Why not capture analog video/audio both lossless?
 

I always captured analogue video in huffyuv/yuv + pcm from VHS.

Is there any reason why not compress audio lossless with FLAC on the fly, please?

a/v sync issues?

With the same line of reasoning, one could wonder why not use x264 with crf =0(lossless) for capturing

Here is my reason, there might be other reasons I do not know of:

The goal when capturing is to use the lightest codec possible with the least load on the CPU. FLAC is a bit more CPU-intensive.

What will happen if your CPU is too overloaded while capturing is that you are going to drop frames. Something you do not want. My guess is with reasonably modern CPUs (like i5/i7 3rd gen or higher) doing FLAC encoding on the fly is probably manageable.

FLAC is likely too resource intensive for realtime compression. Remember, it's per-core speeds that matter in capturing, not the number of cores in modern processors. Core speeds haven't budged much in the past decade, we just added more.

There's also interleave to consider, they relationship between the audio and video. Offhand, I don't believe FLAC can be containerized with video (MKV, MP4, AVI, etc) in this way. FLAC was intended purely for audio, as separate lossless streams.

Audio is somewhat tiny anyway, even uncompressed PCM. No real reason to do it.

FWIW: Considering that audio is a small percentage of the total data stream, video being much greater. IMO in this day of low cost storage the strikes me as being little benefit to compressing audio when capturing. Audio compression can be reserved for the ultimate distribution format as appropriate. One less codec to manage.

Thanks guys for reply

x264 was never intended for edit, cannot capture in plain YUV if I recall.

I doubt Ryzen 9 5950X @ 4.1GHz would be issue or Xeon E5-2697v2 in my computing rig. We used to do it on Athlons, Pentiums guys...I am sure smartphone can do it if one can install capture card :o)

I see your point. I never had issues with old, proven, mature, and "obsolete" tech. Despite I hate it in my SounDevices MixPre recorder...BWF uncompressed WAV 32b float @96kHz is not much fun if you record 1TB data over a week in wilderness.

"Waste of cheap space" strategy...that is exactly how you end up with 1PiB server :-P In my best days of photography, I brought 256GB of RAW photos per week.

For RAW video edit HDD cold storage is crap... SSD tech still costs 100TB cca 50k$. Even copy over 10Gb-eth is annoying.

Every two years, I had to upgrade server ROFL while audio is tiny, like 600MB per hour...1000 hours is 600GB...i like to upgrade my storage every 5years nowadays :o)

Both lossless audio (such as PCM) and compressed audio (such as AAC) can be used, However audio format compatibility with playback devices is very important here, Most devices cannot playback videos with unusual audio formats, or formats that are anything other than 48KHz/16bit. So you have to know what's your playback target, Online file sharing, flash media, offline server playback, physical media formats, also what devices are going to playback those videos, TV, smart device, media player... etc.

Here is a nice write up about some audio formats for video.

As mentioned above by "compatibility issues" is main problem as I have already experienced with legendary VirtualDub2/AVIsynth on Win10. Might be also problem with Win10 that causes to fail. AVIsynth script fed RAW video huffyuv/PCM, filtering, VirtualDub2 compress huffyuv, and transcode audio to FLAC.

It came out ready for Thrash...video ok, audio noises similar to alien language.

BOTTOM LINE is K.I.S.S as usually

Win huffyuv + PCM
Linux ffv1 + PCM

live happily ever after

FLAC audio is supported in MKV files so it's possible if you capture with ffmpeg (and you intend to use it with another ffmpeg-based application) - it won't work in .avi files though as those are more limited in what audio formats are supported so best to just use PCM there.

For long term archival purposes flac+ffv1 in .mkv will give you some slight space saving over just leaving the audio as PCM if you need it.

For non music tapes, such as documentary and dialog, I use AAC @ 192Kbps, 16/48 works like a charm on every device I have tested.

I have recently read about Swiss archival studio working for gov institution. They push with other studios some standards. They use mentioned ffmpeg > ffv1/flac/mkv combo as OpenSource standard for archival purpose.

