less imperfections noise algorithms need to clean up the better.
The discussion of which VCR is best, would lead us to far.

However, I do have some remarks (I'm not a native speaker so if something I type comes out wrong do not take it as offensive please!).
Firstly, why would a VCR do NR in the digital domain? It would have to do a A/D conversion, run its NR, and then do a D/A as it has to output an analogue signal?

Secondly, a TBC in a VCR is not really a necessity as both a DV-cam and a videoprocessor can take on that role (including providing a continuous sync). Davideck has demonstrated this beyond doubt. Moreover, the technology in those devices is more advanced so it is not inconceivable they do a on par or even better job.
We also seem to agree on this so the above can safely be ignored

The 'major' downpoints concerning DV-cam is :

1) If they work as a pass through do they do any processing (sharpness, NR) except TBC?
2) When do they do this? When the signal is still analogue or after is has digitalised it? In other words: TBC errors LOOK like they are corrected in the analogue domain. Therefore, I am assuming NR is done in either the VCR or DV cam (or both) and it (or they) do it in the analogue domain.
3) However, I am firmly against any NR in both the VCR or the DV-cam as I think there are better filters available which can be run on computers or good videoprocessors. For the very same reasons I am against on the fly MPEG encoding as CCE will surely do a better job.

Take notice about one important fact: If lordsmurf is correct, and errors are best dealt with in the analogue domain, my argument against processing in the analogue domain is utterly useless. If however, the digital domain is best at solving issues, my argument makes sense.

My other route with the EDGE (which for me is already an expensive device) is a bit more puzzling to me:

1) does the EDGE do its processing in the digital or analogue domain? It would be safe to say the EDGE would do a good job either way (read better than most VCR's), but it would be unfortunate if it does not do this in the domain best suited for the job.
2) The signal coming out of this device is 576p and must be transported via HDMI. Therefore an HDMI capture card is needed. What comes out of this card? Uncompressed? MPEG compressed? I haven't an idea yet. Does uncompressed mean codecless video? In other words, all one's and zero's captured into a file? Just one bit after the other?
3) Do I need to reinterlace 576p->576i and
4)What will the conversion from 576i to 576p have done to the quality (even more so if it needs to be reinterlaced again.

The above could be summarised a bit like: Given a relatively good source, the main protagonist is the A/D device, which has to correct errors in the analogue domain BEFORE converting it to the digital (again, unless lordsmurf is wrong). Thus, the better this A/D device is, the better quality you get. Minor errors can be fixed in the digital domain applying filters. Given that knowledge the EDGE (I'm starting to sound like a salesman ) has a real edge . Only issue remaining is whether making the video progressive is a killer or not.

I hope you're in the mood for a discussion, as I am really looking forward to your reply!

[2Bdecided]

Firstly, note that there are plenty of threads about this general topic.


My understanding is that a TBC is a digital device. It stores and re-clocks data - it's not exactly easy to do this without sampling (e.g. digitising) it. The digital NR can work with the same buffer as the TBC, adaptively averaging lines and/or frames to remove noise where appropriate. In software there'd be motion compensation too before adaptively averaging subsequent frames, but I doubt VCR DNR does this!

Any VCR NR that doesn't even store a line of video is just filtering the signal - it's a glorified sharpness control, and nothing more. Almost all VCRs have this, and IMO it's pretty worthless.


I'm not quite sure what the edge thing is gaining you. It's deinterlacing when you don't want it to, it's giving you HDMI when you want DV or lossless straight into the PC (i.e. you'll need an expensive HDMI capture card), and it might not handle VHS that well at all (it seems aimed at digital source like DVD and DVB, but maybe there are features aimed at improving antique analogue too).


Finally, in one old thread, Davideck showed a DV camcorder with an analogue input that had an excellent TBC. That doesn't necessarily mean that's the best way to digitise video in 2008. Many people would suggest that a DVD recorder (for those who don't want to process digitally), or a capture card > lossless (for those who do want to process digitally) would be better solutions.

Cheers,
David.