VHS capturing without VirtualDub under Linux

[pulga]

Hi all,

I am trying to digitalise old family VHS (PAL) for three days now and it has been mainly frustrating with very slow progress. Since I haven't found an especially helpful guide that fits for my prerequisities so far, I would like to use this thread to sum up what I might do and kindly ask for help and comments about that.

Initially, I have one of these cheap video grabbers here (I don't know the chipset, there's no description displayed in lsusb or something like that), I believe it's a utv007 though. The bad thing is that the drivers for Windows aren't working, so I cannot capture with VirtualDub. I tried it on several different computers but at least the Linux driver works. I know that this stick is far from being professional, but I would like to get close to the best result achievable with this hardware.

My plan right now is to grab the video somehow lossless for archiving and then do filtering, deinterlacing and stuff like that in VirtualDub2 or Handbrake (?) and convert to a reasonable format/codec/bitrate. So for capturing my command is:

ffmpeg \
-f v4l2 -standard PAL -thread_queue_size 2048 -video_size 720x576 -i /dev/video4 \
-f alsa -thread_queue_size 2048  -i hw:1,0 \
-vcodec libx264 -crf 0 -pix_fmt yuv420p -preset ultrafast \
-acodec flac -channel_layout stereo -report out.mkv

My questions currently are:

1. Is x264 with crf=0 appropriate or should I better use huffyuv or some other codec instead? I have problems playing the videos in VLC and QtAV when using x264, huffyuv works better but is 50% bigger in size.
2. Should I use YUV 4:2:0 (like in the example) or 4:2:2?
3. FLAC or PCM? And what would be the syntax in ffmpeg if PCM is better?
4. With this code I get an aspect ratio of 5:4 for whatever reason. Should I set the ratio to 4:3?
5. Would it be better to do everything at once i.e. grabbing, encoding, deinterlacing, denoising etc.? There are at least options for deinterlacing with ffmpeg, although I don't know anything about them.
6. Is there anything I miss/should change or are there "better" ways than using ffmpeg?

I would be very grateful if someone could have a look at this since I am not very familiar with anything in this topic.

Edit: If capturing doesn't work for you in Windows as well, you may want to check if Desktop-Apps have access to camera and microphone in the Windows privacy settings.

[abolibibelot]

1) Lagarith might be a better compromise than HuffYUV with regards to size, and cause less trouble with media players. Or MagicYUV (slightly less efficient size-wise than Lagarith in my experience). Or FFV1 — but to quote “poisondeathray” from this thread :

Performance wise, UT Video (fastest scrubbing in editors). Compression wise, FFV1 (but slow as molasses).

3) If FLAC works flawlessly with your workflow, by all means use FLAC. But the space saving is going to be negligible compared with the huge sizes of lossless video streams. Some NLE editors might be unable to load FLAC streams.

4) Because by default, display aspect ratio (DAR) = storage aspect ratio (SAR), and 720 : 576 = 5 : 4. So yes it has to be specified that it's an “anamorphic encoding” (non-square pixels). But if you don't plan on putting the converted footage on standard video DVD (for which anamorphic encoding is mandatory as per the specifications of the standard), you can also capture in 4:3 SAR at 768x576, or 640x480 (most experts of VHS digital conversion would say that 640x480 is already more than enough to retain the effective resolution of the analogic source in the best possible conditions, so setting a higher value is a waste of bitrate, or results in a loss of quality if the bitrate is constrained).

My plan right now is to grab the video somehow lossless for archiving and then do filtering, deinterlacing and stuff like that in VirtualDub2 or Handbrake (?)

Handbrake is good as a simple, mostly automated all-in-one video compression software, but doesn't give much control, and is limited if any kind of pre-processing is necessary. In particular, it can't work with Avisynth scripts, so doesn't allow to use superior deinterlacing filters like QTGMC.

