Debayering a DSLR's Bayer matrix.

[Kurdistan]

I checked all connections and it seems I have damaged the camera (I forgot to remove the battery the second time). However, I used another canon 30D and the sensor works well. It is not very clean and there are some residues which needs to be removed. Here is a picture of the white background of a door:

Edited July 1, 2018 by Kurdistan
Typo

[Gina]

Good progress   Get a good magnifying glass (or even microscope) to see those remaining bits.  I wouldn't worry about the edge - it's all too easy to go a bit too far into the blue area (or gold wires).

[Kurdistan]

Here is a flat after cleaning the sensor. It had a lot of scratches, so I removed one layer of silicon completely. It doesn't look bad now. The main problem that I am facing now it cleaning the sensor. It does not get clean completely and there is always some particles there! Even if I do it in a clean room it will come back later as the sensor has no filter or cover glass in front of it now!

Edited July 20, 2018 by Kurdistan

[Kurdistan]

and here is the second sensor that I debayered. this one looks better and there is no scratches. However, particles are still there!

 

[Kurdistan]

Gina, I think you can fix your previous sensor that was not flat. Y mentioned you used a solvent. I think a tiny amount of glue is solved and spread over sensor. The residue from glue makes it like it is not flat. Remove the residue using a solvent (an alcohol) that doesn't attack silicon. I am sure it will be fixed. If not, remove one layer of silicon as I did. I think it will lose sensitivity, but still works.; better than sitting somewhere and gathering dust :). 

[Gina]

I moved on from DSLRs for astro imaging years ago.  It was an interesting exercise but way in the past now.

[Adam J]

I got one of these working well a couple of years back but to be honest its not that much better for narrow-band due to removing the micro-lenses. It is much better for getting luminescence though.  

[michael8554]

Why?

Aren't all the sites recording the narrow band now instead of the nearest colour one?

Michael 

[Gina]

Yes, all the pixels support the full visual spectrum and with the red filter removed respond to Ha.  The sensitivity was reduced by removing the micro-lenses but compensated for by removing the Bayer matrix.

[Adam J]

On 20/07/2018 at 21:34, michael8554 said:

Why?

Aren't all the sites recording the narrow band now instead of the nearest colour one?

Michael 

Its because you dont get 4 x the sensitivity by removing the bayer matrix you get about twice the sensitivity. But as you half the sensitivity by removing the micro lenses you are right back were you started. (this assumes that you bin, if you dont your worse off).

However for luminescence its different because you also collect more light per pixel where as in narrow band you actually collect almost no more light on a per pixel basis, perticually so for Ha via red pixels. You may get a bit more of a boost for OIII as it straddles the green and blue pixels without having full transmission to either. But  not that much in relation to your effort.

You could say you also gain resolution....but in all honesty if you gain resolution then you actually end up with half your original sensitivity.

I did this and did some proper scientific testing on the resulting sensor, but you dont have to believe me, sometimes its good to find out for yourself. Cooling your DSLR would be a much more worth while modification.

Edited August 5, 2018 by Adam J

[nzastronomer]

Hi all, wondering how a Fujifilm XTRANS sensor would benifit from having the color layer removed? For narrowband imaging. Anyone had any experience? 

Cheers! Tony. 

 

[Thalestris24]

13 hours ago, nzastronomer said:

Hi all, wondering how a Fujifilm XTRANS sensor would benifit from having the color layer removed? For narrowband imaging. Anyone had any experience? 

Cheers! Tony. 

 

Hi

I don't know the camera/sensor but it has been done and you can buy a ready monofied one from here. However, I don't know if it's really worth the time, effort and risk, of attempting it yourself these days. You can buy cmos mono astro cameras relatively cheaply and cooled versions for a bit more. Having a cooled camera is important for long exposure imaging preferably set point with a good delta.

Louise

[Gina]

Most DSLRs are not really suitable for conversion as cooling is usually not very effective once you get down to freezing.  As Louise says, with the advent of relatively cheap CMOS astro cameras it's really not worth it.  I spent well over £1000 trying before these cheapish astro cameras came out.  It was a very frustrating exercise - I have a large graveyard of dead DSLRs!!  I will try anything but that was really not going anywhere.  (Except for emptying the bank balance even with cameras off ebay costing just a few tens of pounds.)

[JonC]

13 minutes ago, Thalestris24 said:

Hi

I don't know the camera/sensor but it has been done and you can buy a ready monofied one from here. However, I don't know if it's really worth the time, effort and risk, of attempting it yourself these days. You can buy cmos mono astro cameras relatively cheaply and cooled versions for a bit more. Having a cooled camera is important for long exposure imaging preferably set point with a good delta.

Louise

What are the current "starter" cmos mono cameras? Just wondered what price point I would be looking at.

 

Cheers

[Gina]

Have a look here :- ZWO CMOS astro cameras at FLO

[Thalestris24]

1 hour ago, JonC said:

What are the current "starter" cmos mono cameras? Just wondered what price point I would be looking at.

 

Cheers

Hi Jon

It kindof depends on what targets you have in mind, what scope/lens you plan to use with it and whether you want a cooled camera or not. They are more expensive but perhaps better in the long run. Having said that, some of the new cmos astro cameras are sensitive enough to be able to do short exposures (lots needed) and if used mostly in the cooler, wintry months will likely give quite good results without cooling. Unfortunately, I don't actually have one but plenty of people on here do!

Louise

[Stub Mandrel]

The ZWO cameras are designed so that adding your own cooling at a later date is practical, although unlikely to be as elegant as the all-in one designs.

[Gina]

Yes, I added my own cooling to the ZWO camera I used in my all sky camera.

