The principle problem with "classical" digital imaging

[FDecker]

Frank and everyone else-

I think that you're right...and also a little off...with your reasoning.

I'm an artist, not an engineer, so take what I say with a grain of salt.

I needed to back up a hard drive the other day, and in the process wound up clicking around in folders I had forgotten about.

What popped up we're images from different cameras I've owned, or borrowed from friends, or tried out.

Looking back over a longer arc often gives you a fuller, more impartial view over something than the short term, right?

Why I say you're right on track: some of the photos that surprised me the most were from the Nokia 808 Pureview phone. 41 mini megapixels condensed into 7(?) superpixels. It was lousy in low light, but the pics in daylight are still richer, more detailed and have "thicker" colors than my new Samsung with Google Nightsight app. Nokia was on the right path with the efficient Symbian OS, and leading the way in imaging. Too bad they flatlined.

Some of my other (non-Foveon ;-) favorites came from a Sony R1 from 2006, a Kodak dcs 14n that I bought heavily used and got some great shots before I dropped it in the sea, and candids from a friend's Leica M9 that I borrowed for a weekend. What do they have in common? All CCD sensors.

At the time of using those cams I was caught up in the technology of what they did. Now, looking back over 15 years of photos, the artist in me sees the other, more important, story: the tech only serves to capture light. Tonal transitions, fine reflections, color depth, fog, texture, etc

I would have preferred at least the option of CCD full frame with today's tech, despite drawbacks (bad high ISO relative to Bayer, higher cost etc). CCD sensors went the way of Betamax, Nokia folded. I'm holding hopes that Sigma's 3-layer full frame builds a great tradition, whatever happens in phone and computational photography.

Cheers

From your post I don‘t see there you think that I am a little off. My explanation doesn‘t refer to the way the sensor stores and reads out the data. CCD or CMOS doesn‘t make a difference here. Bayer sensors use CCD as well as CMOS, even though CMOS is now in 99% of all cameras. All Foveon chips are CMOS. CCD has some principal advantages but CMOS did catch up quite fast with a lot of money going into its development.

 

[ascotinitaly]

Frank,

Yes, you are right in a literal sense. I didn't explain myself very well- what I was getting at is that /to my eyes/ ccd sensors give/gave a /very/ film-like result...and Bayer /still/ isn't there. And had they benefitted from the extra decade or so of development that cmos has enjoyed, I'm convinced that ccd would be the superior technology by now. But this comes from someone who also writes, paints, builds etc so I'm not part of the growing number of shooters who can't remember choosing not cameras, but /film/. I enjoyed your post and welcome any technology that brings more accurate, richer color with better separation and "3d", whatever you want that to mean....

Have a great weekend

 

[FDecker]

Frank,

Yes, you are right in a literal sense. I didn't explain myself very well- what I was getting at is that /to my eyes/ ccd sensors give/gave a /very/ film-like result...and Bayer /still/ isn't there. And had they benefitted from the extra decade or so of development that cmos has enjoyed, I'm convinced that ccd would be the superior technology by now. But this comes from someone who also writes, paints, builds etc so I'm not part of the growing number of shooters who can't remember choosing not cameras, but /film/. I enjoyed your post and welcome any technology that brings more accurate, richer color with better separation and "3d", whatever you want that to mean....

Have a great weekend

I am with you. CCD sensors have advantages. I think the most important one is a more even conversion of light intensity into voltage and finally digital information (in the sense of spatial evenness). This is due to non-local conversion whereas CMOS uses local conversion.

In the beginning of digital photography, the was no question about the superiority of CCD over CMOS. As far as I understand, nowadays, the gap has been closed or even more than closed.

Thank you for your comments.

Have a great weekend as well,

Frank

 

[DavidMillier]

Frank,

Yes, you are right in a literal sense. I didn't explain myself very well- what I was getting at is that /to my eyes/ ccd sensors give/gave a /very/ film-like result...and Bayer /still/ isn't there. And had they benefitted from the extra decade or so of development that cmos has enjoyed, I'm convinced that ccd would be the superior technology by now. But this comes from someone who also writes, paints, builds etc so I'm not part of the growing number of shooters who can't remember choosing not cameras, but /film/. I enjoyed your post and welcome any technology that brings more accurate, richer color with better separation and "3d", whatever you want that to mean....

