SONY 4K projector May 1st Redcode vs JPEG2000?

[Nik Manning]

http://www.engadget.com/2007/04/25/sonys-cinealta-4k-digital-cinema-solution-coming-may-1st/

Looks like Sony is releasing there 4k projectors May 1st. It is huge! Super huge, but what I wanted to talk about was their encrypted JPEG2000 format. It takes 300 gb for a 4k movie. Seems to me redcode is more efficient so is there away to make redcode encrypted for something like this? That would be a marvelous workflow to just go back to redcode as the final viewing format.

When Red says they are making 4K projectors are they talking for home , studio , or movie theater usage?

[Graeme Nattress]

REDCODE leverages RAW, which is great for capture, but useless for delivery though....

We're not talking about the projectors other than we're working on them...

Graeme

[Nik Manning]

Thanks for the quick response. I don't understand however what you mean by useless for delivery? A little more info please.

[David Newman]

Projectors will need a developed RGB (or XYZ) colorspace, RAW only has one primary per pixel and requires demosaicing for full res presentation.

[Graeme Nattress]

Yup, making it efficient for capture, but useless for delivery.

Graeme

[Nick Shaw]

What about REDCODE RGB? In camera data rates may be optimised for the ultra-clean images from the camera, but could it not be used at higher data rates for editing and delivery?

[David Newman]

Depends on the advantages REDCODE RGB can offer over DCI JPEG2000, remember bit-rate advanatages (if there is any) is not a compelling reason for the D-Cinema market (and E-Cinema its name for non-DCI formats.) We know a lot about this as CineForm Intermediate just got picked up for a theatre presentation format in India (press release should be out in a week or so.) Our data rates and quality are very similar to DCI JPEG2000, but our player and encoding costs are significantly lower (no hardware required.) If REDCODE RGB is unique in other compelling ways then maybe it could be a theatre format.

[Graeme Nattress]

There's lots of possibilities in this market due to the DCI spec being a bit silly and way too expensive. It was really designed for film scans, not digitally acquired material.

Graeme

[Chris Kenny]

That article mentions $12,651/screen. Did someone misplace a decimal point?

[Graeme Nattress]

Yes, they did muck up the price somewhat....

Graeme

[Blair S. Paulsen]

Graeme's comments about the DCI spec echo others I have heard. Now we find ourselves, as proponents of end to end data based solutions, in a tough spot. Do we:

A) Promote a better alternative to the current DCI spec that would have better image quality, similar or better compression rates and require less expensive infrastructure (easier authoring would be nice). Sounds great but it seems likely to only further muddy the waters with the stakeholders in the distribution chain and fuel the criticism that constant format changes make digital solutions too risky.

B) Accept the DCI specs, even if they aren't that great, just to keep momentum for digital workflows all the way to the screen.

Coming from the indie perspective I am willing to accept the burden of the DCI spec just to engineer out the costs of film and/or tape decks.

[Chris Kenny]

What's the issue with the cost of DCI infrastructure? I'm familiar with some of the technical aspects of the spec, and I can't see a good reason why I shouldn't be able to output a DCI spec movie on the desktop. Are there crazy licensing fees or something along those lines?

[Graeme Nattress]

There are not really many tools to make a DCI package, and it's slow work. Nothing a desktop PC can't handle though.

Also, they use XYZ, which is utterly silly. I know why they did it, but their reason doesn't make any sense at all. It just wastes code values.

The DCI projectors cost way too much, and if they're anything like our Sony 4k, they're a right pain to use.....

Graeme

[M Most]

There are not really many tools to make a DCI package, and it's slow work. Nothing a desktop PC can't handle though.

Also, they use XYZ, which is utterly silly. I know why they did it, but their reason doesn't make any sense at all. It just wastes code values.


They used XYZ because they wanted some degree of future proofing in the event that new capture and display devices are developed (probably electronic) that exceed the color gamuts of current devices.

