Jump to content


Photo

Does Debayering a DSLR improve sensitivity? if so how?


  • Please log in to reply
15 replies to this topic

#1 Starlock

Starlock

    Advanced Member

  • Moderators
  • 291 posts
  • LocationSuffolk, England

Posted 25 August 2013 - 10:10 PM

Hi folks,

 

There are a small number of DIY astronomers either thinking about or actually attempting to remove the Bayer matrix from their DSLR sensors. If successful this would result in a high resolution monochrome camera with a very large chip. Cooling aside the equivalent off the shelf dedicated CCD would cost a fortune so the rewards are great if achieved. So far many attempts have resulted in many dead sensors so its risky business! 

 

The main reason for modding a camera is to improve its sensitivity, the filter removal mod for example will improve a cameras sensitivity to H-alpha which is a very valuable gain over a standard camera, so apart from improving resolution will removing the bayer matrix from a sensor further improve sensitivity? if so what is the science behind this?

 

I for one just assumed that mono chips are more sensitive than colour chips because this is what I've heard here and there. It would be interesting to decide if this is true or false?

 

Anyone want to start the ball rolling :)

 

Best Wishes,

 

Chris



#2 Neil

Neil

    Administrator

  • Administrators
  • 1,493 posts
  • LocationThe Netherlands

Posted 25 August 2013 - 11:58 PM

OK, here is my reasoning on this subject.

 

Removing the bayer layer is definitely dangerous and will only improve resolution, not sensitivity for the following reasons.

 

1) Removing the glass window covering the sensor is a valid mod since it does improve the signal by allowing more of the RED spectrum to arrive at the CCD sensor.

2) Removing the bayer layer will not improve the signal, if anything it will make it slightly worse! The colour filters are very thin and located directly on the CCD sensor, whereas the filters in a filter wheel are not, therefore will absorb more of the transmitted light than the original bayer mask.

 

What you do get is an improvement in resolution, since now all of the pixels can be used in the RED, GREEN and BLUE channels. For narrowband the situation is still the same, an improvement is resolution only.

 

If you want to improve the signal to noise levels you must take some serious steps to cool the CCD sensor. This will not make its characteristic any better, however, it will lower the inherent thermal noise which in turn will improve the signal to noise level.

 

Today, there are a lot of software solutions to debayering a DSLR and they should be considered first. Once debayered, you effectively have the separate channels that a mono CCD sensor would produce and can treat them accordingly. You still have the issues that you get 2x GREEN versus the RED and BLUE channels, which is the only thing this modification solves.

 

My advise, think it through very carefully before you attempt this mod and possibly destroy your DSLR CCD sensor! I truly believe that taking steps to cool the CCD sensor will bring more benefit.

 

Clear Skies

 

 

Neil.



#3 steveward53

steveward53

    Advanced Member

  • Members
  • PipPipPip
  • 786 posts
  • LocationRed Lodge , Suffolk , UK.

Posted 26 August 2013 - 05:49 AM

Hi Chris ,

 

Having followed that painfully long thread on SGL I'd be inclined to go to these guys and get one ready done , costs about the same as 2 x 1100D's which is what most people spend buying 2 x 1100D's and destroying one or both trying the mod ...  :lol:

 

http://www.jtwastron...monochrome.html

 

 

I like the look of their Ultimate Deep  Cooled 1100d too , the dark frames are staggering .

 

http://www.jtwastron...s/ultimate.html

 

 

 


  • Nathancef likes this

Steve.

Mak180Pro , 200PDS , ED80 Pro , ST120 , ETX105.

HEQ5ProSynscan , EQ3-2.

Canon EOS1000D(unmodded) , SPC900NC.Std , SPC900NC Mono-chipped.

SW Synguider thru Mercury 705 , "Scopium" Herschel Wedge.

13'-6" pier for Solar..... 


#4 DENEB

DENEB

    Advanced Member

  • Members
  • PipPipPip
  • 303 posts
  • LocationLondon

Posted 26 August 2013 - 06:32 AM

I also think cooling is more Beneficial, look at this guy who got APOD recently :

 

http://apod.nasa.gov...d/ap130817.html

 

& look how he cooled his DSLR Camera:

 

http://www.pampaskie...pment/cooler_v3

 

Relatively simple to do, Im considering doing the same with my 1100D.

 

Here's an interesting article about Debayering by Craig Stark:

 

http://www.stark-lab...ayering_API.pdf


  • Andromeda likes this

#5 RickS

RickS

    Advanced Member

  • Members
  • PipPipPip
  • 620 posts
  • LocationBrisbane, Australia

Posted 26 August 2013 - 07:28 AM

You will get an effective increase in sensitivity when you collect Luminance data with your naked CCD.  You'll be collecting all the photons that fall on each pixel instead of the approximately 1/3 of the photons that the Bayer filter on each pixel will pass.  As already mentioned,you will see a difference in resolution, but not sensitivity, for R, G and B data.