Have a great weekend

I am with you. CCD sensors have advantages. I think the most important one is a more even conversion of light intensity into voltage and finally digital information (in the sense of spatial evenness). This is due to non-local conversion whereas CMOS uses local conversion.

In the beginning of digital photography, the was no question about the superiority of CCD over CMOS. As far as I understand, nowadays, the gap has been closed or even more than closed.

Thank you for your comments.

Have a great weekend as well,

Frank

The Kodak 14n didn't use a CCD - it was the first full frame sensor and the first full frame CMOS sensor. It was also an appalling quality sensor with terrible noise and colour shading issues and was very quickly replaced by the somewhat improved 14nx module /SLRn sensor.

 

[ascotinitaly]

What does "not local" conversion mean, Frank? As I said, I'm not very tech oriented

re:light intensity

I'm surprised that in 2019 there isn't a sensor that I know of that can record a huge dynamic range in a single capture. In other words- why don't we have sensors whose "light Wells" can re-set once full and re-load in the brightest parts of a scene? Is that not possible. I know a guy who has a successful wedding business. Despite mostly Sony and Fuji gear, he still shoots a Fuji S5 and it's dual-pixel sensor when shooting brides in bright Mediterranean sun.

 

[FDecker]

Frank,

Yes, you are right in a literal sense. I didn't explain myself very well- what I was getting at is that /to my eyes/ ccd sensors give/gave a /very/ film-like result...and Bayer /still/ isn't there. And had they benefitted from the extra decade or so of development that cmos has enjoyed, I'm convinced that ccd would be the superior technology by now. But this comes from someone who also writes, paints, builds etc so I'm not part of the growing number of shooters who can't remember choosing not cameras, but /film/. I enjoyed your post and welcome any technology that brings more accurate, richer color with better separation and "3d", whatever you want that to mean....

Have a great weekend

I am with you. CCD sensors have advantages. I think the most important one is a more even conversion of light intensity into voltage and finally digital information (in the sense of spatial evenness). This is due to non-local conversion whereas CMOS uses local conversion.

In the beginning of digital photography, the was no question about the superiority of CCD over CMOS. As far as I understand, nowadays, the gap has been closed or even more than closed.

Thank you for your comments.

Have a great weekend as well,

Frank

The Kodak 14n didn't use a CCD - it was the first full frame sensor and the first full frame CMOS sensor. It was also an appalling quality sensor with terrible noise and colour shading issues and was very quickly replaced by the somewhat improved 14nx module /SLRn sensor.

You are right. But the answer should go to the post of ascotinitaly.

 

[FDecker]

What does "not local" conversion mean, Frank? As I said, I'm not very tech oriented

re:light intensity

I'm surprised that in 2019 there isn't a sensor that I know of that can record a huge dynamic range in a single capture. In other words- why don't we have sensors whose "light Wells" can re-set once full and re-load in the brightest parts of a scene? Is that not possible. I know a guy who has a successful wedding business. Despite mostly Sony and Fuji gear, he still shoots a Fuji S5 and it's dual-pixel sensor when shooting brides in bright Mediterranean sun.

A CCD uses an external conversion from charge (# of electrons) to voltage to digital value. All charges are moved to an external active circuit (non local). Because all sensor elements are converted by the same circuit, there is no influence on the picture by slight variations in the electronic part. It is much slower than CMOS where every picture element of the sensor has an active circuit which performs the charge to voltage and in some cases the voltage to digital value conversion (local). But on the other hand, variations in the conversion gain of these local circuits are reflected in the final image.

I agree, a sensor as you describe it would be really good in terms of S/N and DR. I think the problem is that the charges must be cleared very fast from a full photo diode and in addition, the photo diode needs a local digital counter which is incremented with each overflow. Maybe, this is asking too much for todays technology.

 

[DavidMillier]

What does "not local" conversion mean, Frank? As I said, I'm not very tech oriented

re:light intensity

I'm surprised that in 2019 there isn't a sensor that I know of that can record a huge dynamic range in a single capture. In other words- why don't we have sensors whose "light Wells" can re-set once full and re-load in the brightest parts of a scene? Is that not possible. I know a guy who has a successful wedding business. Despite mostly Sony and Fuji gear, he still shoots a Fuji S5 and it's dual-pixel sensor when shooting brides in bright Mediterranean sun.

