Is lossless really necessary?

[dellsam34]

So yes, we could "capture" VHS signal at 352x576 without loosing information, but we do not. This is of couse not true for S-VHS and other analog media.

I know you're saying this for the sake of argument but we actually can't, capture cards sample at 720x576 (or 480 for NTSC) at the chip level, You can tell the driver/software to resize to whatever resolution at the output of the software but the actual conversion to digital is always @ 720 at the hardware level.

[lollo]

Folks, stop feeding the troll.

So yes, we could "capture" VHS signal at 352x576 without loosing information, but we do not. This is of couse not true for S-VHS and other analog media.

I know you're saying this for the sake of argument but we actually can't, capture cards sample at 720x576 (or 480 for NTSC) at the chip level, You can tell the driver/software to resize to whatever resolution at the output of the software but the actual conversion to digital is always @ 720 at the hardware level.

Yes to both. My reply was just a "lifesaver" to sophisticles, because I was really arsh with him, although he deserves it

[usually_quiet]

Folks, stop feeding the troll. These type of peeps don't get ashamed by making a fool of themselves on the internet as long as they are hiding behind a screen name, I don't know what's the medical name of this disorder.

Thanks. I'll admit that I haven't read many of the OP's other posts so I wasn't sure I was dealing with a troll. However, I also don't want anyone who runs across this thread while searching for help to be misinformed by what the troll says.

[sophisticles]

The analog PAL signal (as example) is characterized by 625 lines, 50 interlaced half-frames (or fields) per second, 25 frames per second, and has a bandwidth of approximately 5 MHZ. Note that when speaking of analog signal there is no mention to pixels either in terms of horizontal or vertical resolution, but of bandwidth and number of lines.

The analogue PAL signal must be identified with a waveform (normally a voltage variation over time). This wave, which has its own particular shape, is divided into 625 pieces, which are easily identifiable and can therefore be counted thanks to a series of impulses that mark the beginning and end. Each of these 625 pieces of wave (lasting 64 microseconds) contains information and in particular 576 (does this number sound familiar? ) pieces contain the information, a couple contain test signals, the same number contain teletext information, and others all those synchronisms that allow the TV to know where to draw that line.

For what we are interested in, of the 625 lines of the Pal signal only 576 contain video: what we need to understand is how these lines are organized and what video information they contain. Without going into too much detail, to define what information is contained in a line we need to talk about band.

In practice, an analog video line, which as seen is a small piece of wave, could potentially contain very high information, where by video information we mean the number of points, for example white and black, that I can distinguish on a line: it is clear that if in this line I can insert 1000 distinct white and black points (500 white and 500 black alternating), I have much greater information, in terms of resolution, than if I were able to insert 200 points (100 and 100).

The set of the maximum number of distinct points that I can insert in each line and the number of lines (576 in PAL) define the (analog resolution) of the analog signal: greater (analog) resolution = greater ability to distinguish details in an image.

Considering Shannon's sampling theorem and doing some math, in theory a Pal TV signal at maximum quality (5 MHz band) can allow 384 distinct black/white alternations on one line. Given what has just been said (384 distinct white/black alternations), 768 (does this number sound familiar? ) is the number of samples (pixels) that are able to describe without any loss of quality a PAL analogue signal at maximum quality (5 MHz band).

It means that if I take a line of PAL signal, I digitize it using an A/D converter, transforming it into 768 pixels and then from these 768 I obtain a new analogue line, this is exactly the same as the original line (at least in theory, since it is impossible to do such operations without errors and therefore the introduction of noise and other).
It also means that if I use the same procedure with e.g. 900 points, I always get a row identical to the first, but I simply wasted precious bits. On the contrary, if I use 500 pixels, the line I obtained will not be the same as the original, but will have less details (high video frequencies): from a visual point of view I would have a blurrier line.

