Debayering a DSLR's Bayer matrix.

[pixueto]

So my original analysis stands then.  Removing the microlenses (and maybe some additional effect) is reducing the pixel sensitivity by a factor of very roughly 3. 

This has always been my understanding also.  All the pixels will now be sensitive to H-alpha instead of one quarter of them.  But if the price to pay is a reduction in sensitivity by a factor of 3 then not much has been gained. 

I'd love to be proved wrong so if someone has some evidence of the improvement.  I would also happily accept that resultswith  the 1100D may not be typical of what can be achieved with debayering other DSLR sensors.

Mark

Forgive me if I am saying something dumb (which it's probably the case) but... isn't the case that the removal of the CFA nearly compensates for the reduction in sensitivity brought about by the microlenses removal?

Edited April 28, 2014 by pixueto

[pixueto]

Here's one method :

http://www.google.com/patents/US8530356

Plus side, the gold wires won't react with the nitric acid.

But will it eat into the underlying silice substrate? That would be a disaster.

Edited April 28, 2014 by pixueto

[pixueto]

But will it eat into the underlying silice substrate? That would be a disaster.

[pixueto]

Here's one method :

http://www.google.com/patents/US8530356

Plus side, the gold wires won't react with the nitric acid.

Brilliant, I'm reading this patent now. Very interesting. Thanks! Nearly there with this mod.

Edited April 28, 2014 by pixueto

[Gina]

I hope someone understands all that about the BARC layer because I sure don't!

[sharkmelley]

Forgive me if I am saying something dumb (which it's probably the case) but... isn't the case that the removal of the CFA nearly compensates for the reduction in sensitivity brought about by the microlenses removal?

That is what I fear may be the case - the loss of the microlenses may have a crucial impact.  On the plus side, there will certainly be an increase in image resolution - especially for H-alpha.  I continue to watch with interest.

[pixueto]

Here's one method :

http://www.google.com/patents/US8530356

Plus side, the gold wires won't react with the nitric acid.

Well, I've read through and the nitric acid is used to make...explosives! I don't feel so brave now about heating a 69% solution to 60C. Interestingly, it can be purchased quite easily:

http://www.ebay.co.uk/itm/Nitric-Acid-69-500ml-Free-Entry-into-Monthly-Prize-Draw-Same-Day-Shipment-/140573783993?pt=UK_BOI_Medical_Lab_Equipment_Lab_Supplies_ET&hash=item20bad9b7b9

[russellhq]

So my original analysis stands then. Removing the microlenses (and maybe some additional effect) is reducing the pixel sensitivity by a factor of very roughly 3.

This has always been my understanding also. All the pixels will now be sensitive to H-alpha instead of one quarter of them. But if the price to pay is a reduction in sensitivity by a factor of 3 then not much has been gained.

I'd love to be proved wrong so if someone has some evidence of the improvement. I would also happily accept that resultswith the 1100D may not be typical of what can be achieved with debayering other DSLR sensors.

Mark

This is what I don't understand. How can the lens increase the performance by a factor of 3?

[sharkmelley]

This is what I don't understand. How can the lens increase the performance by a factor of 3?

Photons must reach the photodiodes in order to be registered.  But the photodiode itself is only part of the area of the sensor pixel - other necessary components also take up space on the sensor and photons hitting these will not be registered.  Microlenses direct almost all the incoming light onto the photodiode so the "wastage" is reduced.

There's a useful diagram on this page:

http://www.microscopyu.com/articles/digitalimaging/ccdintro.html

Maybe the microlenses do not account for the whole of that factor of 3 but I guess that losing them explains a large proportion of the effect.

Mark

[Gina]

I really don't think it's as much as 3 times.  When I was debayering and testing the results with side-by-side areas of untouched sensor, micro lenses removed and both removed and while the difference between micro lenses removed and not with CFA intact was noticeable, the improvement in sensitivity due to CFA removal seemed more to me.  I'll dig out some sample photos...  The effect is subjective and I have not actually measured it - I didn't know how to.  You really need to do measurements with a narrow spectrum and compare single pixels at the different wavelengths.  With a standard scene it is difficult to tell.

[Gina]

Here's some images image taken when I was debayering 1000D and 1100D sensors.  These are standard capture using the SD card.  When I get going onto debayering an 1100D sensor again I'll provide a better example.

