[NaClH2O]

Interesting article from the "underexposed" blog. Only B&W now and prolly more than several years from commercial application, but the only thing I miss from film is the higher dynamic range.
Panasonic sensor tackles key photo problem--dynamic range | Underexposed - CNET News.com

NaCl(I'm sure that more than panasonic are working on this)H2O

[AVANT]

More and more manufacturers are already doing that.

Even Pentax in the K20D with the dynamic range boost.

I understand it's not exactly the same thing (range boost is still based on 1 photograph), but having 3 pictures taken at rapid successions will be difficult for hand held cameras as we the photos would only merge well with very high shutterspeeds. Merging 3 2" exposures handheld? I think not.

For us camera guys, increasing native sensor dynamic range is still the best way to go.

[KungPOW]

Neat find!

Ok, thinking outside the box here (boardroom speak for "I am clueless, but gotta say somthing")

I would love to see a high res sensor that could approach the dynamic range of film. I could imagine a sensor that instead of having colour receptors for green, red, and blue, just sensed brightness. The red sensor could be replaced with a -2 EV sensor, the green with a 0 EV sensor, and the blue with a +2 EV sensor. Or whatever +/- combo to get the best results. One shot would be taken, and the cameras software would mix the result into HDR artwork.

I could accept that it would be a Black and White image. In my film days B/W was 90% of what i did.

Gasoline and matches will be provided at the door for those who wish to flame me. Just not about how I spell "colour", I'm sensitive about that.

Eric.

[falconeye]

the only thing I miss from film is the higher dynamic range.

Not sure how much future you mean. Right now, sensors aren't digital. Confused?

Well, they are digital as far as they have pixels. But each individual sensor cell is analogue, converting light into electric current or a charge which then gets converted to digital - the AD conversion.

Some day physicists will have solved the problem to actually count the photons in mass produced sensors, which then are digital. It is already done in high energy physics experiments. Once its done, there will be no such thing as an overexposed image anymore and you can get as much dynamic range as you like.

Even before, if every cell gets its own AD converter, you can capture as much light per cell as you like without overexposing. All those HDR tricks are intermediate tricks until then.

When? I don't know. Can somebody please scan the patent database?

[NaClH2O]

Not sure how much future you mean. Right now, sensors aren't digital. Confused?

Well, they are digital as far as they have pixels. But each individual sensor cell is analogue, converting light into electric current or a charge which then gets converted to digital - the AD conversion.

Some day physicists will have solved the problem to actually count the photons in mass produced sensors, which then are digital. It is already done in high energy physics experiments. Once its done, there will be no such thing as an overexposed image anymore and you can get as much dynamic range as you like.

Even before, if every cell gets its own AD converter, you can capture as much light per cell as you like without overexposing. All those HDR tricks are intermediate tricks until then.

When? I don't know. Can somebody please scan the patent database?

If you want to get real technical about it, sensors can NEVER be truely digital as light is primarily an analog mediun (and I don't want to get involved in a quantum/waveform argument) You need an analog sensor to capture analog data. However all of that nit picking aside (and to my mind it's simply semantic nitpicking) digital sensors (as opposed to film) simply do not have the dynamic range that film does. It is the only area that is truely behind film. I don't care how it is implemented whether they take 3 shots, 1 shot and divide it, or simply devise a sensor with greater dyanmic range, I would like to have more DR in my shots. And it is nice to see that progress is being made however slowly it seems now. I hope the K20D has greater DR, but I doubt that it will be significantly greater. If it is I will actually think seriously about getting one, as it stands now my K10D is plenty enough for me.

NaCl(HDR is pretty much the last frontier for digital photography IMO)H2O

[mattdm]

[QUOTE=KungPOW;173813]Neat find!
I could imagine a sensor that instead of having colour receptors for green, red, and blue, just sensed brightness. The red sensor could be replaced with a -2 EV sensor, the green with a 0 EV sensor, and the blue with a +2 EV sensor. Or whatever +/- combo to get the best results. One shot would be taken, and the cameras software would mix the result into HDR artwork./QUOTE]

This is sort of what Fuji does in their SuperCCD for their (Nikon-lens-mount) dSLRs. (It's different from the SuperCCD in their point & shoot cameras despite using the same marketing term.) They've only got two "sizes" of sensors, though, and there's clearly a color filter involved.