768*576 25 frames/sec (does it sound familiar? ) is the "magic" number of pixels after digitalisation that describe a 5 MHz PAL signal without loss of quality. In practice, a video signal, never contains such a high level of information (details, colours, ....) that it requires 768 points for each line, and 720 (does this number sound familiar? ) points are enough to represent the single PAL line of a quality video. Hence the standard ITU-R BT.601-4 (formerly "CCIR-601" or "Rec.601") 720x576.

If instead of a quality PAL signal, we start from a signal stored on a VHS, we could see that 720 is excessive (352x576 is enough, does this number sound familiar? ), but in practice the "Rec.601" standard is used.

So yes, we could "capture" VHS signal at 352x576 without loosing information, but we do not. This is of couse not true for S-VHS and other analog media.

This makes perfect sense.

Despite the personal smear, I thank you for this explanation.

[sophisticles]

So, now I'm curious... Exactly what method did you use to capture VHS in 1994?

Maybe it wasn't exactly 30 years ago, but back in the 90's I had bought a Compaq that I upgraded with what at the time was a massive 128mb ram and I had added a second 10gb hdd to supplement the 10gb drive it came with.

I had added an capture card that had coax in capabilities, and I had run a splitter to split the cable signal going to my TV.

Back then it was possible to bypass the cable box and still get most channels.

I started with capturing The Simpsons, Married With Children and X-Files along with football games.

Somewhere along the lines I got the idea to rent VHS tapes from the local rental place and record them into my computer.

I remember capturing a few of them, basically instead of outputting the signal from the VCR to the TV, I outputted it to the capture card and used a video for Windows application to record.

I quickly ran out of space and the computer did not have room for another internal drive,

I did eventually get an external drive to move all my captures to but by then DVD was out and I had lost interest in VHS.

Unfortunately years ago that drive gave up the ghost and took all those captures along with it.

I still have a JVC S3600U sitting behind me that hasn't been used in years.

[sophisticles]

Folks, stop feeding the troll.

I'm a troll because i have a different opinion than what is commonly accepted and think that it may be time to try different methods, considering how many posts you see in forums all over the place with people having the same problem.

At some point you need to investigate if there is a better way.

[sophisticles]

Folks, stop feeding the troll. These type of peeps don't get ashamed by making a fool of themselves on the internet as long as they are hiding behind a screen name, I don't know what's the medical name of this disorder.

Thanks. I'll admit that I haven't read many of the OP's other posts so I wasn't sure I was dealing with a troll. However, I also don't want anyone who runs across this thread while searching for help to be misinformed by what the troll says.

You're kidding me, right?

You haven't read many of my posts?

We have interacted numerous times and have have posted on a number of the threads i have made where I posted encoding tests.

[usually_quiet]

Folks, stop feeding the troll. These type of peeps don't get ashamed by making a fool of themselves on the internet as long as they are hiding behind a screen name, I don't know what's the medical name of this disorder.

Thanks. I'll admit that I haven't read many of the OP's other posts so I wasn't sure I was dealing with a troll. However, I also don't want anyone who runs across this thread while searching for help to be misinformed by what the troll says.

You're kidding me, right?

You haven't read many of my posts?

We have interacted numerous times and have have posted on a number of the threads i have made where I posted encoding tests.

I don't think so. Encoding tests are not of much interest to me. I rarely read more than the first or second post in those types of threads let alone reply to them. The closest I have come to that subject in the past year was a recent thread titled "Why does x265 suck so bad?". As it happens, you did post there too.

[dellsam34]

I'm a troll because i have a different opinion than what is commonly accepted and think that it may be time to try different methods, considering how many posts you see in forums all over the place with people having the same problem.

At some point you need to investigate if there is a better way.

People having the same problem because they use the same cheap junk you are referring to, I did investigate and used different methods the mainstream don't use and work pretty well for me, But not anyone willing to go that route, Most people with problems go for out of the standard cheap stuff and cheap stuff don't work, simple as that.

