New Kodak sensor?

[David Mullen ASC]

Don't really know what to think but it sounds intriguing... remember that the Bayer filter idea came from Kodak:

AP news article:

Filter Pledges Crisp Photos in Low Light

Jun 14, 12:12 AM (ET)

By BEN DOBBIN
ROCHESTER, N.Y. (AP) - A year from now, capturing a crisp, clear image of a candlelit birthday party could be a piece of cake - even with a camera phone. Eastman Kodak Co. (EK) said Thursday it has developed a color-filter technology that at least doubles the sensitivity to light of the image sensor in every digital camera, enabling shutterbugs to take better pictures in poor light.
"Low light can mean trying to get a good image indoors of your kid blowing out the birthday candles. It can mean you want to take a photograph on a street corner in Paris at midnight," said Chris McNiffe, general manager of the photography company's image sensor business. "We're talking about a 2-to-4-times improvement in (light) sensitivity."
Analyst Chris Chute doesn't doubt that the new filter system, intended to supplant an industry-standard filter pattern designed by Kodak scientist Bryce Bayer in 1976, represents a breakthrough in boosting photo quality - especially when light conditions are not ideal.
"It's often the most simple concepts that can have the most profound impact," said Chute of IDC, a market research firm near Boston. "This could be revolutionary in terms of just changing that very simple filter on top of the sensor and basically allowing companies to use it in all different kinds of cameras."
Kodak expects to provide samples of its new technology to a variety of camera manufacturers in the first quarter of 2008. The technology is likely to be incorporated first in mass-market point-and-shoot cameras and camera-equipped mobile phones beginning sometime next year.
"Typically new features like this would be more likely to show up in high-end products and then trickle down," said analyst Steve Hoffenberg of Lyra Research Inc. "But I think the biggest potential benefit of this may come in the camera phone environment. Camera phones are using smaller sensors to begin with and smaller sensors generally mean smaller pixels, which means lower sensitivity."
When the shutter opens on a digital camera, an image is projected onto the sensor, which converts light into an electric charge. Most sensors use the Bayer mask: Half of the millions of cells on a checkerboard grid are filtered to collect green light and a quarter each are filtered to let through red and blue light. A computer chip then reconstructs a full color signal for each pixel in the final image.
The new method, which has been under development for more than five years, adds "panchromatic" cells that are sensitive to all wavelengths of visible light and collect a larger amount of light striking the sensor. Tailoring software algorithms to this unique new pattern enables faster shutter speeds, which reduces blurring when capturing a moving subject, McNiffe said.

[Steven M. Bailey]

If this could be used with cameras like the hv20 to improve low light shooting, it may prove to be a real improvement on the consumer/indy- circuit. May also increase the feasibility of using slower, cheaper glass. Not necessarily good, but possible.

Thanks for the info, cool. :matrix:

[Chris Gearhart]

Very interesting. I wonder if this would have any snr downside, or if it's the panacea it appears. Worth keeping an eye on.

The fact that it seems to be targetted to mass market first seems interesting--like I wonder if its the quality--or maybe they're just out to make a buck.

[David Mullen ASC]

It may have artifacts that makes it more useful to lower-rez consumer photography, I don't know, or maybe it's too early, too difficult for now to adapt to sensors with a huge number of sensor sites, like a 12MP sensor. The description seems to suggest more that the Bayer filter can be eliminated (since the sensor sites are already panchromatic, otherwise RGB filtering wouldn't work) to gain more exposure and some other method is used to capture/reconstruct the color information other than RGB filters.

[Bruce Allen]

I think they are just talking about an array similar to a Bayer sensor but with some clear pixels mixed in every now and then?

Makes sense and if anything it might increase your SNR and dynamic range - because while some of your pixels would have color filters on them (which would darken things a bit), the clear pixels wouldn't - so they should be able to pick up extra detail in the shadows. Clipping may or may not be a problem - once the "panchromatic" (eg clear-filtered) sensors clipped, they might be able to still pick up color info from the (non-clipped, color-filtered) sensors around them... with an attendant loss of spatial resolution of course, but I might be okay to trade that for slight increase in dynamic range.

Bruce Allen
www.boacinema.com

[Bruce Allen]

Summary to the above: they are trading off color spatial resolution for sensitivity and dynamic range. Kinda makes sense, I think, because luma is the most important - if you are in low light and have a clean luma signal, then you can apply pretty heavy noise reduction to the color channels and people won't notice. The sprinkling of of clear pixels should help black and white photos a tad too...