I've been saying this for a decade. Sensors ought to have infinite dynamic range if each photosite fills, resets, refills, resets , refills. I guess there must be a technical reason why each photosite can't be individually addressed, but superficially it sounds like the obvious solution.

 

[Scottelly]

Frank and everyone else-

I think that you're right...and also a little off...with your reasoning.

I'm an artist, not an engineer, so take what I say with a grain of salt.

I needed to back up a hard drive the other day, and in the process wound up clicking around in folders I had forgotten about.

What popped up we're images from different cameras I've owned, or borrowed from friends, or tried out.

Looking back over a longer arc often gives you a fuller, more impartial view over something than the short term, right?

Why I say you're right on track: some of the photos that surprised me the most were from the Nokia 808 Pureview phone. 41 mini megapixels condensed into 7(?) superpixels. It was lousy in low light, but the pics in daylight are still richer, more detailed and have "thicker" colors than my new Samsung with Google Nightsight app. Nokia was on the right path with the efficient Symbian OS, and leading the way in imaging. Too bad they flatlined.

Some of my other (non-Foveon ;-) favorites came from a Sony R1 from 2006, a Kodak dcs 14n that I bought heavily used and got some great shots before I dropped it in the sea, and candids from a friend's Leica M9 that I borrowed for a weekend. What do they have in common? All CCD sensors.

At the time of using those cams I was caught up in the technology of what they did. Now, looking back over 15 years of photos, the artist in me sees the other, more important, story: the tech only serves to capture light. Tonal transitions, fine reflections, color depth, fog, texture, etc

I would have preferred at least the option of CCD full frame with today's tech, despite drawbacks (bad high ISO relative to Bayer, higher cost etc). CCD sensors went the way of Betamax, Nokia folded. I'm holding hopes that Sigma's 3-layer full frame builds a great tradition, whatever happens in phone and computational photography.

Cheers

Scott I had a Sony R1, and I loved it. I used that camera for years and years, until a wave hit it at the beach, when I was trying to get down low and close to the wet sand with it. I don't miss that camera, because it was so slow, and both the viewfinder and review screen were small and low resolution. The auto-focus was horrendous too - so bad that I just manual focused almost everything. Still, I loved the photos that came out of that camera (at the lowest ISO settings). Its 1.7x crop factor sensor competed well with my Canon 20 D, which I bought it to replace. Its Zeiss-labeled 24-120mm equivalent zoom lens was just great. I missed the ability to change lenses, though I really never did that with my 20 D anyway. When I got my Canon 5 D and three lenses for it, I hardly ever used the Sony anymore. When my Canon 5 D stopped working right I went back to that Sony though, and it wasn't bad, even after shooting for years with the 5 D. I think that says a lot.

BTW, I checked, and the sensor in the R1 was indeed a CMOS sensor:

https://www.dpreview.com/reviews/sonydscr1

 

[ascotinitaly]

Ok, my memory remembered the R1 as a ccd unità, I shoulda looked it up. The Kodak was definitely ccd though, so I must be remembering a different/earlier modwl. I really enjoyed that can, but it met the same fate as Scott's R1- seawater. I got bumped by tourists from behind, so now every camera has a wrist strap around my wrist...as cheap insurance!

 

[D Cox]

What does "not local" conversion mean, Frank? As I said, I'm not very tech oriented

re:light intensity

I'm surprised that in 2019 there isn't a sensor that I know of that can record a huge dynamic range in a single capture. In other words- why don't we have sensors whose "light Wells" can re-set once full and re-load in the brightest parts of a scene? Is that not possible. I know a guy who has a successful wedding business. Despite mostly Sony and Fuji gear, he still shoots a Fuji S5 and it's dual-pixel sensor when shooting brides in bright Mediterranean sun.

I've been saying this for a decade. Sensors ought to have infinite dynamic range if each photosite fills, resets, refills, resets , refills. I guess there must be a technical reason why each photosite can't be individually addressed, but superficially it sounds like the obvious solution.

I think the new type of CMOS sensor that Eric Fossum and his team are working on does something like this. But it isn't likely to be commercial for years, and then only for industrial and scientific uses.

(Eric Fossum is the inventor of CMOS sensors.)