[falconeye]

and I don't want to get involved in a quantum/waveform argument) ... it's simply semantic nitpicking

I think it is nitpicking as well. But hey, Einstein got a Nobel prize for his nitpicking discovery that light isn't an analog medium at all (he got it for the so-called "photo-effect", not "relativity", at a time where everybody else shared your thinking - by the way, today's camera sensors are still based on this effect ). And please, don't disesteem quantum physics. Without, there would be no digital photography, that's for sure. Future progress will come from there.

As far as photon counting is concerned... The situation isn't that bad already: By quantum efficiency (QE) one means how many photons must hit to create one electron charge in the sensor. The quantum efficiencies of the CCDs and CMOS sensors in modern DSLR cameras is about 20 to 50 percent, depending on the wavelength. Top-of-the-line dedicated astronomical CCD cameras can have quantum efficiencies of 80 percent and more, although this is for grayscale images.

The problem now is that the incremental charge (which is digital) is collected until read out as an analog measure and then back-converted to digital electronically. The read-out happens globally by shifting charges around which is slow and causes cells to become "full" or even "bloom" (CCD). One could add a microscopic digital charge counter to every sensor cell and digitally increment/decrement each time the charge changes, then reading counters like reading RAM memory. This makes the sensor more expensive. Also, to reduce thermal noise one should cool down the sensor with a peltier cooling element (as is done in astro photography). All this isn't done because nobody would pay for it.

digital sensors (as opposed to film) simply do not have the dynamic range that film does. It is the only area that is truely behind film.

I agree if you add "current" to "digital sensors". Theoretically, a digital sensor can by far outperform a film in DR. I hope I made this clear.

HDR is pretty much the last frontier for digital photography

Again, I agree. But as I may have illustrated, the main obstacles are by market, not technology. Because nobody can actually display HDR (monitors and print aren't HDR), there is no real market pressure behind. Digital mirror beamers and OLED may change this in the future, though.

---

Last edited by falconeye; 02-09-2008 at 09:00 AM.

[NaClH2O]

*snip*

Again, I agree. But as I may have illustrated, the main obstacles are by market, not technology. Because nobody can actually display HDR (monitors and print aren't HDR), there is no real market pressure behind. Digital mirror beamers and OLED may change this in the future, though.

OK I will accept your explanations. However I tend to disagree a bit about this final statement. True, both monitors and print are far behind the DR our eyes can discern, BUT they are far ahead of what current digital sensors can display. As the article I linked shows, a standard ditgital photo of the lit lightbulb is blown out in the center where as the HDR one isnt. I'd just like to see what I can see in film, even scanned film has much greater DR than the current digital sensors are capable of displaying.

NaCl(I'm not a "good old days" but I still would like the DR I had with film)H2O

[dosdan]

to reduce thermal noise one should cool down the sensor with a peltier cooling element (as is done in astro photography). All this isn't done because nobody would pay for it.

Peltier cooling intergrated into the back of the CCD chip should be doable and not very expensive. It's only a small chip area and the cooling required isn't great.

Thermoelectric cooling isn't very efficient (5-10%), and to reduce the power consumption, can only be on for a short time. I would envision it not being turned on for every shot to reduce trigger lag and overall power usage. Only turned on when you needed to emulate High ISO or when you specifically requested HDR mode.

It would need to fit into auto-focusing/aperture setting time window, so it would need to reach full effectiveness on the photosensors within about 0.3-0.7 secs to be acceptable. The speed of cooling would depend on the thermal inertia and thermal conductivity of the CCD chip. Using a pulse of greater power would help. It all comes down to getting the right balancing act of power consumption/time/cost.

[falconeye]

Thermoelectric cooling isn't very efficient (5-10%), and to reduce the power consumption, can only be on for a short time.

Hi dosdan,

this is all correct. With a high DR chip however (better than 18bit), cooling becomes a major topic. To fully use one more bit of dynamic range you effectively must half your ISO setting. Because this quickly becomes infeasible you must defeat thermal noise. Maybe thermoelectric cooling can't do it but some solution would be required - crushed ice maybe

[dosdan]

With a high DR chip however (better than 18bit), cooling becomes a major topic.

Well we're considering portable DSLR usage. I can't envision the cooling being activated for very long, so the thermal resistance of the hot end-to-air (as opposed to the thermal resistance of cool end-to-CCD) does not need to be very low unless you wanted to get both HDR and a high shooting rate at the same time. How long do any existing thermoelectric-cooled, non-astro DSLRs turn on the chip for and what on-time/off-time duty cycle or shot rate are they capable off?