Of course, if they'd only sprinkle a couple of darker pixels in there as well (dark green, dark blue, dark red, maybe even a non-colored dot of neutral density), then you'd get even better dynamic range... even worse sensitivity and spatial resolution, though!

Bruce Allen
www.boacinema.com

[Jeremy Hughes]

...They don't say anything about the colour quality... eerie...

[number6]

Great thread David, and it's good to read all the debate about what it all means. Helps keep someone like me from getting too excited about what sounds like an unbelievable sensor improvement.

But I'm wondering, if this really DOES pan out to be what Kodak suggests, how will it affect lighting a scene? Does the camera operator become the lighting expert, so to speak?

edit: I'm speaking from a low budget film pov... Anything to save a buck!

[Bruce Allen]

number6, manufacturers have used different sensor arrangements quite often - complementary color in some consumer camcorders, and in still cameras, Sony's "Red Green Blue Emerald" (or cyan...) arrangement, Fuji's high-dynamic range SuperCCD SR, etc.

It doesn't really change the roles on set any more than whether you're shooting with a 1-CCD or 3-CCD camera... the camera operator is still the camera operator.

Bruce Allen
www.boacinema.com

[Tom Lowe]

Visions of the candle-lit scenes in Barry Lyndon are dancing through my head. :)

[Axel Mertes]

It may have artifacts that makes it more useful to lower-rez consumer photography, I don't know, or maybe it's too early, too difficult for now to adapt to sensors with a huge number of sensor sites, like a 12MP sensor. The description seems to suggest more that the Bayer filter can be eliminated (since the sensor sites are already panchromatic, otherwise RGB filtering wouldn't work) to gain more exposure and some other method is used to capture/reconstruct the color information other than RGB filters.

Reminds me of typical HDR imaging attempts. Instead of using a BGRG or the like pattern there are already sensors that have a BG1RG2 pattern (where G1 and G2 are different in density and therefor allow different exposure, thus leading to an HDR result on our most important green channel).

When you extend that idea to a say R1G1B1R2G2B2 or the like pattern you could have dark RGB and bright RGB pixels. You extend your exposure time and generate some kind of HDR image from reconstructed data. Further I'd expect the use of microlenses or the like to enhance light sensitivity, because the above would usually help on longer exposures, not shorter ones. They claim to increase sensitivity, so they need to gain more light at the same exposure times. Could be simply white pixels (no color filter at all) or kind of advanced microlenses.

Anyhow, sensor sensitivity increases 2 times about every two years.

Many different patterns have been tried in the past, even hexagons etc. (which is a bit complex for the sensor design).

But interesting to rethink the Bayer approach here :)

Imagine: Exchanging the micro lense Bayer pattern for the chip is said to enhance sensitivity here. Thats definetly good news, but likely to have some new flaws for decoded images. I expect to see deeper color then, but potentially less real resolution, somewhat bloomy images. They can't do miracles, after all...

Cheers,
Axel

[number6]

number6, manufacturers have used different sensor arrangements quite often - complementary color in some consumer camcorders, and in still cameras, Sony's "Red Green Blue Emerald" (or cyan...) arrangement, Fuji's high-dynamic range SuperCCD SR, etc.

It doesn't really change the roles on set any more than whether you're shooting with a 1-CCD or 3-CCD camera... the camera operator is still the camera operator.

Bruce Allen
www.boacinema.com

Thanks Bruce. I was hoping this new technology would change the amount of light needed. Then we low/no budget guys could use Lowes light fixtures with those little squiggly looking fluorescent bulbs from China for additional lighting.

[Bruce Allen]

Thanks Bruce. I was hoping this new technology would change the amount of light needed. Then we low/no budget guys could use Lowes light fixtures with those little squiggly looking fluorescent bulbs from China for additional lighting.

Or you could use a HV20 and have money for real lights? That was the option I chose, given my lack of cash. I am combining that with trying to write great music video treatments so that people will like my ideas enough to want to give me a budget so that I can rent a Red.

Seriously, lights are cool and you learn a lot using them. Two weekends ago I learned that the ball joint thing on the back of rented Kino Flo Diva-Lites sometimes has little spiky pieces of black metal in it that embed themselves in your finger, it you run your finger over the wrong part, so be careful!

