Red Lenses

[Logan Schneider]

Jim,

 

I do look forward to seeing how the RED film lenses perform. I have been bugging the ARRI and Cooke guys for years to come out with a more affordable second tier in their lens products, and whether or not I find the RED lenses as a good fit for myself, I think that you are providing much needed competition at the owner/operator price point. The only other company that has stayed competitive has been Canon with their Super 16mm zooms, which seems to be piggybacked on their much larger ENG lens production.

 

"Gee, an entire lens set and camera, or one and a half Masterprimes..."

[Max Jacoby]

Max,

 

Congratulations on the project. No negative (no pun intended) from me on shooting 65mm film. What is the project, if I can ask?

 

Jim

Jim

 

The project is not set in stone yet, so I'd prefer not to talk about it for the moment. If I've learned anything from working in this industry it's that until the cameras start rolling, everything can change in a moment's notice. I think you know well how that is.

[Thomas James]

Of course alternate pixel patterns can indeed be simulated using conventional sensors. There is a resolution loss with this conversion but at least Red allows you to start out at higher resolutions so this loss is minimized. As an example if pixels are to be arranged to mimmick the rods and cones of the human eye we can easily use software to convert the 2500 lines of resolution of the Red camera into 500 concentric circles with each circle composed of 500 pixels. So now we are no longer capturing boxes but rather we are capturing truncated pyramids. In the past attempts to simulate the human retina have failed because with standard definition 480 lines of resolution only outputs 96 concentric circles which results in a severely distorted image.

[Paul Clements]

Probably that longer S4 zoom (to replace the 18-100m) that everyone is asking them for :P

 

Hi Max

 

Geoff Chappell from Cooke told Stephen and I that they were working on "stuff" but nothing imparticular with no plans to announce anything new. I asked him whether they were thinking of reworking a new version of the 25-250mm which he catagorically said no too.

 

He finished off by saying that Cooke were so snowed under with the production of their lenses that developing and producing new ones is tricky. Lens production at Cooke is booked up for the next few months they didn't even have an 18-100mm or 15-40mm on the stand because they go straight out of the factory to customers.

 

Cheers

 

Paul

[Stephen Williams]

Hi Max

15-40mm on the stand because they go straight out of the factory to customers.

 

Cheers

 

Paul

 

Hi Paul,

 

They had that zoom, it was an the camera by Les.

 

Stephen

[Paul Clements]

Hi Paul,

 

They had that zoom, it was an the camera by Les.

 

Stephen

 

Hi Stephen,

 

Yeah, thinking back I do recall seeing it. It was the 18-100mm Geoff said they couldn't bring wasn't it. The whole day is a bit hazy for me still! It was a long one! The only zoom in the display case was a HD lens - that's where I'm confusing things.

 

Paul

[John Sprung]

Unless you use concentric circles, your series of boxes can NEVER compete with film, which, uses celluloid, and silver halide crystals.

 

...Just like the human eye.

Film grain, and the rods and cones of our eyes, are arranged in random patterns. The only place this has an advantage over a regular rectangular array of photosites is when you have something in the shot that also has a regular pattern, like a chain link fence. I once looked into using a pseudo-random pattern, designed to be linked like jigsaw-puzzle tiles, to get rid of moire patterns. Of course, the pattern would have to have been standardized so we could exchange images between different pieces of equipment. Bottom line, it was too much work to solve too small a problem. The right answer is to make more and smaller photosites, and listen to Nyquist and Shannon.

 

BTW, small random detail will also alias against any sampling structure, be it regular or pseudo-random. For example, the leaves on distant trees may seem to be fluttering in the breeze, when in fact the air is still. But the windows of a distant high-rise would likely degrade more gracefully with pseudo-random sampling.

 

 

 

-- J.S.

[Thomas James]

A logarithmic polar arrangement of pixels would be a tremendous advantage. First of all we would have an omnidirectional arrangement of the pixels which would be an advantage as we would eliminate the jaggies along diagonal lines. Secound concentric circles spaced logarithmically would allow a space variance of the pixels. This would mimick the human visual system in that the most detail would be concentrated in the central fovea of the retina while less detail would be concentrated in the peripheral areas of vision. This allows an efficient compression system as computational resources would be dedicated to the areas of the picture that need the most resolution while not wasting computational resources on outlining boundaries which are not the focus of our attention. The result is a very natural looking image that seems to come to life. On the contrary with a rectangular array of photosites the image may look artificial and computer generated simply because this is not the way we see things with our eyes.

[Ruairi Robinson]

A logarithmic polar arrangement of pixels would be a tremendous advantage. First of all we would have an omnidirectional arrangement of the pixels which would be an advantage as we would eliminate the jaggies along diagonal lines. Secound concentric circles spaced logarithmically would allow a space variance of the pixels. This would mimick the human visual system in that the most detail would be concentrated in the central fovea of the retina while less detail would be concentrated in the peripheral areas of vision. This allows an efficient compression system as computational resources would be dedicated to the areas of the picture that need the most resolution while not wasting computational resources on outlining boundaries which are not the focus of our attention. The result is a very natural looking image that seems to come to life. On the contrary with a rectangular array of photosites the image may look artificial and computer generated simply because this is not the way we see things with our eyes.

