Slashdot Mirror
True Color in Real Time: The Challenge of Mobile Imaging
rocannon writes "A mobile phone with a color display and built-in camera - it's the quintessential info-imaging tool. But communicating accurate color at high-speed data rates and rendering color on displays that do not deliver even standard screen color rendition are still challenging. Kodak explains why photo and color science are as important as clock speeds and data rates in this expanding market."
Fleshtones are historically the biggest photography challenge. Glad to see that Kodak is leading the way to the pr0n of the future!
.... porn!
Have a look at 10 years old pictures or printouts. Especially those exhibited to light. Make that 10 months for printouts.
Digital images aren't distorted, they just don't fit our colour perception. (AFAIK, among others, more a logarithmic scale in contrast to the linear used in digital images)
"Between strong and weak, between rich and poor [...], it is freedom which oppresses and the law which sets free"
Most geeks don't care about design/looks or anything else poingting in that direction. Geeks want features before fashion. Why should the geek care why the little distorted image looks like a smeared out stamp, the point is that the geek can take the picture, that's it, end of story. I've seen loads of cases when people spend a fortune on their computers and graphicscard. Monitor? The cheapest, biggest.. i.e sampo, samtron or hyundai. Big time degradation (spell) in image quality. On top of that, go buy a no name GeForce card and IT's like watching a TV from the late '80. Put a SONY screen with a Matrox card next to a Sampo with a noname geforce card... Yikes! Oh, I'm not a geek, i'm a perfectionist (spell)
Girls are strange. They don't come with a man page.
-- Michael Mattsson
Not exactly... many films lean towards more green or red, and there are a lot more variables involved than the light source. Differences in developing can also make a very noticable difference in colour. Even so-called black and white images often have a green, blue or reddish cast to them. (I stare at photos all day at my job, professional and not.) The age and quality of the developing fluids, etc, all factor in. As someone who photographs a lot of artwork, I probably see this a bit more often than a lot of people. I usually have to take anywhere from 3-6 photos to get one that looks suitably like the original artwork. I often still have to digitally alter them once they're scanned, anyway. So as far as colour is concerned, I don't think the quality is any better between analog and digital, esp. for the average person (myself included) who doesn't do their own developing and doesn't have much more than a point-and-click knowledge of cameras.
I do agree with you that analog is still currently superior to digital, but that's mostly to do with the fact that the really impressive digital cameras are way beyond the price range of most people, and the less expensive ones leave enough to be desired that film still has the advantage.
My Webcomic: Asylum on 5th Street
Because otherwise Kodak's business would suck. With this magical message, they can make use of this (System and method for generating a universal palette) (and the other 13696 Kodak patents ?
So they solve data rates, they solve porcessor speed, they solve image quality - now I can really quickly download high quality images for the 10 minutes my battery lasts!
<fnord>OBEY</fnord>
Skin tones are hard to do because the visual cortex is wired to recognise humans and spots inconsistencies pretty quickly. I've been working with a friend who produces MMS applications for the Nokia 7650.
One of the applications was a slot machine, and it looked great! The colors are simple, and it came out fine. A lot of the content of the internet is still made up of text and icons, which don't have very stringent requirements on color.
We're still going to see a lot of money being put into palletizing for swapping photos of family and loved (in oh so many ways) ones.
Johns: Well, how does it look now? Riddick: Looks clear.
It's all more videophone sillyness. Very low res, no focus, bad optics, no imaging options, 14kbit on most lines outside of huge cities, what's the point? I can take and email much nicer photos just fine with iPhoto and a real camera. This video cell phone thing is yet another attempt to catch star eyed gadgeteers in a trap of mediocrity for the sake of modularity. "The essence of a thing in its purest and most concentrated form?" Concentrated, maybe. Pure? Not half.
This is almost as dumb as "wireless web." WML sucks, guys. It's even more confusing that the regular web's interface, and 40% of people don't understand that. Combine it with the size limit on WML pages (1400 bytes is the max through tmobile's gateway, and the goddamn XML headers take up 100. Hell, this post wouldn't sneak in under 1300 characters!) and you've got a confusing, bland interface with no real data. Hardly the killer app that's going to change the world.
