2 K RAW on full size

[Sebastian Cramer]

Ok, if've went thru a bunch of threats here and what figured out is there is full size 4K or windowed 2K but no such thing as full size 2 K. From a practical production point of view, I think there is a very big interest in shooting full size, but with a lower resulution like 2K or even below that. For instance, currently I'm always shooting 35mm film and then telecine to PAL SD. So talking of 4K is a hell lot of oversampling, even if it's a Bayer single chip.
I guess there are a lot of big red fans out there (also people normally shooting with 35mm adapters like the Mini 35 P+S) who do anything to work with 35mm full size capturing but simply don't need the resulution, but go for higher fps, smaller files, longer recording etc. I also would be thrilled with the idea of simply having 2 different 2K Formats one in S35 and one in S16 and adjust format according to my needs in terms of DOF, focal length etc. But in the end it's still the same type of file: a 2K image.
If you wizzards at Red can invent magical focus assists, believe me you will make a great number of filmmakers extremly happy if you read out every other pixel of every other line (ok, guess it will not be as easy).
Thanx.

[Anders Holck]

As the Red codec experts has stated that you cant scale a RAW image, scaled 2k RAW is also impossible. So it has to be converted to RGB before the scale.
Scaled 2k RGB was once part of the plan, but now it isn't.
The option now is 1080p RGB which can be scaled from 4k up to 60fps.

Having said that 4k redcode RAW is pretty efficient compared to 1080p Redcode RGB, and is only 2x the datarate of DVCPRO HD at 1080p24.

[Jeff Kilgroe]

Yep, if you want the full 4K area, but in 2K just shoot 4K RAW and scale it in REDCINE. You can shoot the 4K sensor area and encode to 1080p RGB @ up to 60fps on camera. They may also have an option to shoot the 4K area and encode to 2K size in RGB on camera at some point -- I suppose it would make sense to offer that option. But as Anders said, the RGB isn't as efficient as the RAW codec. Also taking RAW from the camera gives you a lot more freedom to play with exposures, LUTs, etc.. in post rather than applying your LUT and gamma in camera and getting stuck with an RGB file to work with. Then again, we shouldn't turn our noses up at that either.. After all it's still wavelet compressed, 1080p 4:4:4.

[Alexander Nikishin]

I dont think a 2k 24fps 35 full frame is as important as a 2k 35 frame at 1-60fps redcode. That is the one thing I'm dying to be able to shoot.

Atleast then I don't have to worry about the crop factor and a wider lens set to match up. If I just had a 2k 35mm frame at 1-60fps I'd have a decent slow mo option which can later be uprezzed to 4k.

This same practice holds for an HVX, shoot 1080p, slow mo in 720p, then upconvert.

[Sebastian Cramer]

As the Red codec experts has stated that you cant scale a RAW image, scaled 2k RAW is also impossible.

Thanks Anders, when was that said, a 2K RAW readout on a 4K area is impossible? Must have missed that. I totally agree the 4K RAW is pretty small compared to RBG, but a 2K RAW would be 4 times smaller...

[esmilis]

As the Red codec experts has stated that you cant scale a RAW image, scaled 2k RAW is also impossible.

Thanks Anders, when was that said, a 2K RAW readout on a 4K area is impossible? Must have missed that. I totally agree the 4K RAW is pretty small compared to RBG, but a 2K RAW would be 4 times smaller...

I think it's in the principle of RAW to record raw sensor data, before debayering and other image processing. You can't scale this kind of thing

[Chris Kenny]

Perhaps it would be possible to sample only 2K worth of pixels (in an appropriate pattern) from the 4K area. This could conceivably give you a 2K RAW image with the 4K depth of field.

[tj williams]

Pardon me to jump in as I'm probably the least computer like on this board but for 2K show shot in 4K raw full sensor, wouldn't it be output to cineon in 2K?

[Thom Steinhoff]

Perhaps it would be possible to sample only 2K worth of pixels (in an appropriate pattern) from the 4K area. This could conceivably give you a 2K RAW image with the 4K depth of field.
Very late in the game for v1 but this sort of thing is a very interesting idea and worth a few experiments. That said, maybe Graeme already went down that path and abandoned it for one reason or another--but it is a good idea!

[Thom Steinhoff]

Perhaps it would be possible to sample only 2K worth of pixels (in an appropriate pattern) from the 4K area. This could conceivably give you a 2K RAW image with the 4K depth of field.
Very late in the game for v1 but this sort of thing is a very interesting idea and worth a few experiments. That said, maybe Graeme already went down that path and abandoned it for one reason or another--but it is a good idea!

