[WerTicus]

What I would like to see is three simultaneous processors using a CCD sensor. One would process at -2EV, one at 0EV, and one at +2EV.

was done by fuji already...in those camera's you see with amazingly high dynamic range scores (for the time) they had two sensors in one however, not three.

and actually isnt this what the k-5's sensor does? but in software? takes two readings within fractions of a fraction of a second to get the dynamic range?

[thibs]

and actually isnt this what the k-5's sensor does? but in software? takes two readings within fractions of a fraction of a second to get the dynamic range?

Nope. There's HDR mode however.

[stillshunter]

For what it's worth, in my experience, even a well processed HDR doesn't approximate the output of a good Sigma image. HDR or exposure blended images can provide better dynamic range but still doesn't have that Foveon 'pop'!

The feeling of seeing your x3f file come alive after SPP (Sigma Photo Pro) has done its stuff is something everyone should experience once in their lifetime.

[philippe]

You know what would be really great? I have an idea for a sensor that doesn't even use electricity -- so there's no readout noise whatsoever. Pixel density is basically unlimited. It uses special light-sensitive chemicals to capture images on a very thin, inexpensive plastic substrate. And best of all, if you want to increase sensitivity, all you do is switch to a different "cartridge" -- there's no need to get a new camera every year! It's the future, I tell you!

AH, finally an other 'film lover'...
So, that's how the past can be the future!

[deejjjaaaa]

With the K-5 Pentax has showed us that the size of the sensor is not the matter...

not Pentax, but Sony Semiconductor (!= Sony Imaging)... and wait for the next generation of FF sensors to put the crop where it belongs again (as the current generation of FF sensors are 2-3 years old designs).

[Pompous Moronox]

Why don't pentax use this technology in its upcoming model. it is supposedly better than the rest of the image processing sensor technology.

Any tech gurus out there, comments will be welcomed

Foveon X3 is not superior to the regular CFA CMOS sensors. Instead the current implementation is quite a bit inferior. For example there is no correlated double sampling for the photosites, thus the reset noise remains a problem. The total read noise is an order of magnitude higher than that of the modern sensors, like the one in the K-5. In addition there is plenty of chroma noise due to the awful colour separation of the technology. Also the green, and especially the red layers provide rather unsharp image data due to crosstalk.

The dynamic range is also lower than that of the modern competeting models - this is because of the excessive noise. It is many stops less than that of K-5 for example.

The colour accuracy is notoriously bad and unfortunately also impossible to fully fix due to metameric failure (which is in nature similar to what happens when one uses led lights to illuminate a photograph - the colours can be perfectly fine, but they can also be horrible and not fixable).

The resolution is very low due to the low pixel count. Also the bottom layers are less sharp than the top one. The red one especially can be bad.

Assuming that the patent US 2010/0155576 filed on Apr 11 2008 describes the sensor in SD1, we can confidently say that that sensor is a massive improvement over the current generation Foveon sensors. This patent describes a sensor with CDS ( -> far less read noise) and a novel and pragmatic approach to some of the problems the three layer pixel structure creates. Namely, there are four times fewer pixels on the red and green layer than on the blue layer. The loss of resoution is neglible and QE will be higher than with a homogenic structure.

However, even that sensor will still be plagued by the colour issue as that is due to the very nature of the colour separation. Also the separation will be as weak as before, increasing noise somewhat.

[Pompous Moronox]

was done by fuji already...in those camera's you see with amazingly high dynamic range scores (for the time) they had two sensors in one however, not three.

and actually isnt this what the k-5's sensor does? but in software? takes two readings within fractions of a fraction of a second to get the dynamic range?

Fujis sensor had two photosites per pixel - a large sensitive main photodiode and one small panchromatic less sensitive photodiode. The reason for the extraordinary highlight ability upto ISO800 was this structure - at ISO800 the DR is actually quite a bit larger than that of the K-5.

Now, this is the "engineering DR", not photographers DR. The problem with the Fuji is that the midtones are quite bad (worse than for example K20D, Canon's very old 20D!), and the shadows are even worse. What this means to the photographer is that the higher minumum quality standard he has for his photos, the lower the DR of the Fuji will be compared to the competition. The Fuji can be a nice speacial purpouse camera, but it's quite bad imaging device for the great majority of us.

K-5 does not do what you guess it does. It is just a well designed and engineered conventional sensor. The critical feature for the low read noise which is the reason behind the high DR is the massively parallel on chip AD conversion.

[stillshunter]

Foveon X3 is not superior to the regular CFA CMOS sensors. Instead the current implementation is quite a bit inferior....etc, etc, etc....

Pompous, have you actually experienced a Foveon sensor? I mean for more than a month or so, not 30 minutes in a camera store or a weekend with a friend's unit, or via some really robust research on the web.

With all due respect, it's really tedious to hear folks rubbish the Sigma cameras either based on specifications, reviews or a casual handling. I say, own one for at least a month and then tell me about how much you prefer your Bayer!

[deejjjaaaa]

Pompous, have you actually experienced a Foveon sensor? I mean for more than a month or so, not 30 minutes in a camera store or a weekend with a friend's unit, or via some really robust research on the web.