They do not share what they do with tapes. Since they are sponsored from our tax money, they may have some studio-grade HW. They also push some archive industry standard with RAWcooked

As hobby, I have already "wasted" at least 10k $ in analogue/digital photography, video is next step. I would like to get some full-frame TBC for max quality. This ancient stuff is hard to get in Europe. VCRshop guys have some, but it is something that Maria Theresia used.

Suggested TBC by lordsmurf are nowhere in Europe....or I didn't find anything yet. All European pro studios claim professional grade analogue/digital conversion...all they do is decent S-VHS deck + Grass valley AVDC 110/300 combo while preaching about poor hobby guys like us, and their "best" RAW video quality.

It is either pure skull poverty/ignorance, or they just exploit the ignorance of normies. I have already done better captures with tourist gear than these "experts". I decided to "waste" another 1k $ to get it done correctly because you will pay same price for poor capture.

It boils down to a few questions.

What formats will be supported over the life of the captured video. If the files will be discarded when you are off the scene that may be a different answer than if you want them read 200 years from now. Also you may need to provide a legacy playback capability, or provide for media conversion over time as things age out. Think 8-track tapes as an example. Look at the history of innovative (at the time) media formats that have died. Sony's MiniDisc anyone? However it is still relatively to setup to play (and clean up) 78 rpm records.

Who is the ultimate customer, now and in the foreseeable future? Will your approach meet their expectations? Will the future customer have the technical expertise needed to use the recorded material?

How much time and energy (and $$$) are you willing to invest in the process?

As an aside: my observation is that few people care about their wedding videos a few years into the future, especially if a divorce is involved.

I am not worried about future because we are currently at brink of extinction so anything intangible, such as digital world, will be pointless :-D Look at crypto mania...we build it, I like the idea, but where you pay for RentalWoman with crypto while in global conflict? There will be no electricity, and as usually, barbaric relics will rule the Earth ;-)

Where do you play with fancy electronic, without own generator. How do you connect to Internet? Post-apocalyptic kids cannot even live without Internet. Such archival worries do not bother me much ;-)

Microsoft works on some glass-based cold storage. As once asked by some great artist - what legacy brings the 1900-2000+ generations...art is dead, humanoids suffer from raging skull poverty. As normal mortal - without A-bomb bunker where you can store all the old-school stuff - why worry about future ;-)

if we survive by some miracle, there will always be another formats.

I keep analogue film from my MF camera. Doesn't exist digital tech capable to store all light...PERIOD...don't get me started with reproduction of digital. I have 20+ old footages that still work huffyuv/PCM. As long as Open Source will work, I have no worries about Open Source formats. They will always be here.

If you worry, transcode. I move my archives every 5 years. There is always new tech, larger, faster.

In Europe you should be looking for Snell & Wilcox devices, They are not just TBCs, they are analog to digital converters with SDI out, The idea of TBC started back in the day when people were doing tape dubs and video editing from one deck to another, That analog workflow is used with consumer capture devices, Big buck archiving business use studio grade A/D converters. S&W managed to get their boxes to half rack size using FPGA in the Kudos+ model series vs the old pizza box proprietary designs.

Look for TBS models, and note that S&W was later on acquired by Grass Valley and they kept the same design but changed model numbering to CRV, Here are some manuals:

TBS800 CRV600
There are more than 8 models I believe based on output options for audio, Some have firewire for DV which is not needed.

Thanks chief. I am betting that is not 500€ toy, even the old ones.

-- merged --

I have found one SW CVR-600AD. There is some weird audio 4pin input.

Do I need to run audio via video sync as well or can I bypass it, and connect directly to capture card, please?

Sadly, many are just dumb newbies themselves. Simply by visiting this site, and reading, you now know more than they do.

They do it for money, not for themselves, not for quality because they know most people have no idea (including themselves...before I gained more wealth, I wasn't aware of some common analogue issues as well...I see now in my 20+y old captures, how crap they are).