Initially, I have one of these cheap video grabbers here (I don't know the chipset, there's no description displayed in lsusb or something like that), I believe it's a utv007 though. The bad thing is that the drivers for Windows aren't working, so I cannot capture with VirtualDub. I tried it on several different computers but at least the Linux driver works. I know that this stick is far from being professional, but I would like to get close to the best result achievable with this hardware.

I have read many discussions on the topic (and still haven't converted a single one of my own tapes ! é_è) : as for cheap capture devices, old ATI All-In-Wonder cards are highly recommended. They can be found on the second-hand market for 10-20 dollars / euros, or with a bit of luck you can even find a complete computer with one of them inside for little more.
But doing the best you can do with what you have right now may indeed be the wisest course of action (wiser than doing nothing and letting years go by, with each one increasing the odds of deterioration of those unique tapes, which is probably a more important factor than the respective quality of the capture device, I may learn this the hard way one day if I finally manage to get on with that task !...).

[pulga]

Alright, I was hoping to get some more answers in the meantime but many thanks to you anyway.

Regarding ATI-cards: I have a laptop only, so I cannot use anything different than USB for this case.

Lagarith might be a better compromise than HuffYUV

What's actually wrong with lossless x264? It's apparently produces way smaller filesize and I guess x264 is more common in general than HuffYUV, Lagarith and all others so it should be more future-proof.

A main point for me right now is chroma subsampling. I don't get if I should use 4:2:2 or 4:2:0 here.

[lordsmurf]

1) Lagarith might be a better compromise than HuffYUV with regards to size, and cause less trouble with media players.

That makes no sense. No media player works well, or at all, with lossless codecs. The problem with Lagarith is it had more resource overhead for capturing. Great intermediary codec, just not often for capturing. Too much resource overhead = dropped frames.

Is x264 with crf=0 appropriate or should I better use huffyuv or some other codec instead? I have problems playing the videos in VLC and QtAV when using x264, huffyuv works better but is 50% bigger in size.

Never x264. Use lossless, sometimes MPEG. But never H.264(x264).

Should I use YUV 4:2:0 (like in the example) or 4:2:2?

4:2:2 is original color. 4:2:0 is some loss. Not as bad as 4:1:1, but still mild loss. This is what DVD/BD uses. Generally you only go to 4:2:0 after editing/etc, the delivery file.

FLAC or PCM? And what would be the syntax in ffmpeg if PCM is better?

FLAC is overhead. Again, overhead is bad for capturing.

With this code I get an aspect ratio of 5:4 for whatever reason. Should I set the ratio to 4:3?

5:4 is PAL SAR (720x576), but it's rectangle pixels. The DAR is still 4:3.

Would it be better to do everything at once i.e. grabbing, encoding, deinterlacing, denoising etc.? There are at least options for deinterlacing with ffmpeg, although I don't know anything about them.

No, never.

Is there anything I miss/should change or are there "better" ways than using ffmpeg?

Linux is really the wrong tool for capturing. Capturing is a Window world. Linux/Mac is just crutch methods, most of them quite terrible.

[pulga]

Never x264.

Could you specify why?

FLAC is overhead. Again, overhead is bad for capturing.

So better use PCM or AAC or something different?

Linux is really the wrong tool for capturing.

So what would you do in my place, buying another grabber and hoping that the software recognises the driver this time?

[pulga]

After hours of pointless debugging I just found out what the problem is with Windows. Sometimes Google is useful too . The device is of course recognised as a camera, so access of Desktop-apps to camera and microphone has to be activated in the privacy preferences of the OS. I don't know why I did not realise that

I don't get a visual preview though, only audio, but recording now works apparently. Is there an especially recommendable guide how to do it best with VirtualDub2 or AmaRec (or any other freeware)?

[abolibibelot]

@pulga

Alright, I was hoping to get some more answers in the meantime but many thanks to you anyway.

You're welcome. I was hoping too! But mid August is a slow period for pretty much everything except herpes and melanoma.

