How do non-bayer cameras fare in UV?

[Fandyus]

This question just came to me and I decided to post here in case anyone has any examples. I know there are various technologies, some for special uses, some outdated, that use different ways of making a color image besides what you usually find in digital cameras today.

The two examples I can list off the top of my head are:

1. Foveon sensors 

2. Some old P&S cameras that I believe have a different Bayer array filter configuration, not two green, one red, one blue but magenta, green, yellow and cyan.

 

Does anyone know how those react to UV?

Does anyone have examples of other obscure sensor technologies?

 

Please post below if you do.

 

Thanks!

 

 

 

[lukaszgryglicki]

I've read somewhere here that Foveon is disappointing - other bayer matrices - no idea... actually that double green always puzzled me - I always thought that it should be RGBF (F-meaning full - should pass entire visible) so algorithms can take an adventage of one value specifying full luminance. In case of removing hot mirror that F would become full spectrum (IR+Vis+UV).

 

My dream would be 6-elements bayer: RGBIUF: I=infrared, U=ultraviolet, F=full spectrum. And then a menu option to choose which of those 6 should be used for R, G, B channels to save a file, then you could do for instance:

R=F, G=F, B=F (and you have full spectrum), R=R, G=G, B=B (and you have a normal camera), R=I, G=G, B=U (and you have very nice separation of different spectrums as colors) R=G, B=B, B=U (you have shifted towards the UV), R=I, G=R, B=G (you have shifted towardsthe IR) and so on...

 

Imagine how much you could do with such a simple approach.

 

[Andy Perrin]

Dabateman has one!

 

[dabateman]

I can answer this as I have many odd ball cameras. 

The Sigma Foveon sensor is basically monochrome.  The top blue layer response to UV. It doesn't go deep though. I have a SD14 and it can see about to 350nm. I tested a SDQ and it can see maximum to 335nm. The output is monochrome. 

 

I have a Lodestar x2c camera, it has the odd GMYC colors on the sensor.  Its wickedly sensitive to UV. I would love a modern digital camera to use those colors. The problem is its far from color accurate.  Its more of like pick the color you want. I can see why the point and shoots dropped that array. Mine is an Astronomy camera used for tracking. So its very low resolution. 

This is its UV sensitivity: 

Versus typically sensor 

[lukaszgryglicki]

What "C" stands for in GMYC?

[dabateman]

My above post with graphs will tell you its Cyan, yellow,  magenta and green. 

 

The Lodestar x2c camera is the only one I know without a fixed lens and that color array. 

[lukaszgryglicki]

That's interesting - need to google that, would like to process its RAWs... do you have any chart of responses of each channel? Like which one shows most in UV - it would generate other "UV false colors" than everything else here, right?

[Nate]

Got me wondering now about my astro cam. It's the asi462, too bad the graph doesn't go further. Now I wish I bought a mono cam.

[dabateman]

The Imx462 looks interesting.  Its silicone is twice as thick as normal to capture more of the 800-1000nm spectral range.  So the IR should be very good off that tinny sensor., its 1/2.8 size.

I wonder if you could get to the silicone limit of over 1100nm using it. 

They sell it as monochrome from 850nm and up as at that region all color see the light equally.  But visible will give you color information. 

[dabateman]

Back to the topic. 

I am looking forward to seeing the teased Panasonic and Sony organic sensors. I have posts about them. Some of the Organic sensors have a green organic layer on top, which would result in very high UVB sensitivity. 

Hopefully Panasonic releases their recently teased organic sensor with 100x fold more dynamic range and global shutter.  That would be really nice and keep me into m43rds. 

[Fandyus]

4 hours ago, dabateman said:

Back to the topic. 

I am looking forward to seeing the teased Panasonic and Sony organic sensors. I have posts about them. Some of the Organic sensors have a green organic layer on top, which would result in very high UVB sensitivity. 

Hopefully Panasonic releases their recently teased organic sensor with 100x fold more dynamic range and global shutter.  That would be really nice and keep me into m43rds. 

Thank you for the responses. I'm not saying anything but you should definitely try more general UV with the GMYC camera you have, it looks really intriguing. Have you ever tried landscapes or flowers with the iconic bullseye pattern? Or people perhaps. People almost always look interesting.

[Fandyus]

9 hours ago, lukaszgryglicki said:

I've read somewhere here that Foveon is disappointing - other bayer matrices - no idea... actually that double green always puzzled me - I always thought that it should be RGBF (F-meaning full - should pass entire visible) so algorithms can take an adventage of one value specifying full luminance. In case of removing hot mirror that F would become full spectrum (IR+Vis+UV).

 

My dream would be 6-elements bayer: RGBIUF: I=infrared, U=ultraviolet, F=full spectrum. And then a menu option to choose which of those 6 should be used for R, G, B channels to save a file, then you could do for instance:

R=F, G=F, B=F (and you have full spectrum), R=R, G=G, B=B (and you have a normal camera), R=I, G=G, B=U (and you have very nice separation of different spectrums as colors) R=G, B=B, B=U (you have shifted towards the UV), R=I, G=R, B=G (you have shifted towardsthe IR) and so on...

 

Imagine how much you could do with such a simple approach.

 

That would be interesting but it would have two major downfalls.

If we assume the Bayer filters would have sufficient OD ratings, you would only be able to shoot UV/IR in monochrome, which would also tie directly into the second pitfall, which would be that if you were to isolate just one channel you'd get 1/6 of effective resolution compared to what you'd get when you would utilize all channels. And also a lot more noise. But it would be the holy grail for general full spectrum shooting, good luck getting a lens that can handle the chromatic abberation though.

Rather than a camera like this, I'd prefer a tri-color camera that is optimized to only see UV. Something like Bernard did but you don't have to painstakingly stitch three images. Same with IR.

[dabateman]

2 hours ago, Fandyus said:

Thank you for the responses. I'm not saying anything but you should definitely try more general UV with the GMYC camera you have, it looks really intriguing. Have you ever tried landscapes or flowers with the iconic bullseye pattern? Or people perhaps. People almost always look interesting.

I now have my Pi camera configured to control it using Phd2 software.  However,  Phd2 seems to shift the pixels slightly off from where Starlight live expects them. So the demoisacing is off. So for best results I would need to write a python script to demoisacing the fits raw files using Numpy, pyfits and what I know from piDNG.  So that a DNG file or even tif or jpeg could be created.  But that project took a backseat to other stuff.

Its the demoisacing the Fits files that really the problem with that camera. 

Starlight live does it best. But I would need to run around with a laptop, with Windows running to get it to work.

I even got Windows 11 installed on the raspberry pi, but it wouldn't recognize my the USB camera (the Lodestar x2c). So that too did really work.  

If I get around to writing that python script,  it would be good. I know the color pattern and orientation.