They buy Grass Valley ADVC110/300, and claim "professional" gear. If you do something for money, there is no wealth involved. You are still poor, earning paper from clueless.

They do some "AI" denoise/deinterlace, compress to MPEG2...that's all folks. Unless, it looks really crap, and the customer still has tourist VCR player from Julius Caesar era...they see no difference. You can be proud, or rich ;-)

We have built Internet many years ago, and modern people are so lazy/ignorant to even read. Some at least think, and see, that 200$ "pro" digitalising is the same as buying the ADVC series themselves. Because some smart modern kids, can do what they inherited from us "pioneers"...Handbrake and similar monkey SWs are the simplest solution if people do some digging first

Common price in Swiss is 40CHF/40$ per hour. Everybody has got at least 4 VHS tapes from past. ADVC is "perfect" solution.

I am for max. amateur quality for reasonable price. I already "wasted" 1000$ on gear, it achieves perhaps 90% of broadcast/studio quality...chasing the remaining 10% is a very expensive hobby. I hoped to meet someone from Europe who already has all this gear, and can digitalise. No luck so far.

It would be the easiest solution, but I do not learn anything ;-) In third wave of my 20+ years digitalising career, I still find gaps in knowledge.

Strange, i allways thought compression during capture, makes that data can be written faster to storage, and uncompressed data during capture is greater in volume, so it will hold up storage onto the storage volume, hence dropped frames...

Write faster to storage: yes, compute faster on CPU: No.

Dropped frames can happen whenever there is a bottleneck, CPU, hard drive writing speed, or even sth else. It does not matter which.

I doubt there are still bottlenecks in modern computing ;-)

CPU idling ~10%
SSD/SATA 300-500MBs sustainable write
SSD/NVMe 1000-1500MBs
DDR4 50GBs?
Even RAMdisk manages at least 1GBs

Of course, if someone uses old rig I understand.

As already written - major issue will be compatibility.

I have tried h264 lossless, it hogs cpu - it cannot be opened in DaVinci Resolve. FLAC not bad, I guess same issue as h264

You would be wrong. Analog video conversion still operates in a per-core world. Having multi-core CPUs simply does not matter. So in terms of CPU overhead, look at core speed, not # of cores. Almost all CPU gains in the past decade were just x86 adding cores, not really meaningfully improving speeds of cores.

Look closer at that measure. It's for all cores. Drill down.

Does not matter. The fragmented nature of SSD can actually cause drops. It fully depends on the 'LC and the caching of drives. Many drives waiver in throughput (not truly sustained), and a dip in realtime write, therefore drops frames. Drive fill becomes more important here.

RAM does not matter at all for capturing, aside from support the base needs of the software. Capturing is almost entirely I/O and CPU

"Old" has nothing to do with it. In fact, "new" rigs tend to give more issues than not. That's because modern computing was not part of the considerations of capture at the time, the 00s. This is why random computers can give random results.

you may be expert in video restoration, but your computing understanding lags widely behind :-P

In modern world, we use IPC (Instruction Per Cycle) measure - I bet all I have that my 16 core Ryzen 9 5950X @ 4.1 GHz beats any 4 core CPU (except the mighty gaming 5600X version optimised for single core performance) in IPC single core performance :-P

What you are basically saying, that single core performance didn't change since Pentium ROFL

I bet again all I have that any latest CPU (except the tourist ones) beats previous generations in single-core IPC performance no matter core-frequency because as we know that is only for PC tuners nowadays. Even the Ryzen 9 3xxx vs 5xxx series jumped >10% in IPC for the former ;-)

In other words, any modern CPU can do single-core/IPC faster than the previous generations. Perhaps except Intel that struggles last 10 years without proper leadership.

Of course, the 10% idle is average for all cores. And as any modern CPU does is switching cores to keep it cool. The only issue I see is that CPU load scheduling may cause some unexpected lag/frame drop that can be easily resolved by NUMA core pinning that forces to use a single core/node per process. I am sure this can be done even on MS Crapdows ;-)

I again bet all I have that even old obsolete HW will NEVER drop frames in Linux. Main frame dropper is in your OS

No. not saying that. The main advantages have been in caching, heat (TDP), RAM communication, etc. The conversation gets too technical, and I don't care to explain it. All that matters is that modern CPUs can drop frames just like older ones, there have not been major leaps in the tech, as it pertains to this exact legacy use, in order to prevent any drops.