There may be issues with the DCI spec (or with SMPTE DC28, it's more "official" standards committee), but the spec is actually quite a bit more flexible than many people realize. For instance, the use of 24 fps is not mandated. Other things are left a bit open as well. But I don't see anything else out there that allows manufacturers to create devices that will work with what the studios are going to deliver, and I don't see anything else out there that will allow D Cinema to move forward - because there has to be a standard before theater owners are going to waste money investing in dead end technology. For better or worse, the standard is here, it works (you were at the Digital Cinema Summit, so you saw what I saw), and it's not that tough to comply with. It covers 2K and 4K projection, and it covers pretty much all the current needs for theatrical presentation, including multiple audio formats, subtitles, and, yes, encryption and security. It's also much better than having theaters (and festivals, for that matter) put in HD projectors and play back HDCam videotape. The best way to improve it is probably to get involved with the SMPTE DC28 group and make relevant suggestions.

[Graeme Nattress]

XYZ is wasteful though, as cameras and film don't produce that wide a gamut. The sensible approach would be to tag your RGB colour space with it's white point and chromaticities, and let the projector convert from the source space (which could be XYZ if it's CGI, or sRGB, REC709 or... Whatever) and convert to the projector RGB space. At the moment, you take your RGB and convert to XYZ, and then the projector takes that XYZ and converts to it's RGB. The reason given is that if you mandate XYZ, it eliminates the need for metadata, but given the rest of the spec all relies on metadata, this is a moot point.

By encoding the image in it's source colour space, rather than the super-wide XYZ, you maximize use of code values and will probably get better compression / precision by doing so, so the point isn't completely academic.

XYZ was chosen as not to limit the gamut. However, this means as projectors get wider and wider gamut, they might start displaying colours you didn't know were there, and things might look different. We can't monitor XYZ directly, so we can't easily check. You can't look at the master in it's native space without it looking totally whacky.

Basically, I think the tagged colour space solution is a better one than the XYZ solution, which just confuses people.

The image formats of 2k and 4k are designed around film scans and ignore the fact that there's a lot of 16:9 material being produced. That doesn't seem clever.

The only fps mentioned in the spec are 24fps and 48fps (2k Only), which to me, is limiting. There was a good question at the event about multiple fps in the same project, which to me, is a good question as you've got to look at world markets and the preponderance of 25fps and 30fps material out there, and especially documentary movies that could benefit from this.

I could go on, picking holes here and there, but quite frankly, it wasn't designed for digital origination.

Graeme

[M Most]

XYZ was chosen as not to limit the gamut. However, this means as projectors get wider and wider gamut, they might start displaying colours you didn't know were there, and things might look different. We can't monitor XYZ directly, so we can't easily check. You can't look at the master in it's native space without it looking totally whacky.

While that is indeed true, the current result seems to work pretty well, remarkably well in fact, especially given that the material is always going through two color space conversions with a middle step that can't be monitored. But the math generally seems to work, and I think perhaps the future proofing is a good idea, even if it presents limitations in terms of total image fidelity for today's formats.



The image formats of 2k and 4k are designed around film scans and ignore the fact that there's a lot of 16:9 material being produced. That doesn't seem clever.

Clever, no. But it does serve the needs of those who wrote the spec in the first place - as a very intelligent speaker brought up during the summit (I think that was Marvin from Modern Videofilm). I agree that this should ultimately be modified, but as was also stated, the spec is still a bit of a work in progress, not something set in stone. Thus the need for more voices of practicality in the SMPTE committee.



The only fps mentioned in the spec are 24fps and 48fps (2k Only), which to me, is limiting. There was a good question at the event about multiple fps in the same project, which to me, is a good question as you've got to look at world markets and the preponderance of 25fps and 30fps material out there, and especially documentary movies that could benefit from this.

