Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Can 10Bit LOG DPX/Cin files from Red be Color Corrected in FCSP2?

Would love to know how you do a 3x3 matrix properly on Log data. You add logs to multiply the equivalent linear values.

Graeme
 
Indeed, I'm not saying it's the right thing to do. I'm just saying log coding could be done. The log data would need to be converted to 32-bit float every time a maths operation was done on it, then the result re-coded as log. A trade-off between processing and storage requirements I guess.

But obviously I don't know the details of how REDCODE works, or what in-camera processing is done. So naturally I have to defer to you Graeme.
 
I was commenting on Praxis comment that top end DI can be done with the math in Log.

Graeme
 
Yes, I suspect what actually happens is that they can read and write log-coded files, and the system makes the conversion to linear and back invisible to the user. But even Color does that, and most high-end folks wouldn't class Color as a high-end DI system.
 
Well, mathematically speaking I'd hope they'd completely linearize the data before working on in, for matrices and the like :-)

Graeme
 
Anyone who invests in DCI spec encoding equipment now is making a HUGE mistake. Lots of things are going to change - faster than folks think.

Laugh at me if you must, but I assure you, when you see the "tipping point" of digital projection in theaters worldwide - they won't be projecting files encoded in DCI spec -
 
Laugh at me if you must, but I assure you, when you see the "tipping point" of digital projection in theaters worldwide - they won't be projecting files encoded in DCI spec -
Click to expand...
What do you mean by that?

I might expect the studios to enforce the security requirements of DCI to try to avoid piracy.
 
Sure, but it needs to get decoded to project, and then you've got a great image you can copy in some way. It just stops access to the 2k or 4k digital files (unless they get hacked just like HD DVD did) and then they're screwed anyway.

Graeme
 
Graeme Nattress said:
Sure, but it needs to get decoded to project, and then you've got a great image you can copy in some way. It just stops access to the 2k or 4k digital files (unless they get hacked just like HD DVD did) and then they're screwed anyway.

Graeme
Click to expand...

Yup. If it can be "decoded", it can be copied. And yeah, those pesky Pirates are good at hacking. The answer lies in new techonology. And DCI is not it.

As soon as start to think about the fact that you can put an entire movie, in 4K on a drive the size of an ipod - it is possible to start thinking about other ways to secure that data - it is possible to create a "media device" that self destructs after one decode - a "disposable, one-time play movie cartridge", etc.

I don't know what it'll be - but it's not the current DCI spec - and I'd lay good money down money down that SEVERAL companies are working hard in their labs to come up with a better answer ... and we will see it within a year or two.
 
Hi graeme,

I've seen that 4K could be compress at 27MB/sec. I just want to know how you could jump to 350MB/sec and up for 4K uncompressed to 27 and still have the same possibility of color grading and compositing manipulation? could you tell us how the compression is done, and do you recommend it for feature film work?
 
Have you seen the movie Crossing the Line - was shot at 24MB/s on an older variant of the wavelet based REDCODE RAW.

We're still tweaking compression details and data rate - I think there's more quality to eek out, and, yes, I'd recommend for feature work.

Graeme
 
offhollywood said:
Yup. If it can be "decoded", it can be copied. And yeah, those pesky Pirates are good at hacking. The answer lies in new techonology. And DCI is not it.
Click to expand...

You can believe whatever you want. But the fact is that the DCI spec DCP is here, it works, it's being used every day in many venues, and it's a required deliverable for essentially every studio distributed picture. So you can poo-poo it if you want, but it is something that is already real and supported, and if you want to fulfill delivery requirements, you have to deal with it.
 
offhollywood said:
Yup. If it can be "decoded", it can be copied. And yeah, those pesky Pirates are good at hacking. The answer lies in new techonology. And DCI is not it.

As soon as start to think about the fact that you can put an entire movie, in 4K on a drive the size of an ipod - it is possible to start thinking about other ways to secure that data - it is possible to create a "media device" that self destructs after one decode - a "disposable, one-time play movie cartridge", etc.

I don't know what it'll be - but it's not the current DCI spec - and I'd lay good money down money down that SEVERAL companies are working hard in their labs to come up with a better answer ... and we will see it within a year or two.
Click to expand...