Too bad Linux (or Mac) never had any capture software that was worthwhile.

don't say "I don't care to explain" if you have no idea. It is ok not to know everything :-P

Why do you insist on GUI? It is for amateurs ;-) You are hardcore "YUY/RAW video" guy, the same hardcore guys use CLI only :-P You just need one stable capture rig where one doesn't waste time by fixing unfixable. Win is notoriously unstable with crap drivers, that never were fixed since Win 1.0

You guys seems to love old school stuff based on AGP, PCI/ATI Wonder in WinXP. With Linux you get most stable ATI/AMD drivers you will never see in Win.

then you type "ffmpeg -i "capture_device" -vc ffv1 -ac pcm capture.avi

get a beer, talk to your family, and live little bit...100% result 0% waste ;-) I never understood people wasting time by unfixable.

Or more accurately, not have time to argue on the internet.

No.

No.

Best is best.

No.

No.

Can't, MS, alcohol really screws with your system, even a single drink. It's not fun at all. It's a disease with more limitations that most people realize or understand. I haven't had my favorite brews in over a decade now. :(

if you never tried Linux, how do you know it is crap :-P it is easy to say NO. Of course, if you expect to come, capture without steep learning curve...you will give up as most. I think in terms of energy/time/wealth/fun ratio. I regularly wasted hundreds of hours by keeping Win alive...it is kinda like cheaper marriage. 1000 hours per year for usual guy with 30CHF/h...you have lost annually 30'000CHF/USD.

You can buy every year the best gear money can buy, and the remaining 700 hours reinvest in gaining wealth. It is said that the average person acquires new wealth every 10'000 hours. If you are already wealthy (say you know the basics of PC operation) it is Albert's phenomenon of compound interest. You are done in less than 3 years (if you don't seek problems where are none, as normies do). You will be living happily ever after ;-)

I have started with UNIX, DOS, Win, in 2000 Linux, in 2011 freeBSD/macOS...since 2011 my path always goes first on macOS, then Linux...Win only for CAD/CAM, occasionally games, and VirtualDub2 again after 20 years. ffmpeg guys do they best, but VD is still king after all years.

@hajes29a

huh?
No there isn't. you are right, which is why we're no longer developing new faster hardware, it is not needed...
ofc there is! There are various cases where you could be computing data using the CPU, and writing to an SSD and the CPU, no matter how strong, it will be holding you back and producing data slower than the write speed of the SSD.

When working on big clusters, writing to Shared (SSD-based btw) storage could take longer than the rate the CPU produces data with. It really depends on the nature of data, and the computation you're doing, the software/code you are running, how it uses the hardware and the computational problem behind. Feel free to doubt bottlenecks, but I know they do exist in general, now if you are talking about video capture (which you did not specify), they also do exist, just less common...

How do you know he never tried Linux? I am not defending him, I am not his lawyer, just curious to know how you make up these assumptions.:hmm:


I have used different flavors of Linux since the age of 10 on my servers/VPS. Started With Centos 5 and 6 in CLI, I never even knew what GUI for Linux looked like until a couple of years ago and I also goofed around on mac and linux to capture videos since 2013-2015. I ended Up redoing many of my captures over the last couple of years. Does it make me worthy to you of stating my opinion since I know what a CLI looks like :question:

The fact you're speaking about "Linux" as being a single thing weakens your positions. Speak about a distro, not the entire family. Because there is a huge gap between different distros.

The fact you say Linux has stable hardware drivers could not be any wronger. The fact is, that Windows has the widest compatibility for hardware. It is statistics and facts, not opinions.
Not even gonna talk about how shitty Linux can be, but also it can be great. It depends on your distro, hardware and your usage.