  • Neil likes this

#6 Starlock

Starlock

    Advanced Member

  • Moderators
  • 291 posts
  • LocationSuffolk, England

Posted 26 August 2013 - 09:43 AM

@ Neil - I see why you think sensitivity won't be increased as you are only considering the situation of replacing the Bayer matrix with filters, and if these filters are at a greater distance from the chip there is perhaps more possibility of light scatter. Have you considered the pure luminance channel? This should be a fair amount more sensitive without any filters, microlenses, and bayer. There must be a reason why people say a mono CCD is just as quick if not quicker to collect data than the colour CCD equivalent, and this is because they take a shed load of lum and bin the RGB channels which takes a fraction of the time. You won't be able to bin say 2x2 with a DSLR so it wont be qucker but if you consider pure lumanence then sensitivity with be increased surely. Some objects look fantastic in mono such as the horse head neb, that combined with the improved resolution might make such a camera a good ally to a normally modded camera?



#7 Starlock

Starlock

    Advanced Member

  • Moderators
  • 291 posts
  • LocationSuffolk, England

Posted 26 August 2013 - 09:48 AM

Hi Chris ,

 

Having followed that painfully long thread on SGL I'd be inclined to go to these guys and get one ready done , costs about the same as 2 x 1100D's which is what most people spend buying 2 x 1100D's and destroying one or both trying the mod ...  :lol:

 

http://www.jtwastron...monochrome.html

 

 

I like the look of their Ultimate Deep  Cooled 1100d too , the dark frames are staggering .

 

http://www.jtwastron...s/ultimate.html

Hi Steve, they look great but I'm thinking of spending 50 quid max on such a project. The camera I have bought off ebay is a Canon 300D with mould on the sensor (cover glass). This has cost me a grand total of 29 pounds plus 10 pounds posatge :D I wont be too worried if I waste the 39 pounds but I still may bottle it when I have the camera stripped down in which case someone out there is going to get the worlds cheapest modded camera :D



#8 Starlock

Starlock

    Advanced Member

  • Moderators
  • 291 posts
  • LocationSuffolk, England

Posted 26 August 2013 - 12:06 PM

I also think cooling is more Beneficial, look at this guy who got APOD recently :

 

http://apod.nasa.gov...d/ap130817.html

 

& look how he cooled his DSLR Camera:

 

http://www.pampaskie...pment/cooler_v3

 

Relatively simple to do, Im considering doing the same with my 1100D.

 

Here's an interesting article about Debayering by Craig Stark:

 

http://www.stark-lab...ayering_API.pdf

Hi Deneb, thanks for the links, very useful and interesting stuff!

 

That APOD image is simply stunning and I didn't realise how effective the cool box and fan would be at cooling a dslr, definately food for thought and relatively simple to do!

 

I've re-read the stark-lab link, I've come across it before I realise. The summary seems to be both colour and mono have there place but one thing to consider which may be more inportant than mono vs colour is the software interpolation. I personally use bicubic interpolation in the DSS settings which I like the resulsts off, nearest neighbour interpolation doesn't seem as good.



#9 Starlock

Starlock

    Advanced Member

  • Moderators
  • 291 posts
  • LocationSuffolk, England

Posted 26 August 2013 - 12:09 PM

You will get an effective increase in sensitivity when you collect Luminance data with your naked CCD.  You'll be collecting all the photons that fall on each pixel instead of the approximately 1/3 of the photons that the Bayer filter on each pixel will pass.  As already mentioned,you will see a difference in resolution, but not sensitivity, for R, G and B data.

Thanks Rick, very well summarised indeed :) I agree +++ :)



#10 Neil

Neil

    Administrator

  • Administrators
  • 1,493 posts
  • LocationThe Netherlands

Posted 26 August 2013 - 01:39 PM

One thing no amount of modding can change is the CCD's response the the incoming signal, its quantum efficiency. Cooling helps to lower the noise floor, changing the signal to noise ratio.

 

I agree with Rick's nice summation on the subject and will add the following point.

 

One of the reasons mono CCD are better is that you can choose whether or not to image the LRGB components and on which nights, hence avoid some (not all) of the gradients induced by the Moon and other light sources due to the weather conditions and any low/high level mist. With a colour CCD sensor you don't have that choice at all and get everything at once.

 

So in the end flexibility is the key weapon of the mono imager L,R,G,B,Ha,OIII & SII; but its very expensive!!!

 

I know the guys at JTWAstronomy, they polished my newton tube rings and made me a nice mounting plate as well. They completely rehouse the DSLR in a new cooled aluminium box etc...

 

Clear Skies

 

 

Neil.