Regarding ATI-cards: I have a laptop only, so I cannot use anything different than USB for this case.

There are USB devices based on ATI chips : ATI TV WONDER 600 / 650 – but they may be hard to find.
See this :
http://www.digitalfaq.com/forum/video-capture/3200-best-ati-wonder.html

@lordsmurf

That makes no sense. No media player works well, or at all, with lossless codecs. The problem with Lagarith is it had more resource overhead for capturing. Great intermediary codec, just not often for capturing. Too much resource overhead = dropped frames.

That's a bit harsh... but you're the expert !
Regarding “too much resource overhead = dropped frames”, is it still the case, with Lagarith (or similar codecs which are more efficient than HuffYUV but more CPU-intensive), considering that the computational power of even a low-end current machine is vastly superior to that of a high-end machine from the Windows XP era ?

Never x264. Use lossless, sometimes MPEG. But never H.264(x264).

Well, x264 with -crf 0 is supposed to be lossless... But arguably not suitable as a capturing codec for the reasons you stated above.
But it would be wise (I think) to keep a x264 -crf 0 + FLAC lossless version as an archive, once the lossy encode intended for viewing / sharing has been made.
Or perhaps FFV1 + FLAC – does anyone know how FFV1 and x264 -crf 0 compare in terms of compression efficiency ?

FLAC is overhead. Again, overhead is bad for capturing.

But quite light overhead, it would seem to me, with a current CPU, and it reduces the data throughput which may be more of a bottleneck on a laptop computer with a relatively slow HDD. (But that's probably irrelevant since the video throughput is much larger anyway.)

@pulga

So better use PCM or AAC or something different?

Obviously not AAC, it's a lossy codec, meaning that you get one “generation loss” right away, if you have to edit anything (if only to normalize), you'll have to re-compress. Ideally you should do a lossy compression only once, and if possible keep a lossless compression version if you ever need to do some further filtering or export another lossy version into a different format or with a different resolution.

[johnmeyer]

Never x264. Use lossless, sometimes MPEG. But never H.264(x264).

You tell him what NOT to use, but don't make any recommendations (other than "sometimes MPEG") as to what he should use.

What video codec do you recommend (current day, not the old posts on DigitalFAQ) for capturing VHS video?

[october262]

Hi all,

I am trying to digitalise old family VHS (PAL) for three days now and it has been mainly frustrating with very slow progress. Since I haven't found an especially helpful guide that fits for my prerequisities so far, I would like to use this thread to sum up what I might do and kindly ask for help and comments about that.

Initially, I have one of these cheap video grabbers here (I don't know the chipset, there's no description displayed in lsusb or something like that), I believe it's a utv007 though. The bad thing is that the drivers for Windows aren't working, so I cannot capture with VirtualDub. I tried it on several different computers but at least the Linux driver works. I know that this stick is far from being professional, but I would like to get close to the best result achievable with this hardware.

My plan right now is to grab the video somehow lossless for archiving and then do filtering, deinterlacing and stuff like that in VirtualDub2 or Handbrake (?) and convert to a reasonable format/codec/bitrate. So for capturing my command is:

ffmpeg \
-f v4l2 -standard PAL -thread_queue_size 2048 -video_size 720x576 -i /dev/video4 \
-f alsa -thread_queue_size 2048  -i hw:1,0 \
-vcodec libx264 -crf 0 -pix_fmt yuv420p -preset ultrafast \
-acodec flac -channel_layout stereo -report out.mkv

My questions currently are:

1. Is x264 with crf=0 appropriate or should I better use huffyuv or some other codec instead? I have problems playing the videos in VLC and QtAV when using x264, huffyuv works better but is 50% bigger in size.
2. Should I use YUV 4:2:0 (like in the example) or 4:2:2?
3. FLAC or PCM? And what would be the syntax in ffmpeg if PCM is better?
4. With this code I get an aspect ratio of 5:4 for whatever reason. Should I set the ratio to 4:3?
5. Would it be better to do everything at once i.e. grabbing, encoding, deinterlacing, denoising etc.? There are at least options for deinterlacing with ffmpeg, although I don't know anything about them.
6. Is there anything I miss/should change or are there "better" ways than using ffmpeg?