The amount of time,investment, and science that has gone into the DCI spec is obviously something most indies know nothing about. Yes your local cinema will be able to rent out their digital cinema with a blu-ray disc player, no it will not look anywhere near as good as a DCI package. As far as where this is all going, the main drive behind DCI isn't to be sending around hard drive packs its to be able to distribute via sattelite and dynamically arrange screens based on revenue estimates. There are estimates out here that this could increase revenue by 20 % at the box. In any case, for theatrical disto DCI is it from the studios.

As far as those ingenius companies ? They're off making really good 3D distribution a reality.
 
Jeff is on the money, we've been looking into this for a local film festival and running a feed into the new digital projection systems is not a trivial task, assuming you bring along your own playout device.
In the past I've done this using the in house video projector (the one used for graphics, it just made the grade) and patching into the theatre audio system.
One of the drivers behind the shift to digital projection is to stop "piracy", by the cinema itself. The distributors demand so much per bum on seat but it's pretty hard for them to know the actual numbers and how many times a print is screened. The financial model the distirbutors use is pretty much if the movies a hit they win, if it tanks the cinemas loose.
 
Graeme Nattress said:
Have you seen the movie Crossing the Line - was shot at 24MB/s on an older variant of the wavelet based REDCODE RAW.

We're still tweaking compression details and data rate - I think there's more quality to eek out, and, yes, I'd recommend for feature work.

Graeme
Click to expand...

Yes I see the PJ movie...I just can't believe how it is possible to get lossless compression and divide the bandwith by 10. Codec is lossless? Do we have to render the codec to view the footage properly? If they is any white paper on this codec, I'd be glad to read it. So basicly, we can do 4K with redcode raw at 24MB/sec, witch make 4K avaible for anybody at home??? Where is the catch!!!!
 
Mat@imageWork said:
Hi graeme,

I've seen that 4K could be compress at 27MB/sec. I just want to know how you could jump to 350MB/sec and up for 4K uncompressed to 27 and still have the same possibility of color grading and compositing manipulation? could you tell us how the compression is done, and do you recommend it for feature film work?
Click to expand...

27MB/sec is RAW so really you should be looking at something on the order of 90MB/sec RGB (3.75MB/frame) if you want to evaluate image quality on your own. I put together an overview for the July/August Showreel Magazine which demonstrates what you should expect from wavelett compression.

But if you want to *really* know how well it works. Go to dalsa's webpage. Download a 4k image. Load up Photoshop and save out as a 16bit Jpeg 2000 @ around 5MB a frame (Redcode is 12bit, so I took the 3.75MB a frame and multiplied by 1.33 for 16bit since there is no 12bit Jpeg2k in photoshop). That will be a very good indication of the quality of REDCode. Or at least put you in the ball park of what you should expect.
 
cinematography.net has a comparison of uncompressed versus redcode... though redcode is being worked on and has likely improved by now. Plus the camera may be lower noise by now, which helps compression.

http://www.cinematography.net/Red/comp-matrix.html

2- Redcode is not mathematically lossless.

It is arguably "visually lossless". Though that phrase gets thrown around a lot by marketing folks. For example, I doubt Prores is visually lossless since there are situations where 4:2:2 codecs can never by visually lossless; but it IS visually lossless if you ingest and output 4:2:2 material.

Anyways, the Redcode RAW compression on the CML site looks excellent... for that particular image, it is visually lossless IMO (I can't tell the difference at 200% zoom) and looks better than uncompressed 4:2:2.

3-
Wavelet compression makes the image a teeny tiny bit softer. That's the catch.
Click to expand...
You don't really see this happen with the Redcode images on the CML site. Though there are errors in the little green holes in the wastebasket thing the girls are holding. But these errors are very subtle... in practice it seems like they would be insignificant.
 
REDCODE is not lossless. It's very good looking, but it's not lossless. You can't get lossless codecs that run much better than 2.5:1.

Wavelets do not make the image soft. If you over-compress wavelets, you can get some softening as an artifact. Be careful you don't imply wavelets do "things" to the image. Wavelet compression can also be totally lossless, but the files are too big and slow for that to be of use for us.

Graeme
 
Graeme Nattress said:
Wavelets do not make the image soft. If you over-compress wavelets, you can get some softening as an artifact.
Click to expand...

Fair enough. He was asking for a 'catch', though, and it seemed the most obvious - is there a clearer artifact of wavelet (over)compression than softening?
 
You must log in or register to reply here.
Back
Top