[lollo]

Despite the personal smear

It was not my intention. Nothing personal, just a criticism to your video-related stuff

[sophisticles]

I'm a troll because i have a different opinion than what is commonly accepted and think that it may be time to try different methods, considering how many posts you see in forums all over the place with people having the same problem.

At some point you need to investigate if there is a better way.

People having the same problem because they use the same cheap junk you are referring to, I did investigate and used different methods the mainstream don't use and work pretty well for me, But not anyone willing to go that route, Most people with problems go for out of the standard cheap stuff and cheap stuff don't work, simple as that.

Simple question.

Assume you had a VHS tape that contained some critical data, birth of a child, evidence from a crime scene, evidence for a lawsuit, something where you needed to preserve the content as precisely as possible.

You wouldn't do what is advocated in forums around the net, you would probably contact a forensic lab and hire them to make a forensic backup of the content.

So the question is what equipment are they using?

I doubt they hook up a VCR to a TV and then bang their heads against the wall trying to fix the captured footage.

What is the absolute best capture card i can buy right now, money no object?

What is the best way to extract the data from the VHS tape before it is passed to a capture card, money no object?

Getting back to the lossless issue, people keep advocating for lossless workflows and yet no one had instructed anyone that for reliable lossless work you really should be using RAID , at which point you really want a minimum RAID 0+1, if not RAID 5.

At the very least, if you are planning on using lossless then at least advise a really fast NVMe that can hit I/O of 1gb/s.

I guarantee you there are people trying to capture lossless using spinning rust.

[sophisticles]

Despite the personal smear

It was not my intention. Nothing personal, just a criticism to your video-related stuff

Not a problem, I tend to have polarizing opinions, like I think CRF is a scam, i consider x264 to be over-engineered and x265 to be even worse, I think lossless is a waste of time and space, useful for archiving but not for actual production work.

I also think that people do things because they have always been done that way or because some cult of personality decreed that such and such is the best way to do something and God forbid anyone questions it.

I believe in the proof is in the pudding and if you have thousands of threads around the net with people basically having the same problems despite following the recommended workflow, at some point you need to ask yourself if this is really the best way.

[lordsmurf]

I guarantee you didn't capture anything in 1994.

How do you guarantee it?

How do you know what I have captured in my life?

Hmm... I was an adult in the 1990s. People used to copy their tapes using a second VCR or point a video camera at their TV screen but I don't think that counts as capturing. There was a consumer editing device and software for computers available then that cost thousands of dollars by itself ...not to mention thousands more for additional equipment (not counting the computer). Tapes were used to store the output.

So, now I'm curious... Exactly what method did you use to capture VHS in 1994?

Yep.

I didn't start to browse digital capture options until about 1995. At that time, it was entirely closed-sourced $$,$$$+ proprietary hardware (essentially what became used for DVD-Video). Or something that was so incredibly crappy that it was essentially a digital flipbook made with crayons -- all for the low price of about $500! And ISA or SCSI, too!

Around 1997, I got to use SGI MediaBase servers, and that was almost like VCD quality for streaming (before the word "streaming" really existed). DVD-Video quality, and better, was possible, and SGI had about 5 years advantage over x86. So I waited, and kept restoring video in the analog domain.

DV was viable starting with the Pemtium III in 1999, and started to get "passthrough" options around the same time. That's also when the Canopus boxes came out. Beta-quality finally-decent capture cards came in 2000 from Matrox and ATI. But it was really 2001 and the ATI AIW Radeon that it became viable for under $500, and with max possible analog quality. I still have the ads from the back of computer and photo/video magazines, showing the options and pricing.

Most of the crappy cards of the era all used BT8x8 (Conexant) chips, from late 90s into 00s. Some of the earliest VH posts about capturing were about problems with those cards.

So, again, claiming 1994 is complete BS.

So, now I'm curious... Exactly what method did you use to capture VHS in 1994?