Give me a cell phone with no dumb games, a nice address book, sizable buttons, a clean look with no precarious plastic, plug & play USB interface that works with mac/linux/palm/pocketpc, and a phone plan that doesn't try to put its tongue in my ear every time I do something interesting and you'll have solved the cell phone puzzle. I'll gladly give you my $40 per month.
Hey freaks: now you're ju
"Mobile phones have screens that suck, feeble processors and not enough bandwidth. So you need to be really clever to transmit a picture of someone that doesn't look weird. Give us lots of money and we might tell you how to do it".
Panurge has posted for the last time. Thanks for the positive moderations.
Recently, Kodak image scientists developed a way to meet these goals by supplementing the distribution of the input colors with a distribution of selected "important" colors [5]. In particular, they found they could supplement skin tones by appending image skin-tone patches generated from a statistical sampling of the skin color probability density function. A major advantage of this approach is that explicit skin detection, which can be error prone, is avoided.
Sounds like this will be great for photographs of white people!
autopr0n is like, down and stuff.
My wife has used a mobile webcam I built from a tablet computer, a USB microcam, using CDPD communications.
One of the biggest problems we found was battery power, always a problems with computers, but enhanced due to the power sucked by the webcam (lots of pixels to move to the computer at high bandwidth, even if they are just to be compressed using JPEG), and the power sucked by CDPD transmission. Of course, the cam was sending out an image every 5 minutes or so, and the camera was "always on."
I think the naysayers on the camera wireless phones are totally wrong. I don't expect everyone to purchase a camera phone, but I think a lot of people (especially young people and several business niches) will. It's really fun!
The uses of a camera phone do not intersect much with a high-quality megapixel digital camera used for "archival quality" pictures. PhoneCams will be used much more for quick little shots where quality matters little...a bunch of friends at a bar (which will totally change Mardi Gras!), to show a potential purchase while shopping, to show a map with directions, to see if you like the night club, or a "hey I'm in Vegas, look" call while travelling.
but when it's on a simple swab of paper, the only variable is the light source.
Yeah, obviously the color on a piece of paper is going to stay constant with constant light, but how do you propose to transfer that image without color loss in the 'analog' world? Obviously a lot of things are going to affect transfer from medium to medium. Probably more so then in the digital medium. After all, an image might look different on different monitors, but the data stays exactly the same.
Btw, this is kind of an annoyance of mine, what's' the deal with calling film cameras 'analog'? They don't use analog electrical signals, a more proper term would be "optical", "Chemical" or even "photographic". Chemical would probably be the best term, since photographic is kind of generic.
Anyway, electro-analog images are terrible at color correction, there's a reason people call NTSC (our analog TV standard) Never Twice the Same Color. And even if it did have good color correction, we are certainly not about to start using NTSC or some other analog image format with our digital cellphones!
autopr0n is like, down and stuff.
You know, they have software that does your spelling for you know.
But yeah, I know what you mean, there is a huge difference in image quality between various monitors and video cards. I used to have a really nice Acer monitor (nice in image quality, anyway, not in it's ability to not fall apart...). I put a cheap S3 video card in there (after they got their 3d act together) and the fucking image rippled at high resolutions. It was ridiculous!
autopr0n is like, down and stuff.
I heard on the news today that Ulrika Johnson was caught pleasuring herself with a mobile phone!
Not the first time she has taken advantage of an Ericsson!!
Digital images aren't distorted, they just don't fit our colour perception. (AFAIK, among others, more a logarithmic scale in contrast to the linear used in digital images)
Actually, nonlinear storage/transmission of colour intensity is quite common (aka gamma). TVs use it and most computer systems do too. For a very good description of this field try Charles Poynton's Colour and Gamma FAQs
You're right, geeks care more about features. And we all know, that to make the really big money, you need to capture the geek crowd, right?
Wrong.
The mass market for mobile telephony lies with the teenagers. Not only that, but fashion-aware teenagers. (For a significant part, this even means "females".)
Sorry to be so blunt, but your market segment is not significant when writing the requirements for a mass-market mobile product.
