RED 65mm

[Shawn Bannon]

RED 65mm sensor? Is RED considering this option at all? do any of RED lenses cover this area? I heard RED lenses are possibly converted still lenses which means its a possibility?

[chuck colburn]

You would need a medium format still lens to cover 65mm. Or an optically modified 35mm cine lens.

[Radoslav Karapetkov]

For Scarlet? Unlikely.

Probably for a future RED II or III.

I wonder if the current RED One body could house a 65mm-sized sensor?

Giving us the possibility to take advantage of the modular design and easier\cheaper upgrades in the future?

[Lexicon]

A RED powered IMAX camera would be awesome but probably not easy to sell since IMAX markets are limited.

[Shawn Bannon]

well maybe more like a VISTAVISION size sensor. The same size as a full frame 35mm/DSLR still camera. Do the RED lenses cover the full frame still area? roughly 36mmx24mm

[J. Bernard Vallon]

There was some speculation that the red zooms are rehoused sigma still lenses, in which case yes they probably do cover 135. Lots of folks are going the nikon road which would allow for the same. I think its a wonderful idea, maybe 5 years down the road.

[Evin Grant]

It's hard enough to pull focus at the current S35mm size.

[J. Bernard Vallon]

agreed. but just imagine how pretty it would be

[David Mullen ASC]

Just for everyone's info:
5-perf 65mm Todd-AO (2.20 : 1) is 52.48 x 23.01mm
8-perf 35mm / VistaVision (1.50 : 1) is 37.72 x 24.92mm

So a 35mm still camera lens would cover 37mm but might not cover 52mm across...

Now perhaps the quickest way to build this Super-RED would be to essentially double the Mysterium sensor horizontally and vertically, thus 48mm across... and you'd get a 8-to-9K Bayer RAW camera, but that's a lot of data. And you'd need medium format optics.

I guess, assuming the same size photosites and spacing, that a VistaVision size sensor, 37mm across, would be around 6K worth of horizontal photosites.

The Phantom 65 sensor, by the way, is 51.02 x 30.5mm (1.67 : 1) but 4096 x 2440. So you get 65mm depth of field for a 4K image, but there may be some heat & noise advantages is spreading those photosites over a larger area, and perhaps some MTF advantage from having a larger target area thus less image enlargement. Though you have to use 65mm optics.

[Shawn Bannon]

if one can pull focus on 15 perf 70mm IMAX... wether it is hard or easy isn't really of concern to many of us. we all know that the larger format sensors can look very spectacular. not for every movie but for some movies...

[David Mullen ASC]

if one can pull focus on 15 perf 70mm IMAX... wether it is hard or easy isn't really of concern to many of us.

Well, maybe it should be a concern... :wink:

[Harry Clark]

Interesting. I've wondered if this might be the next evolution at Red. Rather than try to make an 8K camera with the same Super-35 sized chip, make it with a Vista sized chip. I bet there would be a noise improvement by spreading the photosites more.
I've also wondered this: what if there was a larger chip but, rather than being higher res, addressed the photosites randomly at 4K? Would that help or hurt deBayering? Would it have a more "filmic" feel by mimicking the randomness of silver grain? Would the elimination of the fixed pattern solve anything?
Or is 4K so close already to the limits of human perception it does not matter much anyway.
Cheers,
Harry

[Chris Nuzzaco]

Here's an interesting though...

Why just one chip? I'm really? Why not two chips that make up one larger chip, I'm not an engineer, but perhaps this could help with data coming off the chips, etc..

Any thoughts?

[David Mullen ASC]

The pattern wouldn't truly be random because it would be fixed on the sensor -- it would just be more confusing to deconstruct into RGB than a Bayer pattern. Film grain is random because it's in a different position frame by frame.

Remember that the de-Bayering algorithm has to make sophisticated judgements as to the true color information of the surrounding photosites, otherwise all you are going to get is half resolution for green and quarter resolution for red and blue. So a chaotic RGB filter pattern isn't really helpful for that.

As for spreading 4K worth of photosites over a larger area like the Phantom 65 does, you'd have to weigh the advantages against the inconvenience of not being able to use 35mm cine lenses and the lower depth of field as well.

[David Mullen ASC]

Here's an interesting though...

Why just one chip? I'm really? Why not two chips that make up one larger chip, I'm not an engineer, but perhaps this could help with data coming off the chips, etc..

Any thoughts?

I'm not sure how the CMOS data is read off of the sensor but if it is line by line, let's say, then it shouldn't make a difference whether it is two smaller sensors side by side or one large sensor.

