Interest in designing a better digital camera?

[George Mann]

I am wondering whether or not you'all would be interested in discussing design idea's concerning whats possible regarding the theoretical design of an improved digital camera or capture device?

I have a few theories on the subject that I would like to discuss, with more related design idea's to come.

 

[Вormental_old]

Happy to engage, although my interest lies in the similar (but related) area of re-purposing digital camera technology for other purposes, starting with a non-retarded film scanner. I want to know if it's possible to acquire a sensor development board with a USB interface and a SDK.

 

[George Mann]

I want to know if it's possible to acquire a sensor development board with a USB interface and a SDK.

Most likely made to order at best, as most engineers make their own.

 

[Вormental_old]

Most likely made to order at best, as most engineers make their own.

What are you talking about? Nobody ever makes their own. I want to see you soldering a CMOS sensor to a custom PCB.. Chip manufacturers provide design kits usually consisting of a board and an SDK, here's a page from TI (my world) and here's one I found and bookmarked earlier for imaging chips.

George, I'm afraid we're not going to have a productive conversation

 

[radiant]

I want to know if it's possible to acquire a sensor development board with a USB interface and a SDK.

There are multiple industrial cameras with this kind of capability. You have full control over basically everything you can imagine.

Our ex-local film lab had built his own Super8 / 16mm scanners with these kind of cameras. I remember the 16mm scanner was monochrome and evey frame was illuminated separatelu by R,G,B. (edit: found one video about him, you can see the scanner here start at 2:47).

 

[George Mann]

What are you talking about? Nobody ever makes their own.

How do you think that the early engineers built prototypes?

 

[Вormental_old]

How do you think that the early engineers built prototypes?

I started my career in the early 2000s and even then we had all our prototyping on prebuilt hardware kits with FPGAs before moving to our own PCBs with ASICs. I have no idea which century or industry you're talking about. If you insist on starting from absolute zero, be my guest and waste a year designing a power supply.

 

[George Mann]

Back to the subject at hand. Two limiting factors that I see are,

1. The use color filters that divide the available sensor sites into 3rds.

2. The necessary use of an imperfect process know as interpolation to combine these sites into an effective total sum.

I have always argued that it is better to use a monochrome sensor, which will allow us to yield the sensors full native resolution, while doing away with the need for interpolation.

3rd, a monochrome sensor will allow us to use less damaging filters.

What about color you ask? This can be added during post processing with a sophisticated algorithm.

 

[George Mann]

If you insist on starting from absolute zero, be my guest and waste a year designing a power supply.

I actually find such ventures to be a stimulating challenge!

 

[George Mann]

Have you ever heard of multiple circuit summation? It can be used to increase not only the usable dynamic range of the sensor, but also its signal-to-noise ratio.

Combined with a special feedback loop, we can achieve a fairly linear output.

 

[Вormental_old]

The necessary use of an imperfect process know as interpolation to combine these sites into an effective total sum.

It's as perfect as your own retina! But back to your proposal:

What about color you ask? This can be added during post processing with a sophisticated algorithm.

What algorithm? You skipped the 99% of the meat in your proposal.

 

[George Mann]

It's as perfect as your own retina! But back to your proposal:

Actually, I have always argued the opposite, and believe that it has a detrimental effect.

Do you really believe that the process of combining 3 images
can be done so perfectly as to produce zero artifacts?

What algorithm? You skipped the 99% of the meat in your proposal.

I am not looking to reinvent the wheel, just to improve upon its design.

 

[Deleted member 88956]

Back to the subject at hand. Two limiting factors that I see are,

1. The use color filters that divide the available sensor sites into 3rds.

2. The necessary use of an imperfect process know as interpolation to combine these sites into an effective total sum.

I have always argued that it is better to use a monochrome sensor, which will allow us to yield the sensors full native resolution, while doing away with the need for interpolation.

3rd, a monochrome sensor will allow us to use less damaging filters.

What about color you ask? This can be added during post processing with a sophisticated algorithm.

Hm, the only thing I ever read about imaging sensors is they are all inherently monochromatic. Color filter layer is on top and interpolation is the needed trick to combine them into color image "on the fly". Astro cameras are all monochromatic and shots are taken through filter set for same purpose except each shot is taken at full resolution through each filter, then it is all combined / stacked to create final color image.

 

[Вormental_old]

Yep, you can send your camera to remove the color filter - https://maxmax.com/maincamerapage/cameraconversions

 

[Deleted member 88956]

Yep, you can send your camera to remove the color filter - https://maxmax.com/maincamerapage/cameraconversions

Yes, I was referring to OP about his apparent idea of needing to use a monochrome sensor in the improved digital camera, which they all are since the beginning of time.

 

[George Mann]

Yes, I was referring to OP about his apparent idea of needing to use a monochrome sensor in the improved digital camera, which they all are since the beginning of time.

Of course they are! But the current crop could be improved upon.

 

[George Mann]

Does anyone else have any useful ideas?

 

[Вormental_old]

For solving what problem?

 

[foveon_m]

best to install 3 sensors like still be used in some video cams.

 

[George Mann]

For solving what problem?

For me, it's the same problems overall as found in digital audio, the lack of a true you-are-there realism.

Edit. If you are speaking of the current state of sensor design, then the efficiency level of capture can still be improved upon.

 

[George Mann]

best to install 3 sensors like still be used in some video cams.

Then you still have to rely on an imperfect process to combine the images.

 

[4season]

Not sure what the fuss over monochrome sensors is all about, as it seems to me that use of technology like pixel-shift can yield very high resolution images from a sensor equipped with Bayer array. Downside is that as currently implemented (Olympus, Sony) it's not an instantaneous process and doesn't work well with a handheld camera or with moving subjects. But it's a lot less fussy than manually fussing with R, G and B filters.

But I do wonder if digital sensors can be designed which aren't so sensitive to incident angle of light falling upon their surfaces.Tricks like offset micro lenses have been used to allow lens to be located closer to focal plane (Leica M) but it's an imperfect solution and still subject to light falloff and color shifts. To address such deficiencies, Sony announced the development of curved sensors several years ago:
https://www.dpreview.com/articles/2...ay-allow-for-simpler-lenses-and-better-images
https://cmte.ieee.org/futuredirections/2019/03/29/curved-image-sensors-good-but-late/
And they briefly sold a boutique compact camera employing a curved sensor to asian markets:
https://www.sony.net/SonyInfo/design/stories/dsc-kw1/01.html
But since then, nada. Wondering if they're just too costly to produce.

 

[George Mann]

4seasons, all of the technologies you listed above rely on
inherently imperfect
processes that can introduce unwanted artifacts which I am trying to avoid.

 

[4season]

I fail to see how multiple photos taken with monochrome sensor + separate R, G and B sensors has any advantage versus using pixel-shift with a Bayer sensor? Olympus combines the individual images in-camera, but Sony A7R3 and A7R4 does not, so you can choose your own software to do the deed.

 

[George Mann]

I fail to see how multiple photos taken with monochrome sensor + separate R, G and B sensors has any advantage versus using pixel-shift with a Bayer sensor?

It doesn't. They are all inherently flawed processes.