I would be very grateful if someone could have a look at this since I am not very familiar with anything in this topic.

Edit: If capturing doesn't work for you in Windows as well, you may want to check if Desktop-Apps have access to camera and microphone in the Windows privacy settings.

since your running LINUX, try installing WINE and then you can run windows programs on LINUX.

[lordsmurf]

Never x264. Use lossless, sometimes MPEG. But never H.264(x264).

You tell him what NOT to use, but don't make any recommendations (other than "sometimes MPEG") as to what he should use.

What video codec do you recommend (current day, not the old posts on DigitalFAQ) for capturing VHS video?

He was already discussing it: lossless codecs. Huffy is least overhead lossless.
Sometimes MPEG.

since your running LINUX, try installing WINE and then you can run windows programs on LINUX.

Capture hardware cannot run in VMs. In context of hardware abstraction, Wine is a VM.

[abolibibelot]

since your running LINUX, try installing WINE and then you can run windows programs on LINUX.

https://www.cjoint.com/c/IHoe7sMqqiA
https://www.cjoint.com/c/IHofqKbUnBA
https://www.cjoint.com/c/IHofsExRwvA

[pulga]

since your running LINUX, try installing WINE and then you can run windows programs on LINUX.

I already tried Wine of course and can confirm, that it doesn't work.

He was already discussing it: lossless codecs. Huffy is least overhead lossless.
Sometimes MPEG.

So x264 with crf=0 is not lossless?


I have another question: I have a DVD/VCR Recorder (LG RC388) here which obviously could do capturing "on its own". Currently I'm using my USB grabber connected to this recorder which just plays the VHS cassettes. My question now is, if it is adviseable to let the recorder do anything, without the grabbing stick. The downside of that would be, that I would have to transfer the videos from DVD to HDD and that multiple DVDs may be needed for a single cassette which is expensive, time and resource consuming. But if the quality would be better than I can manage to do with the USB stick I may deal with that.

[amaipaipai]

I'm not a big fan of linux for capturing, but I use Shotcut with MJPEG 4:2:2+PCM.

[KarMa]

Is x264 with crf=0 appropriate or should I better use huffyuv or some other codec instead? I have problems playing the videos in VLC and QtAV when using x264, huffyuv works better but is 50% bigger in size.

Never x264. Use lossless, sometimes MPEG. But never H.264(x264).

I use x264vfw without issue when recording VHS tapes. As long as you record in interlaced mode, save it as 4:2:2, and use a x264 preset that doesn't ever max out your CPU (fast) then it will be fine. Though I don't use x264 for lossless capture but instead use it for lossy captures. For lossless x264 doesn't save much so I usually just go with Lagarith.

That makes no sense. No media player works well, or at all, with lossless codecs. The problem with Lagarith is it had more resource overhead for capturing. Great intermediary codec, just not often for capturing. Too much resource overhead = dropped frames.

Unless you are on some 10 year old CPU, this shouldn't be an issue.

[abolibibelot]

So x264 with crf=0 is not lossless?

I've been told that it was lossless (although not properly decoded by most NLE softwares), but I'll let someone who knows for sure chime in and provide some technical insight.

I have another question: I have a DVD/VCR Recorder (LG RC388) here which obviously could do capturing "on its own". Currently I'm using my USB grabber connected to this recorder which just plays the VHS cassettes. My question now is, if it is adviseable to let the recorder do anything, without the grabbing stick. The downside of that would be, that I would have to transfer the videos from DVD to HDD and that multiple DVDs may be needed for a single cassette which is expensive, time and resource consuming. But if the quality would be better than I can manage to do with the USB stick I may deal with that.