With all due respect, it's really tedious to hear folks rubbish the Sigma cameras either based on specifications, reviews or a casual handling. I say, own one for at least a month and then tell me about how much you prefer your Bayer!

he is just writing about the technical details and he mentioned all (almost all - you can find some more) the valid shortcomings in the current (that you can actually buy) generation of Foveon sensor... and here is the beauty of it - few sensors can be so bad and yet have so many people who like 'em so many years after the first camera w/ such sensor was shipped... who in the right mind will be excited from Samsung 14mp mediocre piece of crap or for that matter from some Canon APS-C sensor after couple years ? sensors that excite people are either "odd" (Foveon, Fuji Super CCDs of S3/S5 or Kodak CCD in M9, etc) or have something technically good in them (current Sony 16mp APS-C, Nikon 12mp FF, etc)... and some like Fuji's are both odd and still good in some technical aspects even now.

[stillshunter]

Right, well if Pompous is talking from a purely theoretical position, then that makes more sense. Doesn't make it any more accurate, just more sense.

...sensors that excite people are either "odd" (Foveon, Fuji Super CCDs of S3/S5 or Kodak CCD in M9, etc) or have something technically good in them (current Sony 16mp APS-C, Nikon 12mp FF, etc)..

So I assume you are suggesting that the Foveon sensor doesn't "have something technically good in them"? I note from your signature block that you own a DP2. So why do you persist with that outdated and superseded technology? A shame it's not a DP1 (have a lonely DP2 already), then I would gladly provide it with a stable, loving and appreciative home

Again, judgement on the Foveon IQ can only be made by someone who has tested it first-hand. The UI of the bodies they are wrapped in can only be made by someone who has spent longer than a dedicated fortnight with it. Anything else is pure theory. As useful as us constructing an argument on who would win in a fight between Superman or The Hulk!

[deejjjaaaa]

So I assume you are suggesting that the Foveon sensor doesn't "have something technically good in them"?

technically good things are 1) demosaicking is not necessary and 2) sensel's density is somewhat less as you stack 3 sensels in one spot -> lenses are less strained to deliver proper resolution

and while the absence of CFA also makes daylight colors @ low ISO nice & lovely (personal & subjective) it can hardly be called a technical achievement (by that we should mean how exact colors can be reproduced under various light, etc)

I note from your signature block that you own a DP2.

indeed

So why do you persist with that outdated and superseded technology?

because I like the "look & feel" of the images that I personally (here goes the proverbial text about the mileage) can get using my raw conversion/pp tools and how easy I get what I personally like from it - but at the same time, I acknowledge all of its shortcomings...

A shame it's not a DP1 (have a lonely DP2 already), then I would gladly provide it with a stable, loving and appreciative home

yawn... cheap rhetoric... glad that you did not ask "Have you stopped beating your wife?" or something along these lines of loaded questions.

Again, judgement on the Foveon IQ can only be made by someone who has tested it first-hand.

absolutely not... your first hand testing will not correct metamerism, abolish the laws of physics and so on... what you mean by IQ is essentially your person taste... and we can't argue about tastes

The UI of the bodies they are wrapped in can only be made by someone who has spent longer than a dedicated fortnight with it. Anything else is pure theory. As useful as us constructing an argument on who would win in a fight between Superman or The Hulk!

stream of consciousness...

[stillshunter]

Appreciate your honesty deejaa! And respect the fact that you are speaking from a position of experience. And hats off to you for your array of cameras - KX, GH2, DP2 and LX3 - Niiiice!

Well we'll part company then, agreeing that it's a case of "different strokes for different folks". In closing I have to say that I enjoy my K-r as much as I do my DP2s - just for two very different reasons.

[NewTake]

I have the SD14.... three words: can't stand it. I bought it because the IR filter is easily removed by hand. The functionality of the camera drives me nuts. It is very primitive, and very buggy. I have the most up to date firmware and the camera periodically will freeze up, mainly when shooting in bracket mode. It seems like the processor cannot keep up and it is very slow (I shoot RAW).
Not sure if the SD15 made a large improvement upon the SD14, but the SD14 needed a lot of work.
The compact Sigma cameras may fare better, but I rarely use the SD14 anymore, just because it is so tedious and frustrating. I really want to like it, and I actually like the unique 'pastel' colors it produces (with the IR blocker on of course).
It just isnt worth me carrying around as a 2nd camera to the k7, even if it can capture IR. Not to mention the image size is that of a 5mp camera, which can be a bit annoying too.

[deejjjaaaa]

...

well, people are buying Sigma cameras only for the sensor alone... cameras sans sensor are simply bad (or rather 5-10 years behind the state of technology), not worth any discussion outside Sigma forums, unlike the sensor... however based on postings @ dpreview (sigma section) it seems that SD15 operation-wise is really better than SD14...

[jeffshaddix]

well, people are buying Sigma cameras only for the sensor alone... cameras sans sensor are simply bad (or rather 5-10 years behind the state of technology), not worth any discussion outside Sigma forums, unlike the sensor... however based on postings @ dpreview (sigma section) it seems that SD15 operation-wise is really better than SD14...

For a long time now I've been hoping for a Pentax/Sigma cooperative effort. A K-5/7 body with a foveon sensor would be my dream come true. Alas, that's only a pipe dream, but it's still fun to think about. Too bad the mounts aren't compatible with a simple adapter.