New sensor design - DPChallenge Forums

Olympus 40-150mm f/2.8 PRO M.Zuiko

06/01/2005 10:13:33 PM · #1

Here's a thought: how about a Foveon-esque hybrid sensor designed for wide dynamic range photos?

The Foveon method does away with the Bayer grid and uses a multi-layer approach to give each pixel full RGB info. The "standard" method only collects one color per pixel and then interpolates them to get RGB.

If we stick with the standard Bayer RGB we could add a "highlight" layer that would get much less light and therefore be useful in retaining highlight detail than is now lost when overexposure occurs.

I think Fuji has something similar, but all on one layer, with two different sensors per pixel.

Ok. So, who wants to build it?

Brad

Olympus OM-D E-M1

06/01/2005 10:25:29 PM · #2

I'll get right on it - no problem.

(well maybe just a little one - lemme find a dictionary...)

ElGordo

Canon EOS-400D Rebel XTi

Sigma 18-200mm f/3.5-6.3 DC OS for Canon

06/01/2005 10:27:58 PM · #3

I have been watching for a larger Foveon sensor to arrive on scene but they seem to be content with the 3.34 megapixel (X3 layers) currently in the SD9 and SD10 cameras. The picture quality is nothing short of amazing but with some caveats, like higher noise susceptibility and, as yet, the only recorded format is a proprietary one that requires their software for retrieval. The prices are dropping on both models, an SD9 went for $460 on ebay with lots of nice extras and 28-300mm lens. In my opinion, almost any current production DSLR with 6.1 or better rez will outperform either of these models. But the technology seems to be inherently superior to Bayer mask sensors whether ccd or cmos.

Message edited by author 2005-06-01 22:29:10.

Olympus 40-150mm f/2.8 PRO M.Zuiko

06/01/2005 10:30:06 PM · #4

Originally posted by BradP:

I'll get right on it - no problem.

(well maybe just a little one - lemme find a dictionary...)

Ok, just remember it's my idea and I'd like some of the profits... =]

Brad

Olympus OM-D E-M1

06/01/2005 10:34:00 PM · #5

Originally posted by BikeRacer:

Originally posted by BradP:
I'll get right on it - no problem.

(well maybe just a little one - lemme find a dictionary...)

Ok, just remember it's my idea and I'd like some of the profits... =]

Agreed.

(soon as I can translate all that into english..)

kyebosh

Canon EOS-20D

Canon EF 80-200mm f/2.8L

06/01/2005 10:36:50 PM · #6

That actually does make sense to me... Talk to Sigma and try to get them to make an SD24 :-)

Neil

Fujifilm X-T30

Nikon AF-S NIKKOR 24-70mm f/2.8G ED

06/01/2005 10:37:24 PM · #7

Originally posted by BradP:

Originally posted by BikeRacer:

Originally posted by BradP:
I'll get right on it - no problem.

(well maybe just a little one - lemme find a dictionary...)

Ok, just remember it's my idea and I'd like some of the profits... =]

Agreed.

(soon as I can translate all that into english..)

You can make it out of old Mercedes parts. Perhaps a radiator might be a good start. It might be a little large, but hey, they say a bigger sensor is better ;)

Olympus 9-18mm f/4.0-5.6 M.Zuiko Digital ED

06/01/2005 10:38:37 PM · #8

Originally posted by ElGordo:

I have been watching for a larger Foveon sensor to arrive on scene but they seem to be content with the 3.34 megapixel (X3 layers) currently in the SD9 and SD10 cameras.

...

In my opinion, almost any current production DSLR with 6.1 or better rez will outperform either of these models. But the technology seems to be inherently superior to Bayer mask sensors whether ccd or cmos.

I agree, but, you have to wonder if Bayer sensors were processed as if each of the sub-pixels only contributed to a larger "super-pixel" would that yield better results than the Foveon design. That would mean a 6MP sensor would only give you 1.5MP though.

jonr

Olympus OM-D E-M5

06/01/2005 11:08:21 PM · #9

I've always wondered why sensors don't have RGBC (C = Clear) instead of RGBG design. It would make true B/W option possible on a camera. And probably give some increased dynamic range...

Sigma 24-70mm f/2.8 EX Aspherical DF for Canon

06/01/2005 11:20:23 PM · #10

Originally posted by jonr:

I've always wondered why sensors don't have RGBC (C = Clear) instead of RGBG design. It would make true B/W option possible on a camera. And probably give some increased dynamic range...