Hard to say without having first-hand experience with both devices... But, generally speaking, from what I could gather over the years while staying practically none-the-wiser :
– You could use a bunch of DVD-RW to reduce the cost and bulk.
– The advantage is that this method is relatively “unattended” : you put the tape, press record and can come back later, if the recorder is in good working order the result should be reliable and readily watchable. With the computer + standalone capturing device method, you get more control but also more hassle to make sure that everything works together as it should, that every driver is accounted for, that there are no dropped frames or synchronization issues, and so on... then more headaches to post-process the output... Potentially, the quality can be significantly improved, by capturing to a lossless format with a good capturing device and doing some wise filtering before compressing to a delivery format. But practically, with the tools and skills you currently have, that's not even guaranteed, or there's a high risk or giving up before completion considering the amount of learning and tweaking that it entails (see the quotation below).
– Among DVD recorders, the ones which get highly recommended are the ones based on a “LSI chipset”, such as the JVC DR-M100 (I got one used for 30 bucks years ago but haven't used it at all beyond a few tests – and that's when I was motivated to complete the task é_è).
http://www.digitalfaq.com/forum/video-workflows/4073-video-hardware-dvd.html
– Quoting “johnmeyer” from this recent thread :

Very few people, especially those who are new to video, ever finish these project which is why, despite the "best quality possible" requirement, I usually recommend the same thing as manono: a VHS to DVD combo. The advantage is that it works, pretty much out of the box, and that you just feed it tapes and blank discs and then have it crank.
The worst experience I had was around fifteen years ago where a guy would call every year around Thanksgiving, wanting to know how to transfer his tapes. He followed the advice I gave him the first time we talked and had indeed purchased the required hardware. I even went to his house one time to actually transfer a tape. It worked just fine, and exactly like it was supposed to but, because of all the "moving parts" of having a computer, VHS deck, capture equipment, capture software, editing software, and authoring software (I'm probably forgetting a few things) he could never make it to the finish line, even for one tape.
He did finally get the combo and AFAIK he did actually finish the project using that.

– But the VCR part of all-in-one VHS-to-DVD recorders is generally considered substandard... So it's tough to take a decision with so many seemingly contradictory statements, with so many moving parts, with a limited budget, a limited amount of time to put into this, and little opportunity to actually run some first-hand tests and decide based on the outcome (and also a limited experience to even determine what makes one outcome superior to the other, although said other may be more palatable at first glance, thanks -- for instance -- to an agressive noise reduction processing which effectively removes unwanted fluctuations in the picture but also washes out all fine detail, apparently Panasonic recorders do that a lot and are not recommended for that reason, too bad the first thing I purchased when I decided to tackle the digitization of my tapes was a Panasonic recorder...).

[pulga]

I came across the posting of "johnmeyer" as well and, to be honest, I'm a bit surprised, because there are devices for 10-20€ (like I have) that come with software for capturing, so you basically just need to plug connect them to VCR and PC and press play. I understand the "best quality possible" stuff but bad quality in this case is better than nothing. So to sum up, the risk of throwing everything away is not significant in my case. But I would lie if I say that it doesn't start to annoy me a little bit, since it does take a lot longer and gets a lot more complex than I imagined.

I found a JVC HR-S9850 for 50€ in my surrounding, which is listed in lordsmurfs VCR buying guide. Maybe this in combination with the Hauppauge WinTV USB Live2 would be a good (budget) idea?

[pulga]

Another question: I'm playing around with VirtualDub2 now and noticed, that it obviously records with 30 fps. This makes no sense to me, since my source is obviously PAL and all other recordings I did so far were done with 25 fps. I cannot change the framerate in the capturing settings either. But I can't encounter anything strange when watching the recordings of VirtualDub, liveview works as well, what's the point here?