Maybe it wasn't exactly 30 years ago, but back in the 90's I had bought a Compaq that I upgraded with what at the time was a massive 128mb ram and I had added a second 10gb hdd to supplement the 10gb drive it came with.

I had added an capture card that had coax in capabilities, and I had run a splitter to split the cable signal going to my TV.

Back then it was possible to bypass the cable box and still get most channels.

I started with capturing The Simpsons, Married With Children and X-Files along with football games.

Somewhere along the lines I got the idea to rent VHS tapes from the local rental place and record them into my computer.

I remember capturing a few of them, basically instead of outputting the signal from the VCR to the TV, I outputted it to the capture card and used a video for Windows application to record.

I quickly ran out of space and the computer did not have room for another internal drive,

I did eventually get an external drive to move all my captures to but by then DVD was out and I had lost interest in VHS.

Unfortunately years ago that drive gave up the ghost and took all those captures along with it.

I still have a JVC S3600U sitting behind me that hasn't been used in years.

At very earliest, you did that in 1999. And you surely had a BT8x8 card.

[lollo]

I believe in the proof is in the pudding and if you have thousands of threads around the net with people basically having the same problems despite following the recommended workflow, at some point you need to ask yourself if this is really the best way.

It’s the opposite.

The thousands of threads around the net with people having the same problem is because they do not follow the recommended workflow (high-end S-VHS with TBC, optional external TBC, recommended capture card, recommended software, lossless YUV 4:2:2 interlaced capture)

You can easily verify it by yourself even here on videohelp forums!

[dellsam34]

Simple question.

Assume you had a VHS tape that contained some critical data, birth of a child, evidence from a crime scene, evidence for a lawsuit, something where you needed to preserve the content as precisely as possible.

You wouldn't do what is advocated in forums around the net, you would probably contact a forensic lab and hire them to make a forensic backup of the content.

Forensic labs have moved on from VHS, they no longer have such equipement, But for the sake of argument, let's say a cold case have been reopened after finding a video tape somewhere in storage, The investigators would try to source the equipment if they can, be it a time-lapse VCR for security cameras, a consumer VCR for low speed consumer tapes or a pro VCR for SP tapes which a high end consumer VCR can do better due the fact that pro VCRs are worn out from heavy use and will require service, or a camcorder if the tape is a camcorder format. But since they lack the knowledge in such equipment they will most likely put a bid for someone to come and help, A knowledgeable hobbyist like some of us here with different types of equipement pro and consumer, Or a well established business in high quality work like Colin from video99.co.uk, Obsolete video service, digitalfaq.com and the likes, people that repair and tune their own equipment just like some of us the hobbyists, not got-memories, west coast digitizing, kodak, legacybox, these are the fast food joints not restaurants with chefs.

There is no such "what's the best capture card money can buy?", People who often ask this question are ignorant people, and I don't mean by ignorant insult, people with very limited knowledge in the field like yourself, just like me going into retro gaming community or retro computing and ask what is the best retro unit money can buy today? they would laugh at me because clearly shows I know nothing about the field and I really don't, not a hypothetical scenario.

[Alwyn]

I believe in the proof is in the pudding and if you have thousands of threads around the net with people basically having the same problems despite following the recommended workflow

The problem (for you) is that they don't have any problems if they follow the recommended workflow.

Or a well established business in high quality work like Colin from video99.co.uk,

Wazzat, capturing into DV-AVI??!!

[sophisticles]

Or a well established business in high quality work like Colin from video99.co.uk,

That Colin guy seems like the real deal.

Also, thank you for proving me right, at least partially.

We usually provide DV-AVI files of 13GB per hour from most video formats. These are much higher quality than the compressed files you would get from a cheap video to USB converter gadget, and are much better suited to editing. In the case of mini dv and Digital8 tapes, these DV-AVI files are a pure lossless copy from the digital tape contents which even retain the original time/date information which some software can access. Additionally I offer MPEG4 files of about 4GB per hour, ideal for playback on most modern TVs as well as portable devices, the extra cost for these smaller files is very minimal. Some professional studio tapes are available as other video file formats including 10-bit YUV (huge file sizes though).