While that's also true, the way the spec is written it's a bit more vague than it may appear. It uses terminology such as "up to" instead of a direct specification, and according to others who know more about it than me, there actually are provisions in the updated spec (i.e., the unfinished DC28) for 25fps projection. Having said that, I also thought the question about multiple frame rates was really brilliant. And I'm not saying that just because I was the one who asked the question :biggrin: ...

[GlennChan]

1- Re: 2K resolution
To plagarize Charles Poynton's opinion... the 2048x1080 resolution has little technical merit to it, only marketing merit. They should adopt 1920x1080 resolution since there's economies of scale behind it (i.e. in projectors... Sony's 4k projector I believe is 4 times 1920x1080). And when you version for DVD (which practically every theatrical release will), it's simpler because you don't need to rescale or crop.

Too many formats is a bad thing.


XYZ is wasteful though, as cameras and film don't produce that wide a gamut.
From what I understand, digital cameras can produce XYZ gamut (ignoring color accuracy). Certainly in color grading, increasing saturation will push colors outside their normal gamut. Or you might introduce special effects with wide gamut colors... i.e. lasers in scifi movies.

2- As far as monitoring wide gamut...
Yes it will be challenging. However, there are ways to visualize colors that your monitor can't produce. You can for example lower saturation on everything so that you see the relationships between colors, have your system highlight all out-of-gamut colors as yellow or whatever, etc.

As future systems get wider gamut, the colors will (chances are) look better. So you could say that certain colors will shift hue as you get wider... but to me, that's not a big problem. Color accuracy is not that huge a deal because we generally don't notice it / filter it out (the real world isn't color accurate to begin with).

I would think that your real problem is with projectors that are smaller gamut that what the colorist saw. Bad gamut mapping can degrade image quality in ways you don't forsee (i.e. if you just let colors clip after you use a matrix to adjust the colors, you get shifts in luminance; this looks pretty bad).

In a similar vein, you could argue that going with wide gamut removes the WYSIWYG-ness of color grading.

3- On the other hand, you could argue that it's a better idea to stick to a fixed gamut. You have financial and technical limitations so you gotta pick your battles. For marketing and aesthetic reasons, resolution and contrast ratio make more meaningful impact on picture quality. Contrast ratio makes the biggest difference in image quality. Resolution is very marketable. 3-D might also be worthwhile (??).

Wide gamut isn't very marketable (difficult to show people, is not a simple concept, etc.) and doesn't make that big an impact on picture quality. In most feature film material, you just don't get a lot of extremely saturated colors that hit the boundaries of film gamut. And in de-saturated films like se7en, wide gamut doesn't do anything for you. If you consider that and the complexities of working in wide gamut (i.e. not being able to see out-of-gamut colors), then maybe it isn't really worth it to do wide gamut.

3b- I believe that even if your display device could do wide gamut, going with a smaller gamut would raise light output and let you increase contrast ratio. If you look at the original NTSC primaries, they defined a much larger gamut than current TVs. But it got abandoned in the race towards brighter output.

In theatres, you will have a pressure towards brighter output to get larger screen sizes, extend lamp life, (and I believe) or to get a higher contrast ratio.

Thus it might be reasonable to stick with a fixed gamut system (though you have to choose very carefully, not too wide or small).

That being said, I haven't really seen a wide gamut system in practice so I can't say about the image quality.

3c- In color grading there may be an art to making colors look very vibrant within a fixed gamut.

4- Compression efficiency:
I don't think this is a big deal. Just throw lots of bandwidth / storage at it... instead of introducing complexity or other problems trying to get size down.

5- Perhaps the digital cinema industry and the home theatre/video industry could sit down and have sort of a truce.
The digital cinema industry is trying to stay one step ahead, while the home theatre industry is trying to deliver theatre quality in the home. So you have things like DCI adopting 2k resolution so that they can market 2k resolution over 1080 (while 2048x1080 and 1920x1080 are practically the same, and might actually be the same because it's cheaper to post in 1920x1080). That's stupid.