I also cannot change the volume like mentioned in this guide here, the Volume Meter just displays the current volume, no possibility to change something here. But I think it's not such a big deal.

Edit: If I set the device to "Microsoft WDM Image Capture" I can change the framerate and "format" is not greyed out like when selecting "DirectView". But I still cannot change the volume.

[johnmeyer]

There is no "hard-wired" 30 fps with VirtualDub. Keep looking through the capture settings for your hardware. That is where the fps will be set.

[lordsmurf]

"Bad quality is better than nothing" is arguable.
Sometimes quality sucks so much that you get a headache trying to watch it.
Literally.

H.264 (including x264/ffmpeg version) is not lossless, period.
- Less-lossy, sure.
- Lossless, no.

[KarMa]

"Bad quality is better than nothing" is arguable.
Sometimes quality sucks so much that you get a headache trying to watch it.
Literally.

H.264 (including x264/ffmpeg version) is not lossless, period.
- Less-lossy, sure.
- Lossless, no.

Firstly we are talking about VHS here with VHS quality. You said MPEG (2?) can be used, which is less efficient than x264. And you even offer H.264 transfers for your customers.

x264 does support lossless, period. I don't use it for that but it does support it. There's some quibbling over if you need to use CRF=0 or CQ=0 to get lossless, I forget which or maybe both give lossless. But it does support it.

https://forum.videohelp.com/threads/358901-is-x264lossless-really-lossless

[jagabo]

There's some quibbling over if you need to use CRF=0 or CQ=0 to get lossless, I forget which or maybe both give lossless.

Both work. If you specify CRF=0 and use Mediainfo to check the resulting file you'll find that x264 used QP=0.

[johnmeyer]

I wouldn't get too hung up on whether a codec is "lossless" or whether it is "lossy". The real issue is how it looks, and you have to do your own tests, with your own material, to know about that.

Instead, I would concentrate on whether a codec is designed as an intermediate codec or whether it is designed for final delivery.

The best example of what I am talking about is Cineform. It was created almost a quarter century ago as an "intermediate codec." The term "intermediate" refers to the fact that it is designed for editing, which is the process between acquisition (taking the video with a digital video camera, or capturing from an analog video source) and final delivery. During those intermediate steps, the video is often manipulated and stored several times. While ideally you want to edit and then save to the final, delivery format all in one step, there are hundreds of situations where this is not possible, and you have to edit, then save the result, and then use that result for further editing, and subsequent re-compression.

The beauty of Cineform is that, while it IS a LOSSY codec, you can use it multiple times without seeing any noticeable degradation. This means you can have all the advantages of a compressed format (easy to read from disc, so no dropped frames or stuttering playback; and less disk space used), without having to worry about degrading your masterpiece.

So my advice is to look for codecs which were designed as intermediate codecs, such as ProRes, DNxHD, Cineform, or DPX.

[lordsmurf]

x264 does support lossless ... I don't use it for that

No.
Not for capture.

This is why:

concentrate on whether a codec is designed as an intermediate codec or whether it is designed for final delivery.

H.264 is delivery, extreme CPU overhead for real-time (appliance required).
Intermediaries are want you want. Some capture better tan others, again because of overhead. Huffyuv is safest. Lagarith works. ProRes, maybe. A few others get more iffy, problems, Ut Video, MagicYUV, etc. I really have no opinion on Cineform or DNxHD, never tried those for capture, just NLE work.

You said MPEG (2?) can be used, which is less efficient than x264.

MPEG is a wide spec format, supporting SD bitrates of 50mbps, 4:2:2, and GOP can be altered as needed. It's not worse quality than H.264, not at all. And best yet, it's a capture format, and editing format, and high-end professionally, something that is NOT true of H.264 without special hardware. Also a distribution format, what it's mainly known for (and why people are confused about MPEG abilities).

And you even offer H.264 transfers for your customers.

Never captures.
Captures are almost always lossless AVI, with a few MPEG where appropriate.