Notice anything?

Maybe I'm not that ignorant after all.

[Alwyn]

That Colin guy seems like the real deal.

Also, thank you for proving me right, at least partially.

You don't know what you don't know. Or are you joking?

[dellsam34]

Wazzat, capturing into DV-AVI??!!

He does actually do lossless upon request or any other encoding format, it just happen that most people in the UK prefer DV.

[dellsam34]

Maybe I'm not that ignorant after all.

Not really, DV is just what his customer base prefer, he does offer lossless AVI especially since we are talking forensic and valuable. What he means by cheap USB devices is the lossy software that comes with those chinese knockoff devices like easycap, they convert directly into mp4, he actually proved you wrong since you suggested one of those businesses.

Me personally DV is an obsolete codec just like MPEG-2 and most people here in the US don't know what to do with DV anyways, so it will endup converted to a modern lossy codec which incur 2 losses, one during DV conversion from lossless and the other during the conversion from DV to the modern codec.

Native DV and D8 tapes are transferred as such, there is no choice there, but it's up to the customer if he wants raw DV files or have it converted to a modern codec, I've done few D8 and DV tapes and almost all of them wanted HD mp4 container, So stripped DV out into bare AVI, de-interlaced with QTGMC, resized to 1440x1080 and encoded into h.264 4:2:0.

[Alwyn]

Hang on, Video99UK says himself he captures into DV "For most domestic formats". He makes no mention of Lossless>MP4 so I assume he's doing DV>MP4, exactly what you're criticising here.

I wouldn't be considering him "best practice" at consumer video conversion. He is appears to be a tech whizz, but using DV as a base for a paid service would ordinarily attract howls of condemnation here.

Anyway, back to topic...

[Sharc]

The analog PAL signal (as example) is characterized by 625 lines, 50 interlaced half-frames (or fields) per second, 25 frames per second, and has a bandwidth of approximately 5 MHZ. Note that when speaking of analog signal there is no mention to pixels either in terms of horizontal or vertical resolution, but of bandwidth and number of lines.

The analogue PAL signal must be identified with a waveform (normally a voltage variation over time). This wave, which has its own particular shape, is divided into 625 pieces, which are easily identifiable and can therefore be counted thanks to a series of impulses that mark the beginning and end. Each of these 625 pieces of wave (lasting 64 microseconds) contains information and in particular 576 (does this number sound familiar? ) pieces contain the information, a couple contain test signals, the same number contain teletext information, and others all those synchronisms that allow the TV to know where to draw that line.

For what we are interested in, of the 625 lines of the Pal signal only 576 contain video: what we need to understand is how these lines are organized and what video information they contain. Without going into too much detail, to define what information is contained in a line we need to talk about band.

In practice, an analog video line, which as seen is a small piece of wave, could potentially contain very high information, where by video information we mean the number of points, for example white and black, that I can distinguish on a line: it is clear that if in this line I can insert 1000 distinct white and black points (500 white and 500 black alternating), I have much greater information, in terms of resolution, than if I were able to insert 200 points (100 and 100).

The set of the maximum number of distinct points that I can insert in each line and the number of lines (576 in PAL) define the (analog resolution) of the analog signal: greater (analog) resolution = greater ability to distinguish details in an image.

Considering Shannon's sampling theorem and doing some math, in theory a Pal TV signal at maximum quality (5 MHz band) can allow 384 distinct black/white alternations on one line. Given what has just been said (384 distinct white/black alternations), 768 (does this number sound familiar? ) is the number of samples (pixels) that are able to describe without any loss of quality a PAL analogue signal at maximum quality (5 MHz band).