Out of the Palm based PDAs Sony has the best screen. Someone said here that geeks don't care about what it looks like as long as it can do it. Well I'm not buying this. What it looks like is VERY important. I wanted to by a new PDA but they still insist on 160x160 resolution with sucky colors. Why should I switch from my 2 year old handspring? More effort should be put into the displays everywhere. And not only what they look like but also how they perform battery-wise.
I hate the fact that you people don't salute me
Not likelyl. However if it was a BW image PRINTED on colour paper (developed via RA-4) then yes, you can get a colour tone to it
:)
There are other possibilities. A true BW image is silver based - what you see are fixed silver grains. You could have that printed selenium toned, which would give you a slightly purple print and leave behind Selenium Sulfide (i think...)... which is the most practical permanent print you can make. No way for that to be oxidized and removed... very very permanent.
Now you've also mentioned development of film- yes. This plays quite a roll but believe it or not film is developed (as in research) to be quite tolerant of processing mistakes. Yeah you can give film a colour cast thru improper development but more likely than not what you are seeing is a printing error. The Pro lab I used to work for charged $4.00 for a 4x6 that was custom printed. Yes you bet that skin tone was dead on- a 4x6 machine print cost $0.46. I could make 10 of those for the same cost, and probably get to the right tone, but then you are limited by machine buttons (which was I believe 15% increments, or about 10cc (colour correction) filter levels).
Most all developing fluids are seasoned (you refer to them as aged). Thats out of necessity- it's far cheaper to buy and replenish than to mix anew, and it's also more reliable from a standpoint of quality. Occasionally you have to dump a tank and start over but that's a very traumatic experience for all involved (and usually involves running lots of seasoner (exposed but unimportant film/paper) to bring the Bromide levels back to normal).
The biggest problem in photo is metamerism from your description- how can you get a system with 3 dyes to mimic what you see in artwork where the spectrals are all over the place.... even with extremely careful colour management you are going to suffer. I think there is a place that uses 10 inks to reproduce artwork (might be you?)... it's difficult, expensive, laborious but gives great results. It's all about covering the spectrum... and 3 is good, but 10 is much better
Anyway, I've rambled quite far from BW colour casts.... post processing is most likely responsible for it.
Damn. U.S. Patent 5,664,080 is a patent on the dithering that paint programs have been doing for as long as I can remember. The first claim covers reserving some entries for the GUI, finding a set of colors that best represent the original image, and dithering the image using a repeating halftone pattern. Most paint programs that I have used use this algorithm.
Will I retire or break 10K?
I'll tell you why they'd say that right now -- because that's what their company does. They make money selling imaging devices and technology. Obviously they need to look out for and push the market for what their company sells.
I'm sure the RIAA could tell us why Digital Right Management is more important than clockspeeds and data rates -- it doesn't mean it's true :P
It seems to me this "feature", just like text messaging, is being pushed by a culture that just doesn't understand Americans. Just because SMS is all the rage in Norway and the Japan has the highest per capita camera ownership doesn't mean some asshole American like myself simply doesn't want to have just a phone to, you know, fucking talk to people.
SMS is great for the same reason email is great - it's asynchronus. I *could* leave a voicemail message asking a friend when/if we're meeting for dinner, but text messaging is far simpler. Most of the time, we're either working or in some other setting where the phone is turned off, so you can't "just talk to people" very easily.
Dithering, color quantization, and palette choice have been known and in use for decades, and back then, desktop computers were less powerful than phones are today. Kodak is a late-comer to the area. I don't see anything particularly new or "enabling" in what the Kodak folks are writing about.
Don't forget that broadcast TV is analog. PAL, NTSC etc. are atrocious for accurate colour. When we want accurate colour, we adjust gamma etc. and match colour profiles end to end or use systems like Pantone. We use calibrated light sources. We use carefully selected emulsions and develop them carefully.
In this application we don't want accurate colour; we want colour that's inoffensive to human visual perception, i.e. renders skin tones decently, and no banding or obvious posterisation. You don't care if your car's colour -- the exact shade of blue -- is not accurately depicted on your friend's Palm Pilot.
Unlimited growth == Cancer.