Converted copies are H.264 (almost never x264, rarely encoded with freeware), often deinterlaced for LAN/streaming, in addition to higher quality archival.

[jagabo]

H.264 is delivery, extreme CPU overhead for real-time (appliance required).

In my experience x264vfw at the ultrafast preset, two threads, QP 0 (lossless), all I-frame, interlaced, YUV 4:2:2 (input as well as internally), compresses just as fast as huffyuv v2.1.1 and achieves slightly more compression.

[october262]

since your running LINUX, try installing WINE and then you can run windows programs on LINUX.

I already tried Wine of course and can confirm, that it doesn't work.

He was already discussing it: lossless codecs. Huffy is least overhead lossless.
Sometimes MPEG.

So x264 with crf=0 is not lossless?


I have another question: I have a DVD/VCR Recorder (LG RC388) here which obviously could do capturing "on its own". Currently I'm using my USB grabber connected to this recorder which just plays the VHS cassettes. My question now is, if it is adviseable to let the recorder do anything, without the grabbing stick. The downside of that would be, that I would have to transfer the videos from DVD to HDD and that multiple DVDs may be needed for a single cassette which is expensive, time and resource consuming. But if the quality would be better than I can manage to do with the USB stick I may deal with that.

try checking out some of the youtube videos on VHS capturing using LINUX - https://www.youtube.com/results?search_query=VHS+capturing+-+Linux

[lordsmurf]

try checking out some of the youtube videos on VHS capturing using LINUX -

Most of those are pretty terrible.
And honestly, that advice isn't much better than "search Google" ... which is probably why the OP is here.

[pulga]

I want to thank you all for helping me so far!

Currently I'm trying to borrow a professional VCR from an acquaintance. I also worked my way a bit into avisynth and QTGMC using mainly this guide (it's in german, but the linked video is english). However, there are a few questions in the air for me, maybe someone can help me with that too:

1. Apparently the capturing device I have is a Macrosil, not a UTV007. The driver seems to work, I've already done capturing on Windows with it, but I'm starting to think about if I even get lossless signal with it. I read here that this device seems to compress the signal beforehand? Here's the mediainfo of one test capture with VirtualDub 1.9.11 (the package by lordsmurf), no filters or denoising, just HuffYUVMT:
   

   General
   General
   Complete name                            : /run/media/Videos/tddzdztdzudzutdtzutdtuztdz.avi
   Format                                   : AVI
   Format/Info                              : Audio Video Interleave
   Format profile                           : OpenDML
   File size                                : 3.55 GiB
   Duration                                 : 7 min 55 s
   Overall bit rate                         : 64.2 Mb/s
   
   Video
   ID                                       : 0
   Format                                   : HYMT
   Codec ID                                 : HYMT
   Duration                                 : 7 min 55 s
   Bit rate                                 : 62.6 Mb/s
   Width                                    : 720 pixels
   Height                                   : 576 pixels
   Display aspect ratio                     : 5:4
   Frame rate                               : 25.000 FPS
   Standard                                 : PAL
   Bits/(Pixel*Frame)                       : 6.041
   Stream size                              : 3.47 GiB (98%)
   
   Audio
   ID                                       : 1
   Format                                   : PCM
   Format settings                          : Little / Signed
   Codec ID                                 : 1
   Duration                                 : 7 min 55 s
   Bit rate mode                            : Constant
   Bit rate                                 : 1 536 kb/s
   Channel(s)                               : 2 channels
   Sampling rate                            : 48.0 kHz
   Bit depth                                : 16 bits
   Stream size                              : 87.1 MiB (2%)
   Alignment                                : Aligned on interleaves
   Interleave, duration                     : 50  ms (1.25 video frame)

   My question now is: Would it be worth to use a different capturing device, like the Hauppage USB Live2? I cannot find the ATI TV-Wonder 600 USB, so that's seems to be no option. I also would like to use it capturing Hi8 and Super8 videos, if I can get a suitable player for that.
   