Sounds good too. But, instead of clear, maybe ND+2 or something like it.

hankk

Canon EOS-5D

06/02/2005 09:23:33 AM · #11

Originally posted by BikeRacer:

Originally posted by jonr:
I've always wondered why sensors don't have RGBC (C = Clear) instead of RGBG design. It would make true B/W option possible on a camera. And probably give some increased dynamic range...

Sounds good too. But, instead of clear, maybe ND+2 or something like it.

Or use some sort of electrically changeable filter (like an LCD) to select the density of the filter.

muckpond

Nikon D200

Sigma 15mm f/2.8 EX Diagonal Fisheye for Nikon

06/02/2005 09:28:54 AM · #12

aren't you guys gonna need a flux capacitor for that thing?

gloda

Canon EOS-5D Mark II

Canon EF 24-70mm f/2.8L USM

06/02/2005 09:35:12 AM · #13

And 1.21 gigawatt.

Tamron SP AF 28-75mm f/2.8 XR Di for Canon

06/02/2005 10:25:14 PM · #14

Originally posted by gloda:

And 1.21 gigawatt.

That, or a dozen car batteries!

kyebosh

Canon EOS-20D

Canon EF 80-200mm f/2.8L

06/02/2005 10:29:58 PM · #15

if it's that much better, i'll be glad to carry around a little more battery weight...

yeoua

Canon PowerShot G2

06/02/2005 10:53:12 PM · #16

Originally posted by BikeRacer:

Usually the reason for more G is that our eyes are more sensitive to green, so having hardware/software that is more sensitive to green is a good idea. You can see this in some picture formats where more bytes are devoted to storing green than red or blue. I'm pretty sure our sensitivity is green, red, and blue, where green is most sensitive and blue the least.

nicklevy

Canon EOS-350D Rebel XT

06/02/2005 10:54:18 PM · #17

There is surely a technical disadvantage to that foveon construction. The idea is very old. Sony build 3CCD camcorders for 10 or 15 years and you can get here a very cheap 3CCD camcorder that use the same idea as the foveon for a couple hundred box. If this technology is so incredible then why our 1k$ and more camera doesn't have it all?

Message edited by author 2005-06-02 22:55:49.

Nikon AF-S DX Zoom-Nikkor 18-70mm f/3.5-4.5G IF-ED

06/03/2005 12:35:18 AM · #18

Originally posted by nicklevy:

But, the difference with the Foveon sensor is that one sensor records all info in three layers. With a 3CCD design, the light is split (introducing light loss) and sent to three different CCDs. Foveon has its issues with needing the light to fall perpendicular to the surface of the sensor (to insure it falls down the "wells") but, is a better and more modern design than optically splitting incoming light.

gbautista87

Nikon D70

06/03/2005 01:13:13 AM · #19

Originally posted by BikeRacer:

Originally posted by nicklevy:
There is surely a technical disadvantage to that foveon construction. The idea is very old. Sony build 3CCD camcorders for 10 or 15 years and you can get here a very cheap 3CCD camcorder that use the same idea as the foveon for a couple hundred box. If this technology is so incredible then why our 1k$ and more camera doesn't have it all?

But, the difference with the Foveon sensor is that one sensor records all info in three layers. With a 3CCD design, the light is split (introducing light loss) and sent to three different CCDs. Foveon has its issues with needing the light to fall perpendicular to the surface of the sensor (to insure it falls down the "wells") but, is a better and more modern design than optically splitting incoming light.

Don't all digital camera sensors (Bayer/Foveon and CCD/CMOS) need light to fall perpendicular to the sensor, anyway?

Message edited by author 2005-06-03 01:13:47.

Nikon AF-S DX Zoom-Nikkor 18-70mm f/3.5-4.5G IF-ED

06/03/2005 01:44:40 AM · #20

Originally posted by autobahn123:

Don't all digital camera sensors (Bayer/Foveon and CCD/CMOS) need light to fall perpendicular to the sensor, anyway?

Yes, but it's my understanding that because light has to travel far down the stack of photosites at each pixel location in a Foveon sensor, and still be able to fill each "bucket", perpendicular light is more critical to a Foveon sensor than with traditional Bayer sensors.

gbautista87

Nikon D70

06/03/2005 01:51:16 AM · #21

Would there be light loss in a Foveon sensor because the light has to pass through each layer of photosites?