It means that if I take a line of PAL signal, I digitize it using an A/D converter, transforming it into 768 pixels and then from these 768 I obtain a new analogue line, this is exactly the same as the original line (at least in theory, since it is impossible to do such operations without errors and therefore the introduction of noise and other).
It also means that if I use the same procedure with e.g. 900 points, I always get a row identical to the first, but I simply wasted precious bits. On the contrary, if I use 500 pixels, the line I obtained will not be the same as the original, but will have less details (high video frequencies): from a visual point of view I would have a blurrier line.

768*576 25 frames/sec (does it sound familiar? ) is the "magic" number of pixels after digitalisation that describe a 5 MHz PAL signal without loss of quality. In practice, a video signal, never contains such a high level of information (details, colours, ....) that it requires 768 points for each line, and 720 (does this number sound familiar? ) points are enough to represent the single PAL line of a quality video. Hence the standard ITU-R BT.601-4 (formerly "CCIR-601" or "Rec.601") 720x576.

If instead of a quality PAL signal, we start from a signal stored on a VHS, we could see that 720 is excessive (352x576 is enough, does this number sound familiar? ), but in practice the "Rec.601" standard is used.

So yes, we could "capture" VHS signal at 352x576 without loosing information, but we do not. This is of couse not true for S-VHS and other analog media.

This makes perfect sense.

^^^This, and looking at the same from a slighly different viewpoint, based on the number of the quoted resolution lines of consumer tapes:

Assuming S-VHS has a maximum horizontal resolution of 420 lines gives 420/2=210 black/white cycles (transitions). A PAL scanline duration is 53.333 microseconds. This gives 53.333/210=25.4 microseconds cycle duration or 1/25.4us= 3.94MHz spectral width to represent a resolution of 420 lines. According to Nyquist-Shannon the theoretical minimum sampling frequency must be twice as much, means 2x3.94MHz=7.9MHz. In any practical implementation it must be higher in order to prevent aliasing and ringing. Rec.601 specifies 13.5MHz sampling rate (valid for PAL as well as NTSC) which includes some margin for practical implementation limitations. So for 420 discernible S-VHS lines there is not ample "waste" of bandwidth or digital pixels. Moreover, above consideration applies for "sinusoidal" vertical bars. Lines with sharp (steep, step, trapezoidal etc.) transitions have a wider spectral content and require a correspondingly higher sampling rate to represent (catch) these sharper/steeper transitions properly (means not smoothed/blurred or aliased).

Capturing high quality tapes at lower sampling rates means one would readily risk to loose information which may be on the tapes, apart from a non-standard sampling mess with individual chip manufacturer dependent PARs (PixelAspectRatio) etc. like in the early digitization age.

Edit: This excursion doesn't answer the original question whether lossless is really necessary ...

[usually_quiet]

Maybe I'm not that ignorant after all.

Not really, DV is just what his customer base prefer, he does offer lossless AVI especially since we are talking forensic and valuable. What he means by cheap USB devices is the lossy software that comes with those chinese knockoff devices like easycap, they convert directly into mp4, he actually proved you wrong since you suggested one of those businesses.

Me personally DV is an obsolete codec just like MPEG-2 and most people here in the US don't know what to do with DV anyways, so it will endup converted to a modern lossy codec which incur 2 losses, one during DV conversion from lossless and the other during the conversion from DV to the modern codec.

I agree that DV-AVI is certainly obsolete as a distribution format. New computers can play it (if a DV codec is installed) but no other new consumer electronics can. However, even if it isn't lossless, DV-AVI isn't a terrible choice for somebody in the UK who wants to use it as archival storage for their standard-definition interlaced PAL home video. DV is noisier (DCT ringing) and preserves less detail than lossless compression but PAL DV uses 4:2:0 chroma subsampling and consumes less storage (13GB per hour of video).