3 1000D examples and 2 1100D.  I think the first 1100D image (annotated) shows that the area with CFA removed is lighter than that with normal sensor but with the colour change it's difficult to tell just how much.

[russellhq]

Gina, have you noticed a layer below the bayer matrix? When I've been debayering the 1100D, once I've gotten the green layer off with a scraper, there's a dull gold layer underneath. If I then use a polishing compound, the dull layer is removed and there's a second gold layer.

[alistairsam]

hi,

Been a while since I posted..

One of the members in IIS did a comparison with a colour and mono qhy5 and this might provide some indication although its not the same as a dslr. 

http://www.iceinspace.com.au/forum/showpost.php?p=1075166&postcount=1

and here are some replies to relevant questions on that finding

http://www.iceinspace.com.au/forum/showpost.php?p=1075315&postcount=8

Conclusion was that you'd need 1.5x more exposure with a colour sensor to get the same snr as a mono. 

Loss of signal with the loss of microlenses don't appear that noticeable with "L" channel images I've tested on my mono 350D with exposures of over 4 minutes. 

Below is a comparison I did sometime back with Lui's mono 350d Ha image vs a qhy9 mono image. the 350d was not cooled so the difference in noise is obvious but I think the details are very close and getting this result from spending less than $200 on a camera is worth it. 

https://www.dropbox.com/s/rvij6c77e9umenf/Eagle-QHY9vsmono350d.jpg

Also just noticed that Luis has taken some fantastic images with his mono 450d and 350d and LRGB as well

http://www.flickr.com/photos/37419943@N08/14009936175/in/photostream/

I have been meaning to do an LRGB with my 350D to compare with a colour 350D. 

The major benefit with a mono dslr apart from narrowband would be for the luminance channel.

using all the pixels for details in the image for the L channel would add to the detail in an LRGB image as this won't be the same as a synthetic L by combining the RGB from a colour dslr. 

My 600d mod was a failure but i'm still on the lookout for a sensor. 

as for the 1100d, the price of a new one has dropped so much, so worth the risk. 

Cheers

Alistair

[Gina]

450D example - there is a strip on the RHS where the sensor is untouched which may be compared with the centre region with CFA and micro lenses removed.  This is a flat using white light in greyscale.  CR2 image converted to TIFF and then resized in Ps and saved as PNG.  (I'm still looking for representative examples among my dozens of debayering images in various folders ).

[Gina]

Gina, have you noticed a layer below the bayer matrix? When I've been debayering the 1100D, once I've gotten the green layer off with a scraper, there's a dull gold layer underneath. If I then use a polishing compound, the dull layer is removed and there's a second gold layer.

Yes, I have.  I don't think this blocks much light though and variations in the amount of this layer removed should be able to be compensated by using flats.

[Gina]

hi,

Been a while since I posted..

One of the members in IIS did a comparison with a colour and mono qhy5 and this might provide some indication although its not the same as a dslr. 

http://www.iceinspace.com.au/forum/showpost.php?p=1075166&postcount=1

and here are some replies to relevant questions on that finding

http://www.iceinspace.com.au/forum/showpost.php?p=1075315&postcount=8

Conclusion was that you'd need 1.5x more exposure with a colour sensor to get the same snr as a mono. 

Loss of signal with the loss of microlenses don't appear that noticeable with "L" channel images I've tested on my mono 350D with exposures of over 4 minutes. 

Below is a comparison I did sometime back with Lui's mono 350d Ha image vs a qhy9 mono image. the 350d was not cooled so the difference in noise is obvious but I think the details are very close and getting this result from spending less than $200 on a camera is worth it. 

https://www.dropbox.com/s/rvij6c77e9umenf/Eagle-QHY9vsmono350d.jpg

Also just noticed that Luis has taken some fantastic images with his mono 450d and 350d and LRGB as well

http://www.flickr.com/photos/37419943@N08/14009936175/in/photostream/

I have been meaning to do an LRGB with my 350D to compare with a colour 350D. 

The major benefit with a mono dslr apart from narrowband would be for the luminance channel.

using all the pixels for details in the image for the L channel would add to the detail in an LRGB image as this won't be the same as a synthetic L by combining the RGB from a colour dslr. 

My 600d mod was a failure but i'm still on the lookout for a sensor. 

as for the 1100d, the price of a new one has dropped so much, so worth the risk. 