Now I suppose one idea would be to double the number of photosites so that half could be ND'd to capture more highlight information, whether that means doubling the size of the sensor to fit 8K and then create 4K from it -- or making the final resolution only 2K from 4K worth of photosites. But that idea would be tricky with a Bayer pattern. I guess you'd have to go to something similar to Kodak's new extra sensitivity RGB sensor pattern, which sacrifices color for speed.

[Chris Nuzzaco]

I've also wondered this: what if there was a larger chip but, rather than being higher res, addressed the photosites randomly at 4K? Would that help or hurt deBayering? Would it have a more "filmic" feel by mimicking the randomness of silver grain? Would the elimination of the fixed pattern solve anything?
Or is 4K so close already to the limits of human perception it does not matter much anyway.

Are you referring to the actual bayer filter pattern on the sensor? I don't think that's possible, if thats what you meant.

As for the larger photo sites, yes, a big help, but I've also seen that larger chips can have some vignetting issues because light should optimally strike each photo site directly on. Doing this is far more easily done with smaller chips like 2/3 and smaller. Those cameras and their lenses are designed to basically focus all of the light straight on to the sensor, visualize a narrow beam of light, thats basically what happens... The earliest Canon 35mm digital cameras had this issue vignetting issue, because 35mm lenses don't focus their light into a tight beam on the sensor, it goes from being straight in the center, and then gradually becomes angled at the edges.

SO...

If you had a really big chip, there is a risk that you will have not only vignetting around the edges, but also more noise since those photo sites would be underexposed.

I've heard that micro lenses over each photo site can help address this issue, but as for how they are designed, I would guess that the lenses over the center photo sites would be different than the lenses over the outer photo sites.

So complex!

[Chris Nuzzaco]

Now I suppose one idea would be to double the number of photosites so that half could be ND'd to capture more highlight information,...

This is currently being done in the still camera world. I can't remember what camera does this, but last I heard, its slow and complicated.

[Rodney Johnson]

A RED powered IMAX camera would be awesome but probably not easy to sell since IMAX markets are limited.

They are growing all the time, they just signed a deal with AMC theatres for 100 screens in the next 2-3 years. They will go from about 300 screens world wide to 400 with one this one deal. IMAX is going digital with the first projectors scheduled to be installed summer 2008.

[brandon herman]

Here's an interesting though...

Why just one chip? I'm really? Why not two chips that make up one larger chip, I'm not an engineer, but perhaps this could help with data coming off the chips, etc..

Any thoughts?



I thought this was done on the JVC HDV camera with the interchangeable lens...

I remember the first generation yielded weird split-screen problems, with slightly off exposures, WB, etc... They eventually fixed it, I think...

[David Mullen ASC]

I don't think the JVC had two chips stuck together, it was one chip where they read the charge off of one half then the other half. I think... could be wrong. I believe the problem was fixed with some software changes.

[Lachlan Ward]

I say stick with a s35mm size sensor! but use 2 and a prism. I want more latitude god damn it, to hell with 8k, more latitude, higher bit rate. 32bit with 32 stops of latitude. That would be awesomely radiculas. Great fun for the colorist. The only thing that could be a sticky wicket would be focus.

[Chris Nuzzaco]

I say stick with a s35mm size sensor! but use 2 and a prism. I want more latitude god damn it, to hell with 8k, more latitude, higher bit rate. 32bit with 32 stops of latitude. That would be awesomely radiculas.

I'm not sure you'll ever see 32 stops of latitude. Even film can't do that, and even if there were a format out there that could, you would need a much better monitor (which doesn't exist yet) in order to view that stuff. Add on top the need for a good understanding of viewing LUTs etc... HDR cinema would certainly be cool to play with though, assuming anyone ever makes it. I've messed around a little HDR photography.

I can certainly understand a desire for more latitude, but I think many people mistakingly believe more latitude = automatically better images with less effort. Not quite the case, really good film lighting can be pretty hyper realistic looking. Latitude simply makes the capture stage more flexible, but it wouldn't make a hard lit old womans face look any nicer.

Sorry if I busted your bubble! :bleh:

[David Mullen ASC]

A prism block with 35mm-sized sensors would not only be quite big & heavy, but you'd be limited by the flange depth in terms of what lenses you could use.

[Michael Brennan]

Im sure that some readers of this list think that recording 16bit RAW improves dynamic range.
Once the photo site has had its fill of electrons and is saturated that's it.

CMOS is not the best of sensors when it comes to dynamic range.

One thing electronic sensors do promise in the future is increased resolution.

We need a apple to fall on someone's head to radically improve CMOS or CCD dynamic range....