2. Regarding Hi8: I still have the camera used for recording the tapes (Sony SteadyShot Handycam CCDTR910E PAL), it has Composte and S-Video ports. Is it a good idea to capture directly from camera or are there better options? I certainly think, that there are professional devices for that which would be a slight overkill for me, but if the camera is the worst of all options I would think about getting something different for that.
   
3. I also would like to know if I can treat VHS. Hi8 and Super8 the same workflow wise or if there are significant differences between the formats.
   
4. Since I'm planning to keep the lossless captures are there standard codecs for that? I noticed, that there are different versions of HuffYUV that are obviously not compatible so that may be not the best idea, I also have problems capturing with Lagarith. What about capturing with HuffYUV and converting the videos to x264 CRF=0?

[KarMa]

What about capturing with HuffYUV and converting the videos to x264 CRF=0?[/LIST]

FFV1 has a better compression ratio than x264 lossless, so it's not the best for that. FFV1 pretty much beats everything besides the crazy slow YULS (not worth it), this is my main reason for not using x264 lossless. But I use it for lossy capture. I'd capture to Huffyuv or Lagarith and then do a final FFV1 lossless conversion.

[abolibibelot]

@ pulga

I cannot find the ATI TV-Wonder 600 USB, so that's seems to be no option.

I couldn't either, searched repeatedly... é_è
Is there a place where this device (or the 650 one) can still be purchased, from Europe, at a “reasonable” price, shipping included ?

I noticed, that there are different versions of HuffYUV that are obviously not compatible so that may be not the best idea

What do you mean by “not compatible” ?

What about capturing with HuffYUV and converting the videos to x264 CRF=0?

Based on KarMa's reply, the most efficient yet practical option for archival would seem to be FFV1 + FLAC.

@ KarMa

FFV1 has a better compression ratio than x264 lossless, so it's not the best for that. FFV1 pretty much beats everything besides the crazy slow YULS (not worth it), this is my main reason for not using x264 lossless. But I use it for lossy capture. I'd capture to Huffyuv or Lagarith and then do a final FFV1 lossless conversion.

At least here are some clear-cut answers for one small part of the conundrum !
Could you provide approximate estimates regarding the respective compression efficiency of those codecs ? Or are there reliable comparison results available somewhere, including at least all the codecs mentioned above ? (Except perhaps YULS which I had never heard of before and seems to be of little practical use, even for archival purposes.)

[KarMa]

FFV1 has a better compression ratio than x264 lossless, so it's not the best for that. FFV1 pretty much beats everything besides the crazy slow YULS (not worth it), this is my main reason for not using x264 lossless. But I use it for lossy capture. I'd capture to Huffyuv or Lagarith and then do a final FFV1 lossless conversion.

At least here are some clear-cut answers for one small part of the conundrum !
Could you provide approximate estimates regarding the respective compression efficiency of those codecs ? Or are there reliable comparison results available somewhere, including at least all the codecs mentioned above ? (Except perhaps YULS which I had never heard of before and seems to be of little practical use, even for archival purposes.)

Can check these ratios out, they are old but a good guide. https://forum.doom9.org/showthread.php?p=966606#post966606
With "FFV1 (AC, large context, keyint=50)" being roughly what I would use. A decent comparison of speed vs compression ratio for lossless codecs, http://yuvsoft.com/downloads/download/lossless-codec/lossless_codecs_comparison.png which comes from a well known Moscow University comparison program. That MSU codec comes from the Moscow University, and even though they claim it to have great compression I have not found that to be the case and its slower than YULS in my opinion.

FFV1 is commonly used by archives for its practical high efficiency and non-proprietary nature. FFV1 also inspired the new image format FLIF, as a higher efficient alternative to PNG. FFV1 + FLAC in an .AVI or .MKV would be recommended for long term lossless.