As for FFMPEG, I am not even gonna tell you how many times I had to fix and recompile it myself. So no, even your almighty ffmpeg has issues and all of its features are well-tested.

Many things changed in CPU architectures. Old capture software and hardware drivers are not written to make the best use of modern CPUs.


most of what you're saying is just you being happy with what you have and forgetting to take into account many other factors, such as the fact the many of the people are not stupid and you are not the only smart guy, that others have different experiences, they just do not feel insecure to need to state it here to make a point about an OS and capture software, etc also, You're forgetting convenience and reliability when it comes to capturing.


Yes, you can find something that works fine on Linux, but not everything will always work fine.

I am not even going to talk about how unstable Linux can be.

Oh c'mon, do not be one of them. This is the first and second item in the "I just installed Linux starter pack". Just live and let live, nobody cares about the superiority of non-noobs, or non-normies.

Also why are you bothering trying to categorize people? People are not sets. Humans are not a normalized brainless cult. "You DO A and B like your brother. Therefore you have to do C since your fellow brother does it." No. Each person is free to like YUY/raw/uncompressed, without having to have the same tastes as others. People are free to disagree.

And to answer your question, people use GUI for convenience. Back when I was studying Ergonomics and HMI we were able to prove that there are things better done with CLI, others better done with GUI. Somethings require hybrid approaches. Like launching GUIs with predefined parameters and finishing the work while previewing.



You need to realize when you do multiple things that are all different, you end up being objective about OS and understanding they're tools, not your own kids. They all suck. They just suck differently. There is no good OS/OS family. We just use each for what it does best.

If you find your happiness with X and Y on OS Z, then be happy. Just do not decide that everyone who uses other methods is amateur, noob, outdated, etc.

General advice has to be general, and suitable for the widest array of people with different hardware and a reasonable amount of knowledge without huge learning curve. You are free to experiment outside of the common methods and you can still get good/excellent results if you are lucky and know what you are doing. It does not make us all inferior to almighty you because you captured using CLI.

This whole argument about OS is useless, People use what works for them, there is no better or worse. If someone managed to make Win11 works as good as Linux with no issues then what's wrong with that? I know I have, I do post process in script though, but hey, some people post process in NLE, but they make good looking videos.

I use Linux desktops daily, and actually prefer Linux for many tasks. I'm platform agnostic, equally using Windows, Mac, and Linux on a regular basis. My Win7, Win11, M2 Pro, Intel Mac, and Xubuntu share a desk. Mint on another desk, XP on others. These are just tools for tasks.

Linux for video capture is like trying to use a screwdriver to drive in a nail. Or the dishwasher for your underwear.

I started dabbling with Linux in the mid/late 90s, hated it. Tried it again around 2011, used it ever since. Referring only to desktops here, not servers. (Instead of playing with Linux in ~2011, I should have played with mining Bitcoin. Oops.)

That's it. Video capturing has needs, you need to meet them, using the correct tools.

Yep.

Yep.

That actually does not matter as much as some folks think. And many that think "they know __" actually have no clue. The tool has limits, and the tool will be the actual limiter when you have in-depth knowledge.

Yep.

Bottom lines from my soap box:
- Use what works for you and best meets your needs. We do not all have the same needs.
- All commercial software availability and updates are market driven. There have to be enough buyers to support development of new updated releases through purchase. No income stream, not development.
- Some of this is planned obsolescence, force users into new hardware and software (a self feeding spiral). I recall a marketing video (from Ms) in the 1980 when they put out a new version of Excel to compete with 123. (Excel was a real hog but the video included Bill Gates in a cameo rolling out a cart of new PS and the tag line is "Now you have a reason to get a new PC.")
- An issue of a computer reseller trade mag had a cover story that the noted Windows 95 was great because at last users would no longer be able to setup/configure their own PC; consultants would be required meaning more money for the resellers.

IMHO, the only significant improvements to word processors since the 29K WordStar.COM of the 1980s are spell checking, fonts, and WYSIWYG, and maybe the ability to embedded graphics.