[Anders Holck]

I think simply dropping pixels would be a very bad idea, leading to massive aliasing. It would be like doing a nearest neighbor scale in photoshop.

[Sebastian Cramer]

My Nikon D2 can aquire different formats on a single chip with no windows and no quality loss. All I'm saying, I don't think I'm the only one who would be extremly interested in recording 2K Redcode RAW in S35 format. I guess there a lot who'd love to do that.
Im just curious to get some feedback from Greame & Co if this is technically possible.

[Rob Lohman]

ready2roll: in RAW? Sounds like that's in RGB (where it is 'easy') instead of RAW.

[Gavin Greenwalt]

People are throwing around the wrong word: Impossible.

Technically a chip could capture 4kRAW -> Demosaic -> Scale -> Mosaic -> output 2k raw.

But that would require quite a mosaic process which might over tax the existing DSP.

Also technically a sub sampled 4k chip could output 2k raw. The problems I wouldn't imagine wouldn't be aliasing. It would be the same spatial sample ratio as 2k windowed. The problem would be sensitivity. You would lose ?2 stops? of sensitivity since the lens would be spreading out the available light across 4 times as much surface area. This last bit could be wrong though, I'm just speculating. Graeme would this be correct?

[esmilis]

Technically a chip could capture 4kRAW -> Demosaic -> Scale -> Mosaic -> output 2k raw.

That technicly wouldn't be raw, more like some weird Mosaiced new codec. The main principle behind raw, as far as i know, is to record the very bits coming from chip before any processing, and you can't scale that. So i still think it's impossible

[Chris Kenny]

It would be the same spatial sample ratio as 2k windowed. The problem would be sensitivity.

I would think you would have the same sensitivity as 2K windowed. Yes, the light is spread out 4x as much as using a 16mm-format lens to shoot 2K. But you're using a 35mm-format lens. A T2 35mm-format lens produces an image just as bright, at any given point, as the one a T2 16mm-format lens produces. The need to produce an image of the same brightness, but ~4 times the size, is why 35mm-format glass is so much larger and heavier than 16mm-format glass at a given T-stop.

[Tom Lowe]

It's really a bummer that even if you choose to shoot at 2K, you cannot use REDCODE, full frame, and overcranking at the same time.

There seems to be no option for Full Frame>Overcranked>RAW, regardless of resolution :( Unless I am mistaken, what this does is force anyone trying to shoot a feature with overcranking to go to the lowest demoninator -- 1080 RGB.

So you lose the resolution, you lose fullframe DOF, and you lose RAW. If this is the case, I think there are other cameras that might be options for feature shoots requiring overcranking.

[Sebastian Cramer]

The question is simply can you read out a reduced number of pixels on the full sized chip to get a lower resolution and still get Redcode RAW? P+S and Redrock are selling hundreds of image converters, because it seems that everyone wants to shoot 35mm format. And I'm totally sold on the benefits of Redcode RAW (so RGB isn't the choice here), but I don't think everone does require the full resolution of 4K. I personally would prefer slim data and storage etc. And it would be a major plus if the user can choose between 2K in S35 and 2K in S16 according to the needs of the shot (DOF, fps, focal length etc), but it still would be a constant data format for further processing.

[MikeCurtis]

Chris - what you're proposing is exactly what they are saying is problematic.

In the meantime, shoot 4K RAW, Redcine down to 2K/HD for a gorgeous image. If other options are possible, they'll make them available in due time.

In the meantime, it is still a better deal than anybody else.

-mike

[Dan Blanchett]

Bottom line, if you want to overcank a dozen or so shots (John-woo style) in a feature film that is destined for a 2K film out for projection, is it better to shoot at 1080p or 720p and punch in to try and match your sequence, or shoot at 24p 4K RAW and use Shake/Twixtor/? to approximate a decent slomo or ramp.

So 4K raw + software vs. lower res, less DOF/FOV + resize in post?

[Dan Blanchett]

...or rent a Phantom on those days if you can afford it. (I can't)

[Rob Lohman]

The question is simply can you read out a reduced number of pixels on the full sized chip to get a lower resolution and still get Redcode RAW?

Sure you can, but it will either reduce your field of view (ie, S16 lenses) or you will get aliasing and whatnot. There might be ways to do it, but let's first get the 4K camera out with the "simple" stuff!

[Sebastian Cramer]

Thanks for your answer, Rob, I'm not into cmos/bayer things at all. All I'm saying is a reduced resolution in Redcode RAW on the full size chip would be my faforite configuration. If there is a technical way of making this happen, I know you guys would go for it :-)