 

Advantage for WHAT? Then you get jaggies on vertical and horizontal lines! So whats the benefit? Plus jaggies are not a such problem on high def and beyond ANYWAY so it's no advantage whatsoever - especially if it's sampled at a higher resolution and scaled down. Plus dude, our eyes may see less detail in peripheral vision, but if we look at a screen, we can look at any part of it, including the peripheral areas, so it needs as much detail in the corners too... since the screen is not locked our eyes. Even if the screen was on goggles attached to you head, you could still scan any part of it so it would need as much detail everywhere. And a group of people all watching a screen, are all going to be looking at slightly different things on the screen. This would only work if you put the audience's head in a vice, clamped their eyelids open, and stuck pins in their eyes to fix the direction in which they look to the exact centre of the screen.

 

This idea makes no sense, whatsoever.

[Emanuel A Guedes]

So Stephen... I have to put you on the spot. What was your opinion of the RED footage. Does it look like 2K as suggested by one poster? Or Video?

 

You have confirmed that you are not a fanboy so your opinion will carry some weight here.

Understood. Your reasons are well understood. Nevertheless, no one should think that all fanboys behave like fanboys. If "fanboy" will be or even is actually the right definition for that matter... To be a RED customer doesn't necessarily mean to conduct an opinion survey or whatever like a "fanboy".

 

Just for the record.

Edited September 11, 2007 by Emanuel A Guedes

[Damien Molineaux]

...

 

Jim, while you're here, is there any word on when overcranking will be implemented? I don't suppose you guys are considering having Lens Data System or Cooke iIntellegence lens system compatibility? It would be a nice option. I'm interested to see how the 18-50 does at it begins testing in the open market.

 

Hi Logan,

 

I guess Jim didn't see your question... Red camera and lenses use Cooke /i Technology. There's an article here :

http://www.studentfilmmakers.com/news/article_1372.shtml

 

With my regards,

Damien

Edited September 11, 2007 by Damien Molineaux

[Jim Jannard]

Hi Logan,

 

I guess Jim didn't see your question... Red camera and lenses use Cooke /i Technology. There's an article here :

http://www.studentfilmmakers.com/news/article_1372.shtml

 

With my regards,

Damien

 

Sorry, I didn't see your question. The RED camera has a provision for /i but it is not enabled yet. Our two RED zooms (18-50mm and 50-150mm) will be /i enabled.

 

Jim

[John Sprung]

This would mimick the human visual system in that the most detail would be concentrated in the central fovea ....

The fundamental error here is that we look *with* our eyes, not *at* our eyes. The image on the retina is upside down, but we don't, therefore, hold all our photographs upside down. The folks in the movie audience choose where to look with their foveas -- you can't impose an artificial fovea location on them.

 

So, we don't need or want to mimic the human visual system. What we want to do is mimic the portion of the real world that falls within the range of wavelengths, brightness, and resolution that the human visual system sees.

 

 

 

-- J.S.

[Thomas James]

Photographers impose their will on audiance all the time. Photographers choose the zoom the camera angle and the speed of their pans and tilts and they also choose which objects to focus. We also have frame rates shutter speeds exposure and white balance and compression that are under full control of the photographer.

 

And now the photographer can control the resolution of the camera whether it be 720p 1080p 2K or 4K. What I am proposing is that the photographer can gain an additional control over the resolution of the camera with the technology of space variant pixels. When a movie is going to projected on a big screen IMAX movie theatre it simply does not make sense to concentrate pixels in the area of the periphery of human vision where they will hardly be noticed but rather to concentrate the pixel density in the fovea or central areas of interest. This is not to say that the human visual system will be exactly replicated since this may produce too much of a bulls eye targeting effect and would require additional scanning with rapid eye movements to emulate the sacading of the human eye in order to avoid a tunnel vision effect. However as a general principle the technology of space variant pixel design is legitamite however it may have to be toned down a little bit to generate a more subtle effect. For example a true android machine vision system may be composed of 1000 concentric rings and the number of pixels per ring being an invariant number of 1000. An alternative hybrid system could be developed that includes both space variance of the pixels as well as numerical variancy which corresponds to each ring. Thus with some experimenting one can determine what would be the proper ratio.

 

Fortunately alternative space variant pixel arrangements can be utilized using conventional sensors using software emulaters. Of course there will be a resolution loss with such a conversion however at least Red allows you to start off with 12 million pixels which should be plenty enough resolution to produce a decent emulation which would e far superior to the low resolution logarithmic polar chips that have been produced so far.

 

Now what is the purpose of all of this? The purpose is to produce a more natural organic image and to increase compression efficiency.