#11 klaatu2u

klaatu2u

    Advanced Member

  • Members
  • PipPipPip
  • 283 posts
  • LocationPortland, Oregon, USA

Posted 26 August 2013 - 03:14 PM

There was a fellow some time back who had a debayered DSLR or two on Astromart, I sent him some PM's and he kindly replied.  he had posted some great Ha widefield and I was very impressed.  He felt that the cooling was a huge factor (as and far as I remember his weren't mod for that) for longer exposures.  What makes me think this an interesting option that the 16803 has 9um pixels and cost here (US) ~ $10k...


  • Neil and Starlock like this

Jim in Portland, OR.  Some images I've taken

14" Orion goto Dob that amazes me for visual, Lots of eyepieces, and a few binoculars

A few nice refractors and 8300M CCD with full set Astrodon filters 5nm Ha, 3nm SII and OIII -->Veloce RH200 f/3 on order :)

A 'ROR' and a concrete pier in my white LP zone backyard observatory where I image from (yes, in the city!)

TheSky6, Skytools3 pro, MaxImDL pro, Photoshop CS5, PSP7 with PhotoX2, many shelves of Guides, Atlas, handbooks, you name it...


#12 Starlock

Starlock

    Advanced Member

  • Moderators
  • 291 posts
  • LocationSuffolk, England

Posted 26 August 2013 - 04:03 PM

Hi Chris ,

 

I like the look of their Ultimate Deep  Cooled 1100d too , the dark frames are staggering .

 

http://www.jtwastron...s/ultimate.html

Holy Moly these look like amazing cameras and they don't cost much more than a Atik 314L! The cases are also beuatiful which is not always the 'case' with hypermodded cameras.

 

Thanks for the link Steve :)



#13 Starlock

Starlock

    Advanced Member

  • Moderators
  • 291 posts
  • LocationSuffolk, England

Posted 26 August 2013 - 04:16 PM

There was a fellow some time back who had a debayered DSLR or two on Astromart, I sent him some PM's and he kindly replied.  he had posted some great Ha widefield and I was very impressed.  He felt that the cooling was a huge factor (as and far as I remember his weren't mod for that) for longer exposures.  What makes me think this an interesting option that the 16803 has 9um pixels and cost here (US) ~ $10k...

Yes this is the thing I can't help thinking :) my Canon 300D has 6.3 MP and the pixel size is 7.4um. It cost me 29 pounds from Ebay, I already have a couple PC cooling fans with heat sinks laying around so potentially quite a saving on the equivalent spec CCD camera (putting the inherrent quantum efficiency of the chip aside as Neil points out). I'm not saying it would be as good as the equiv spec CCD but it wouldn't be 100 times worse and it would be 1/100th the price.



#14 Neil

Neil

    Administrator

  • Administrators
  • 1,493 posts
  • LocationThe Netherlands

Posted 27 August 2013 - 03:23 AM

Yes this is the thing I can't help thinking :) my Canon 300D has 6.3 MP and the pixel size is 7.4um. It cost me 29 pounds from Ebay, I already have a couple PC cooling fans with heat sinks laying around so potentially quite a saving on the equivalent spec CCD camera (putting the inherrent quantum efficiency of the chip aside as Neil points out). I'm not saying it would be as good as the equiv spec CCD but it wouldn't be 100 times worse and it would be 1/100th the price.

Exactly, you can already achieve a lot with a HA modded and cooled DSLR, then decide if you want to risk the mono conversion! This is definitely a low cost, mid performance entry point for anyone who is considering imaging and would like a larger CCD sensor.



#15 klaatu2u

klaatu2u

    Advanced Member

  • Members
  • PipPipPip
  • 283 posts
  • LocationPortland, Oregon, USA

Posted 28 August 2013 - 03:33 AM

Just located my email copy - the fellows name is Jim Chung and i believe he may be from Canada but unsure - he was selling some modified Nikon DLSR's. My correspondence was in January of this year. 


Jim in Portland, OR.  Some images I've taken

14" Orion goto Dob that amazes me for visual, Lots of eyepieces, and a few binoculars

A few nice refractors and 8300M CCD with full set Astrodon filters 5nm Ha, 3nm SII and OIII -->Veloce RH200 f/3 on order :)

A 'ROR' and a concrete pier in my white LP zone backyard observatory where I image from (yes, in the city!)

TheSky6, Skytools3 pro, MaxImDL pro, Photoshop CS5, PSP7 with PhotoX2, many shelves of Guides, Atlas, handbooks, you name it...


#16 Nathancef

Nathancef

    Newbie

  • Members
  • Pip
  • 1 posts

Posted 22 April 2015 - 08:31 AM

Thank you : And one more question. Can i use web cameras for Motion Capture?like Logitech




Thank you for your interest in our Astronomy Forum...
It looks like you haven't joined us yet?
Why don't you register to get all the features of being a member! Its free!