[Ken Mitchell]

I just debayered the sensor of my old camera, I believe all went ok. There are still some small parts left that I will clean next time.

My question is why are the images pink/blue and not black and white like in the OP? Do I need to convert it? I tried changing the wb but it always stays one color.

Thanks in advance.

Ken

 

[tooth_dr]

I took my first images at the weekend with a debayered 20d. I will post the results here.  I only got 4 x 600s as I was manually controlling. I’ve a lead ordered to automate the exposures. 

[YellowPacman]

D600/D610/D750/a7/a7-2 or any other camera with the IMX128 sensor can be debayered, the most tricky part is the glass removal.

Also you need a cleanroom to get rid of those dust/particles on the sensor for a perfect image. I still have some of it even after 7 cameras debayered. 

I aslo did some simple test of brightness with debayered d610 vs non debayered d610. Both in monochrome mode, same settings. The non-debayered one needs about (not exact) +0.7ev in lightroom to compensate exposure to make images the same. The illumination is led lamps (no uv, no ir), it is a night outside, no sunlight. Under a daylight or for the astronomy purposes (with IR or UV) i suppose the debayered one will get more than 1ev advantage.

 

Edited December 14, 2018 by YellowPacman

[angryowl]

Just thought I'd add my recent experience with debayering an ICX453AQ sensor for those interested.

The sensor in question came from a Nikon D50 which was listed as "taking black pictures" and while I hoped it was a shutter failure, it turned out it was two of the very fine golden wires inside the sensor that had come off the sensor array. While this meant the sensor was beyond repair I thought I'd open it up and have a go at debayering it since I've heard Nikon sensors are supposedly easier to tackle.

The top glass came off easily after a quick blast with a hot air gun. For debayering, I've used toothpicks to gently scrape away at the surface until both the CFA and lenses were removed. Upon close inspection, the toothpicks did not seem to scratch the underlying surface at all. But I was impressed at the sheer ease at which the CFA came off with the lightest of pressure on the sensor.

Sadly, I threw away the sensor and have no images of the final result and obviously testing the sensor after debayering would have not been possible, but just wanted to add my experience to this massive thread. 

I have also previously successfully debayered a Canon 20D sensor using this exact method, but was a lot more involved which resulted in many leftover scratches.

[Noobulosity]

Finally made it through the massive amount of posts in this thread.  Whew!  What a loooooong journey!  I actually joined just so I could post my own adventure.  Unfortunately (fortunately?), I cannot claim my spot in the DSLR Killer Club.

With so many successes sprinkled throughout the thread, and despite a heap of dead sensors from various attempts, I was inspired to try this myself.  I have an astro-modified Canon T3i that I had considered testing this on.  But, I was hesitant to use that camera.  Instead, I found a cheaper Canon 7D to test this on.  (Nearly $100 less than the T3i's typically go for right now.)

I ended up having very little trouble removing the sensor cover glass with heat and gentle prying with an X-acto blade.  When it popped off it nearly scared me, lol.  But, I just made my way carefully, slowly heating up the frame until the adhesive turned white and loosened right up.  A little extra lift from a thin blade and it popped right off.

From there, the hard part was removing the micro-lenses and CFA layer.  These were quite hard.  I started with a simple toothpick, which had trouble making any headway at all.  Then I tried adding some fine polishing compound to the toothpick, and that started doing something, but it mostly seemed to be only affecting the micro-lenses.  The CFA wasn't doing anything.  Then, after seeing the post regarding an aluminum rivet, I decided to give brass rod a try.  I flattened the tip of it at an angle, and that had very little trouble making it through the CFA layer, but didn't seem to cause much trouble for the sensor beneath.  I was very careful to stay away from the edges, so there's still a CFA border around the sensor.  To get the last of the CFA off the center of the image, I used the toothpick again to polish off the little bits left over.  Still need to do a little more clean-up, though.  (See flat below.)

I did end up with a tiny scratch or two, and one horizontal and one vertical line of dead pixels.  Fortunately those dead lines are long the perimeter, so they can be easily cropped out of the image.

Overall, I'm quite happy with how this turned out.

Progress pics:

Here you can see a bit of uneven scraping, but not horrible:

My first monochrome image!  Taken through my WO Zenithstar 103 at a distant power pole.  I opened this in IRIS and it converted it automatically to monochrome, but wouldn't let me change the debayering to get rid of the pixelation.  So, I went with DCRaw instead, and that did the trick.

A flat taken against a white monitor, using my Sigma 18-35mm f/1.8 Art lens, settings were 35mm FL, f/1.8, ISO100, 1/250s

Heavily-stretched.  You can see the scratches.  And it seems there are bits of leftover CFA on the sensor that need removing.  I can't see them with my own eye, but they must be there.  So, I'll have to clean things up a bit more.  Might just do that right through the bayonet mount with the mirror up.

 

Edited June 15, 2019 by Noobulosity

[Gina]

Well done!  Great result.  Just take care with the odd little bits and don't be tempted to go any nearer to the edge or you might ruin all your good work.

[Noobulosity]

On 15/06/2019 at 00:53, Gina said:

Well done!  Great result.  Just take care with the odd little bits and don't be tempted to go any nearer to the edge or you might ruin all your good work.

Well...  I must have hit something.  I have a feeling that using cotton swabs is a bad idea, because the fibers can stick out to the sides and slip underneath the frame the filters used to utilize.  I must have clipped one of those gold wires trying to clean things off, because now half of the sensor works fine, and the other half is darker and has lines to it.  You still get picture on both halves, but only the one side is really usable anymore.

I ordered another sensor from eBay for $65, so I'll give it one more shot and try to be a bit more diligent with my next attempt.