Cheers

Alistair

Thank you for that -= I'll look through the links with interest and reply later

[alistairsam]

thought I'd share this cooling mod/body for a mono 350d built by an incredibly talented teen. Jo, hope you don't mind..

this approach to a sealed body is a lot easier than one with the tec on the outside. 

http://www.iceinspace.com.au/forum/showpost.php?p=1025258&postcount=32

http://www.iceinspace.com.au/forum/showpost.php?p=1025256&postcount=31

Alistair

[russellhq]

thought I'd share this cooling mod/body for a mono 350d built by an incredibly talented teen. Jo, hope you don't mind..

this approach to a sealed body is a lot easier than one with the tec on the outside. 

http://www.iceinspace.com.au/forum/showpost.php?p=1025258&postcount=32

http://www.iceinspace.com.au/forum/showpost.php?p=1025256&postcount=31

Alistair

Thanks for the links, I especially like this image! Nice idea for keeping the camera sealed!

[GuillermoBarrancos]

The sensitivity to Ha is definitely increased, presumably because the spectral response of CFA red pixels falls off in the deep red.  Likewise the sensitivity to OIII is increased because the spectral line of OIII falls between the green and blue responses.  SII is even deeper in the red than Ha.  So a debayered DSLR achieves better sensitivity for NB as well as having twice the resolution in both axes.  For RGB the sensitivity may be slightly lower but the resolution is still doubled.  This is definitely a very worthwhile mod for astro use.  I have yet to compare a debayered 1100D directly with a mono CCD astro camera but would still expect the CCD camera to win, particularly with regard to noise.  But the cost of a 12mp astro camera is considerable.  Even with the cost of all the DSLRs I've bought for this project it is still only a small proportion of the cost of a 12mp astro CCD camera.  But also, of course, I enjoy the experimenting

In response to the lensing discussion.

If you look at the QE graphs of Mono CCD´s. Here is an example of the KAF8300, then you will see that the micro lenses can increase QE sensitivity up to over 40% in the visible spektrum. That´s almost double.

Tho in the far red, in general the effect is less to about 20-30% or so. Depending the sensor. Drop off is more in one sensor, less in the other.

PS. As note. An unmodified Canon 1100D has a peak QE of about 36%.

[russellhq]

I will deliver a flat frame tomorrow of the D40 and the D3000.    

The CFA has been removed with a qtip and a little patience (20Minutes?) while moving the qtip in circles over the sensor.

We are using the partly dissolved CFA as a helper, the CFA settled on the qtip on the first run over the the weakened CFA by the dichlormethane.

a toothpick,gently moved over the sensor did some good work in the areas, where the qtip was too bulky.

The last two attempts with scrubbing powder or diamond lapping paste removed the CFA VERY easily, but did various damage to the Sensor. (Scratches, "blobs" in the sensor)

Do you remember what grade the diamond paste was? I was considering trying a very fine grade (0.1um) diamond paste:

https://www.cromwell.co.uk/ENG2571000A

Or an extra fine (0.05um) alumina suspension:

http://www.agarscientific.com/general-consumables/polishing-grinding-materials/polishing-aluminas.html

This would be used to clean up the sensor after scraping to remove the remaining bits.

[Gina]

40% increase in sensitivity is about half a stop - not a lot   Though "every little helps" I guess.  I'm not losing any sleep over losing a 40% sensitivity increase

[GuillermoBarrancos]

40% increase in sensitivity is about half a stop - not a lot   Though "every little helps" I guess.  I'm not losing any sleep over losing a 40% sensitivity increase

I think this is a very nice example of just how much impact that extra 40% in sensitivity has  :

http://stargazerslounge.com/topic/214878-leo-triplet-tidal-tail-in-2-hours/

[alistairsam]

40% increase in sensitivity is about half a stop - not a lot Though "every little helps" I guess. I'm not losing any sleep over losing a 40% sensitivity increase

Hi

The increase with microlenses is actually 15% at 550nm and 10% at 650nm in the graph.

Alistair

[Gina]

Hi

The increase with microlenses is actually 15% at 550nm and 10% at 650nm in the graph.

Alistair

Very little then.

[russellhq]

Hi

The increase with microlenses is actually 15% at 550nm and 10% at 650nm in the graph.

Alistair

Here's a bigger chart, looks like a 50% gain to me.