Mike Brennan

[Nathan Garofalos]

I dont know much about the Mysterium sensor, but if they are able to crop it down to a s16mm sized sensor, would it be really hard to just make it bigger to a 65mm sensor, or set 2 inside a camera, like one side by side. I am probably really really really REALLY off, but i know is more complicated that that, or maybe they could use mirrors or soemthign to collect the data from the lens reflect it into the prossesor. Maybe like in LOTR when they had pippen or i dont know what hobbit saving the kings son from being burned and they set up a mirror reflecting the fire so he can jump in an not get burned...

[David Mullen ASC]

I'm not sure which problem you're trying to solve with your description.

[Nathan Garofalos]

I was trying to think of a way to give a more of a dynamic range and more latitude. Like a 65mm sensor, or a way to give the equivalent to a 65mm sensor without making a whole new sensor.

[David Mullen ASC]

If you go for a two-sensor approach to increasing dynamic range, they have to be taking exactly the same image at the same time, just one exposed darker than the other, so side-by-side doesn't work. You'd either have to go with a prism block to split the light in two directions, or create a pattern in a single sensor that had 50% of the photosites filtered darker.

Yes, you could also try some sort of partial mirror trick to reflect some of the light to a second sensor at a right angle to the first, but again, you have the same problem as the prism block method -- you need more space and you'd have limitations on the flange depth of the lenses.

As for making a 65mm-sized sensor, that's not so hard... but then you need 65mm / medium-format optics to cover the sensor area.

[Chris Nuzzaco]

I've been thinking again (go figure!).

This probably isn't the "ideal" solution to getting more range, but one worth playing around with at least.

Many people are creating these little 35mm adapters using ground glass as a sort of rear projection screen. Might it not be possible to somehow split the image on the reverse side of the screen (the side the camera is on) and then have two cameras shooting the same scene?

I know its kinda clumsy and "guerrilla" style, but for lower budget shows shooting exteriors (heck, even naturalist cinematographers), this might be a cool solution. You can get the 35mm DOF, and at the same time, have a somewhat HDR image.

The only cons I can think of right off the bat are...

- Gen Locking, not all lower end HD cameras can do this.

- Editing could be a pain if you want to make use of dual exposures, and combining them together into a seamless image requires a mixture of compositing and colorist work (I could do this, but I'm sure not everyone else out there knows how).

- Light loss could be high due to the extra splitting optics.

- Twice as heavy rig (and all the complications that brings).

- CA issues with the two different cameras. The lenses need to be pretty darn close to perfect, or at least exhibit the same CA issues to the same degree.

On second thought, unless you do landscapes and super telephoto daytime shots of wildlife, this might not be the best idea out there. However, I'd be willing to play with one if I had the resources to make it.

[J. Bernard Vallon]

Many people are creating these little 35mm adapters using ground glass as a sort of rear projection screen. Might it not be possible to somehow split the image on the reverse side of the screen (the side the camera is on) and then have two cameras shooting the same scene?


Chris,
I don't think this would work in practice because of the image alignment. To get usable HDR footage you need to have EXACTLY the same image photographed twice. The parallax between your two cameras would probably be enough to throw every pixel out of place just enough to ruin your image.

I think if we want a system with a bigger sensor and less DOF, but still want RED ergonomics and compatibility with existing gear, we should be asking for the vistavision sensor with nikon lenses.

As for increasing dynamic range, what about what the fuji DSLRs do? They have photosites coupled with "micro" photosites that let in less light, and combined the data for a more accurate exposure with better highlights. It works really well, I've always been happy with the DR out of my s5

[Chris Nuzzaco]

You have an s5? How much range does that get?

[J. Bernard Vallon]

I've never formally tested it, but its the same exact sensor as the s3, which got me about 11~12 stops DR and it came out in 2004. 14bit a/d converter.

[Chris Nuzzaco]

Nice! You know I'm not far from you in northern VA.... Check your PM's.

[Timothy Carr]

For Scarlet? Unlikely.

Probably for a future RED II or III.

I wonder if the current RED One body could house a 65mm-sized sensor?

Giving us the possibility to take advantage of the modular design and easier\cheaper upgrades in the future?

Wow we have come a long way :D

[Pablo Moreno]

It's hard enough to pull focus at the current S35mm size.

You are right about that 100%.

[James Falco]

RED did mention way back a 645 sensor and had concept renders of it. I personally have been waiting for it eagerly as I shoot a lot still photography with medium format digital back. Not sure why anybody would be asking about it in the Scarlet section of the RED forum. It's like asking about aircrafts in the bicycle section.

[Jeff Kilgroe]

OMG.. ZOMBIE THREAD!!!

Die! Die, die!!!