Hmm... yeah, a higher sensitivity sensor would be nice...

Bruce Allen
www.boacinema.com

[number6]

Or you could use a HV20 and have money for real lights? That was the option I chose, given my lack of cash. I am combining that with trying to write great music video treatments so that people will like my ideas enough to want to give me a budget so that I can rent a Red.

Seriously, lights are cool and you learn a lot using them. Two weekends ago I learned that the ball joint thing on the back of rented Kino Flo Diva-Lites sometimes has little spiky pieces of black metal in it that embed themselves in your finger, it you run your finger over the wrong part, so be careful!

Hmm... yeah, a higher sensitivity sensor would be nice...

Bruce Allen
www.boacinema.com

Bruce, Before they came out, I was considering the Panny HVX??? (whatever), but when I learned they were uprezzed 960X something, something... I decided not to go there at that time. Then comes the RED. I stalked the RED on this forum for a few months, and then when I could see a financial way to oun one, I sprang.

Granted, I could afford more varied movie-making equipment if I had spent less on the cameras (later I reserved a second RED). But my theory is that my first 2, maybe three movies could be cash positive due to the stories themselves. My strength is making stuff up, but my weakness is trying to know something about everything. So I'm hoping that my strength and my weakness will get me a couple of movies made on the cheap, and that will get me an in with the big boys who will then get my better (but bigger budget) ideas made by professionals. I do not aspire to be a director or actor or technical person on a movie. I just want the stroke to get what I have to say, made.

But Bruce, back to why I want the RED cameras to make my leetle frienz (movies), I figure if I can get my scenes recorded in the highest possible amount of data, should I get into a festival here or there and should someone like the story, a distributor will come nearer picking it up if they have more to work with digitally to bring my work up to movie theater standards. If not, then I've still got more to work with on my oun to provide a better product for whatever outlet I find.

I'm just thinking it is easier to throw away data than it is to invent new. Although with the technology getting where it is, I could be wrong about that.

[Bruce Allen]

But Bruce, back to why I want the RED cameras to make my leetle frienz (movies), I figure if I can get my scenes recorded in the highest possible amount of data, should I get into a festival here or there and should someone like the story, a distributor will come nearer picking it up if they have more to work with digitally to bring my work up to movie theater standards.

Number6, if you want to record your scenes "in the highest possible amount of data", in many cases, a HV20 with a lighting kit will record more usable and beautiful data than a Red without one!

Of course the #1 way to collect quality data is to get a good DP and art department and give them enough time to work?

Bruce Allen
www.boacinema.com

[number6]

Number6, if you want to record your scenes "in the highest possible amount of data", in many cases, a HV20 with a lighting kit will record more usable and beautiful data than a Red without one!

Of course the #1 way to collect quality data is to get a good DP and art department and give them enough time to work?

Bruce Allen
www.boacinema.com

Bruce, I've read many of your posts and from that know that you are many multiples over my head in knowledge, so I won't dispute your statements. I suspect it's like many of the sound posts on this forum and others that say "It's better to have an inferior mic placed closely than use a superior mic at great distance."

But I've already put doun my deposits on the RED, and with the information Jin Jannard put out today, should be able to muster the "geetus" when my numbers is up...(ooooooohhhh! that sounds spooky.) Anyway, I'm in no hurry to get my cameras because I can't start full-out on the projects I've got in mind for now anyway. It's a health issue thing (not my oun) and I am just guessing that I am pretty well nailed to the floor for that amount of time.

My (non) scripts are not necessarily time-stamped. One is slightly in the future and one, though current, will remain current for a number of years considering the current world situation. Also, a possible third one will never be dated... only the dialogue will change a little.

I probably need this time before beginning... imagine getting to pick the brains of people like David, Jeff Kilgore, Evin, yourself, and hundreds of others on this forum for a year. I won't be an expert but I may know enough to get by. Sure, I would like to hire a good DP and even a Director to translate what I envision onto HD platters (or maybe flash, by camera time). But Bruce I HAVEN'T NOW, NOR WILL I HAVE IN THE IMMEDIATE FUTURE, THE MONEY TO MAKE THOSE HIRES!

Sorry Bruce, I wasn't shouting at you... I was just venting my frustrations.

On the other hand, if I can pull this off with limited funds, even partially... that would be a poot!...er, hoot.