[sophisticles]

Not really, DV is just what his customer base prefer, he does offer lossless AVI especially since we are talking forensic and valuable. What he means by cheap USB devices is the lossy software that comes with those Chinese knockoff devices like easycap, they convert directly into mp4, he actually proved you wrong since you suggested one of those businesses.

Go back and reread what I suggested, I said Kodak because I thought they were Kodak and figured they would have some fancy equipment that a hobbyist would not have.

And for the record, I am not Chinese but all this talk about "Chinese knockoff" and "Chinese crap" that is used on here is messed up.

China has 1.5 billion people and they manufacture all sorts of things, while I am sure some of the low quality stiff is made in China, I would be surprised if the high quality stuff isn't also made in China.

So can we keep the racial slurs out of this? Low quality stiff can be made anywhere as can high quality stuff.

[dellsam34]

Chinese knockoffs are chinese brands made by chinese shops to mimic fully functioning products from brand name companies that are also made in china but not by chinese manufacturing entities, have you heard of the chinese iphone? Does that mean the real iphone is not made in china?, That's another area you lack knowledge of.

[sophisticles]

Hang on, Video99UK says himself he captures into DV "For most domestic formats". He makes no mention of Lossless>MP4 so I assume he's doing DV>MP4, exactly what you're criticizing here.

Of course he does, because lossless is a complete waste for VHS capture, and i would argue, for most things.

These people want to act like they are experts, and for the sake of being civil and not causing needless arguing, let's just say they are hobbyists with some equipment that have done some captures and in their mind it justifies their belief that they are experts.

But let's be honest with ourselves, in this thread, and others, they have claimed the following:

That lossless is lossless, there are not different types of lossless. I am not going to call anyone out by name, but anyone that doesn't understand that utvideo 8-but 4:2:2 is not the same as 10-bit 4:4:4 FFV1 should not be in this discussion.

It has been claimed that I am ignorant and I don't know what i don't know, and a person is held up as an expert yet that expert uses something they spent numerous posts denouncing.

They advise to use lossless but apparently RAID to them is something to kill cockroaches, because nowhere do they inform people that if you want to capture lossless you really should be using RAID 0+1.

PDR, who is a very knowledgeable person and may be one of the more knowledgeable in this forum, has contradicted himself. In another thread he pointed out that lossless is only lossless if the NLE treats it as lossless, and as he pointed out, most do not, because they were not designed with ingesting lossless formats. Yet in this thread he has advocated for lossless use and tried to use the example of EXR plates in movie production, failing to address that this applies to Sony CineAlta F65 Camera captures, most digital cameras record to RAW and ProRes simultaneous.

The lossless capture crowd remind me of the people that used to use pirated copies of CCE to perform a 9 pass MPEG-2 encode or use the placebo preset for x264 and x265 or the people that used to take their commercial Blu-Rays and re-encode them to x264 because they "wanted the highest quality".

Or the people people that will deinterlace using QGTMC with the placebo preset.

I how people that chance upon this thread will read everything carefully, maybe experiment on their own and make up their own mind.

And keep in mind that anyone that says that asking "which is the best capture card" or "the best VCR" is proof that the person asking the question does not know what he is doing yet then turns around and claims that people should be following the recommendations of the users on this forum, not realizing the inherent contradiction, may not be the most trustworthy person to take advise from.

With all due respect.

[sophisticles]

Edit: This excursion doesn't answer the original question whether lossless is really necessary ...

This is something that people need to answer for themselves.

I have terabytes of footage from various sources converted to various formats and the only thing i have seen lossless useful for is creating huge file sizes and achieving high PSNR scores.

You can't see the difference between lossless and a high quality near lossless format and I can prove it very easily.

Take any source you want and convert it to I frame only AV1 using libaom-av1 + yuv444p12le with CRF 0 and CRF 1.

Are you going to notice the difference?

Only in the file size difference.

[sophisticles]

Chinese knockoffs are chinese brands made by chinese shops to mimic fully functioning products from brand name companies that are also made in china but not by chinese manufacturing entities, have you heard of the chinese iphone? Does that mean the real iphone is not made in china?, That's another area you lack knowledge of.