And then in the home video side, they're adopting wide gamut color because the digital cinema folks are doing it. But the home video side will likely have a lot more trouble implementing it due to backwards compatibility and needing to apply color management (mapping out-of-gamut colors into displayable colors). Color management done poorly won't look very good (and it's likely this will happen because you need to keep costs down in consumer equipment).

[GlennChan]

Some pictures...

In the first attached image, I increased saturation (via chroma gain) to create colors outside rec709 gamut. These colors aren't color managed. Clipping occurs and this results in shifts in luminance (since monitors can't produce negative colors, or colors that are whiter than the monitor's white).

In the second attached image, rudimentary color management is applied (luma is kept constant). This brings out detail in the clipping areas... i.e. you see more folds in the red/magenta/pink shirt.

This is actually not a very good algorithm since the hue will shift. As well, the proportions of saturation are not maintained; saturation gets clipped. However, it does show that no color management looks crappy. And in consumer devices, this might happen to keep costs down / the engineers may not understand the specs fully so they let stuff like this happen (i.e. we currently have the rec.709/601 color inaccuracies, and the chroma bug in DVD players).

If home video went in its current direction (a fixed gamut), you can easily avoid the first situation. The colorist would be looking at image #1, and can do the appropriate color corrections to avoid it. But in a wide-gamut system, the colorist would see #2 but some home viewers would see #1. If the colorist isn't aware of what might happen to the image, then some home viewers see #1 (which just looks bad). And if the colorist is aware of what might happen to the image, he/she might do a grade like #2. Which defeats the point of wide gamut, since there really aren't that many wide gamut colors anymore.

[M Most]

1- Re: 2K resolution
And then in the home video side, they're adopting wide gamut color because the digital cinema folks are doing it. But the home video side will likely have a lot more trouble implementing it due to backwards compatibility and needing to apply color management (mapping out-of-gamut colors into displayable colors).

There's theory and there's current and foreseeable future reality. That reality is that the primary method of presentation in theaters now and likely for some time, certainly worldwide, is film. Current DI practice is to use the color gamut of print film as the target for all color management in the pipeline, including the video deliverables. Having a wider gamut available means little if the primary presentation format isn't digital cinema, because it won't really be utilized until that changes. Right now, color management is applied in all DI work that maps the print film target gamut to Rec 709 for HD video delivery. It will be some time before that changes.

[GlennChan]

Hi Michael,
Could you please clarify what you're saying? (I'm afraid I'm probably missing some context.) As I understand it, you're saying that:
Grading is done towards the gamut of print film (since most movies will be seen with film projection). Therefore, there is little point in specifying a gamut larger than print film (or little point in grading for it) until digital cinema is more common.

[M Most]

Hi Michael,
Could you please clarify what you're saying? (I'm afraid I'm probably missing some context.) As I understand it, you're saying that:
Grading is done towards the gamut of print film (since most movies will be seen with film projection). Therefore, there is little point in specifying a gamut larger than print film (or little point in grading for it) until digital cinema is more common.

Yes, that's generally what I'm saying, but with a caveat. In any post pipeline, you tailor it for the primary deliverable. That doesn't necessarily mean lowest common denominator or highest, it just means that you must decide on the primary delivery target. Right now, for all theatrical releases, that's film, even if the picture generates more revenue from its video distribution. In part this is due to studio demands and desires, but it is also very much due to archival considerations - and film separation masters are currently the archival standard. Another technical consideration is that while it's not particularly difficult to achieve a film color pallette for the video deliverable - 3D luts can do this quite well - it is far more difficult -almost impossible - to achieve a "video" color pallette on film, due to some serious saturation and brightness differences. Of course, in the future, all of this can and most likely will change, at least in terms of theatrical exhibition, as electronic projection becomes more widespread. For this reason, it is very desirable to have in place a long-term digital cinema standard that can accommodate a transition to wider color gamuts that might be possible with electronic capture and distribution, even if that has little practical use in today's cinema.

[Graeme Nattress]

Yup, we need to plan for wider gamuts, and that's why I think a tagged colour space is the way forwards, not mastering to XYZ....

Graeme

[Stephen Gentle]

Wow, that's one huge projector. How many RED 4K projectors do you think yo're going to be able to fit in that?

[M Most]

Yup, we need to plan for wider gamuts, and that's why I think a tagged colour space is the way forwards, not mastering to XYZ....

Graeme

Graeme, I hear you, but you really ought to get involved with the SMPTE DC28 committee if you want this to happen. Ultimately, there are going to be standards. They need voices that are expressing different opinions than the studio tech chiefs.

[Graeme Nattress]

Is it not already a done deal though? I mean people are already making DCI spec files and projectors? I'm very much of the KISS school, and whenever I read a SMPTE spec they're always hideously complex and that really hinders their real world implementation, like MXF or MPEG7 for instance.

I really hear what you're saying, I'm just not convinced they'd listen to a word I'd suggest, which is pretty much the idea I got when I head them all speak. You can prove me wrong though - might be best to continue this in email.

Graeme

[Mark L. Pederson]

Soon, DCI will be DOA.

[M Most]

Soon, DCI will be DOA.

And what evidence or information convinces you that this is the case?

[Craig W. Bickerstaff]

I've never been to a digi cinema so could anyone enlighten me as to how a widescreen movie is handled?

I'm talking about Cinemascope 2:35:1 by the way

[Mark L. Pederson]

And what evidence or information convinces you that this is the case?

No comment. Ask me again at IBC.

[Graeme Nattress]

Craig, I think they crop top and bottom.

Graeme

[Dennis Guskov]

That article mentions $12,651/screen. Did someone misplace a decimal point?

I agree, it does seem a bit high... $1,265.10 is better and more affordable :-)

[Michael Schrengohst]

http://www.tvtechnology.com/features/news/n_sony_cracks_the_4k.shtml

More like $95,000 - a lower power one is "only" $60,000 plus lens.....

So if the HDCAM is $100,000 and the RED is $20,000 ish....

Are we looking at a "REDjector" for around $20,000 ish as well???

Now all we need is a REDpopper for the popcorn and we are in business!!

[Nik Manning]

So I am pretty sure that if RED can come up with a 4K projector that is close in capabilities to Sony 4K projector and obviously projects 16:9 for under $30K they could break into that market and be profitable. I don't think that was the market you were targeting with your 4K projector but I don't need a 4K projector for my 80 inch screen in my apartment.

So I say set the challenge. 4K projector for theaters for under 30K by 2009! GO RED!!!
No I don't know anything about that market so don't listen to me. How many screens are there in usa anyway?

[DogSoldier]

I always thought a good business to start would be to act as a middle-man between the theatres, the distribution companies and the projector manufacturers.

Right now, the theatres don't want to spend money on a projector that will only save the distributors money. The distributors don't want to spend money to put equipment in someone else's theatre. The projector manufacturers aren't churning out tons of these things and thus lowering the cost because there's no one stepping up to the plate to invest in them.

So why not have a company that buys the projectors, installs them in the theatres for no charge (and will even keep up with the maintenance), then contracts with the distributors to put their movies on the screen cheaper than the cost of making copies on film in all the theaters where they have a projector.

It's a win-win-win.

And now Reds going to come out with a projector...

.... huh....

[Gavin Greenwalt]

The real enemy of independent distribution are those frakin sattelite dishes on the roof of every digital theater.

The cost of projecting an independent film digitally is like 10 times that of renting a theater for a 35mm projection. Why? Because they have to send it to their corporate headquarters 1,000 miles away who then beams it back to the projector. Either that or you have to pay a few $k for an "engineer" to hook up your HD-DVD player.

[GlennChan]

Right now, there are theatres you can rent with projectors hooked up to a DVD player (or HDCAM player or whatnot). The trick is still marketing your film to get people into those seats.

Quality may not be as high as very good digital projection (or 2K/4K), but I don't think technical quality is really the barrier.