You really are full of yourself aren't you?

[poisondeathray]

PDR, who is a very knowledgeable person and may be one of the more knowledgeable in this forum, has contradicted himself. In another thread he pointed out that lossless is only lossless if the NLE treats it as lossless, and as he pointed out, most do not, because they were not designed with ingesting lossless formats.

Where is the contradiction ??? Why am I on the "hitlist" . Selective reading on your part

I provided a truly lossless workflow for NLE's too... v210 for 10bit 422 is uncompressed and lossless and gets treated as lossless in Premiere Pro. Same with UYVY for 8bit422. So instead of huffyuv, you use UYVY - it's lossless and treated as YUV . Or instead of Prores you use v210. Hardcore editors used

I provided a truly lossless workflow for up/down sampling YUV<=>RGB conversion too. 32bit float and nearest neighbor up/downsampling . eg.. vapoursynth , EXR export/import

I was pointing out in the other thread that people who use "lossless" captures like huffyuv just assume it's lossless - but it's often not because of mis-handling in programs . There are a lot of "gotchas" for people that don't know the in/outs of programs and colorspaces

Yet in this thread he has advocated for lossless use and tried to use the example of EXR plates in movie production, failing to address that this applies to Sony CineAlta F65 Camera captures, most digital cameras record to RAW and ProRes simultaneous.

Yes, I mentioned EXR for VFX renders. The rendered CG elements and visual effects are done by computers and software ; did you think that stuff happens in camera ? . For visual FX scenes, and/or elements composited into live action. I was making a point that lossy renders at that stage never happens - you don't render CG stuff to Prores for example at that point . All the CG stuff requires 32bit linear float , scene lighting , for compositing to look realistic. You can't do that with Prores

For some scenarios - high end intermediate VFX pipelines - None of them - ProresHQ or XQ4444/Cineform/x264/x265 etc...are good enough - They are never used at all. 10 or 12bpc is not enough. Only 32bit EXR is used. Prores is only used near the end of the pipeline near delivery

Note that I did not advocate lossless for everything - I wrote it depends on the scenario. If your end goal is Youtube, it's not going to make a noticable difference if you used lossless or ProresHQ for capturing

It depends on the scenario , and criteria.. obviously. If you're looking at the end of a lower quality distribution format , it's usually more than good enough.

Actually near lossless is great for 99.9% of scenarios. I suggested x264 qp 1 when you need high quality - because the quality is 10-20 db higher then ProresHQ . For the scenarios where ProresHQ is not enough

Nice that you bring that up - x264 10bit 422 intra is fully adjustable, you can get about 10-20 db PSNR higher than ProresHQ at qp1, if ProresHQ wasn't "good enough" . It doesn't get mishandled like codecs in lossless mode either in many programs like NLE's. You have field order, SAR (aspect ratio) signalling.

[_Al_]

quote by op:

PDR, who is a very knowledgeable person and may be one of the more knowledgeable in this forum, has contradicted himself. In another thread he pointed out that lossless is only lossless if the NLE treats it as lossless, and as he pointed out, most do not, because they were not designed with ingesting lossless formats.

Besides his own answer to defend for himself,
you and many others that panic , shouting save yourselves, flee! But, it is all because of illegal levels. If you have legal levels fixed, it is just YUV to RGB and back errors, which can be greatly diminished by using floating point in Vegas for example, heck, it is SD only. So much fuss in that thread or here for basically nothing, even after it was brought up again and again. Just because of bloody level legalizing, which is very known to Vegas user who visits forums. Those users know what to do with particular footage, or even type of camcorder, what to do set things right. They fix the footage. The others, using a "recommended" NLE, don't do anything and watch blow up colors anyway on their screen when YUV to RGB goes on monitor and everything above 235 is gone anyway.