Debayering a DSLR's Bayer matrix.

[Luis Campos]

BTW Louis, would you be so kind as to email me some of your flats? I would like to do a comparison, so we can figure out the best method.

Cheers Mark,

Many thanks for the tutorial, GREAT write up

It's great to work this solutions amongst all of us, so many cool ideas flowing, this is great

I uploaded a Flat image made with my 350D and 200mm f/4 newt. and also the ghost nebula TIFF raw file if anyone wants to have a go:

Ghost nebula 19x8 min. @ 800 ISO:

https://www.dropbox.com/s/dmw9i6xlz7gj9mt/Praying%20ghost%2019x8%20min%20800%20ISO%20%28sigma%20median%203%29.tif

Flat image:

https://www.dropbox.com/s/jky75u9n011vuk5/flat%20mono%20350D%20200mm%20f4%20newt..tif

Cheers,

[JamesF]

I attacked it with a hammer and chisel

Gina!!

James

[Gina]

Herewith, photo of partly debayered sensor.

OK so I didn't take enough care with the gold wires but this was a dud anyway There are still a couple of parts with gold wires intact so I can try epoxy in those areas and see how that goes.

Edited August 10, 2013 by Gina

[Gina]

It's great to work this solutions amongst all of us, so many cool ideas flowing, this is great

I uploaded a Flat image made with my 350D and 200mm f/4 newt. and also the ghost nebula TIFF raw file if anyone wants to have a go:

Ghost nebula 19x8 min. @ 800 ISO:

https://www.dropbox....a median 3).tif

Flat image:

https://www.dropbox....mm f4 newt..tif

Cheers,

Here's a quick stretch with Ps Resized and saved as PNG then uploaded here...

Edited August 10, 2013 by Gina

[Gina]

Been looking for slow setting epoxy resin - how about this standard Araldite? - says it give a long working time - http://www.ebay.co.uk/itm/ARALDITE-PRECISION-STANDARD-Epoxy-Resin-Adhesive-Extra-Strong-Glue/160921674426?rt=nc&_trksid=p2047675.m1851&_trkparms=aid%3D222002%26algo%3DSIC.FIT%26ao%3D1%26asc%3D286%26meid%3D476560980903253599%26pid%3D100005%26prg%3D1048%26rk%3D5%26rkt%3D5%26sd%3D271132932907%26

[Luis Campos]

Herewith, photo of partly debayered sensor.

OK so I didn't take enough care with the gold wires but this was a dud anyway There are still a couple of parts with gold wires intact so I can try epoxy in those areas and see how that goes.

You're getting there Gina

I also had a firt try with a practice one....poor sensor, but it took me on the right path.

Yep, the epoxy sollution seems the most secure way, I guess any good quality slow setting epoxy will do the trick.

[Gina]

Thanks Luis

[Rottweiler]

Been looking for slow setting epoxy resin - how about this standard Araldite? - says it give a long working time - http://www.ebay.co.u...d=271132932907

Looks good

[Gina]

Thank you I'll order some.

[Gina]

Ordered - should arrive Tuesday

I've used Araldite for "donkey's years" and always found it excellent. More recently I've used the newer "Rapid" which also works well where you want a quick job and don't need any adjusting time. I've found it becomes more "runny" if it and the work is warmed slightly but this reduces the working time as the chemical reaction is speeded up. Does mean though that you can run it into joints that can't be parted much to apply normally. I would think that warming the sensor slightly would encourage the epoxy resin to run into the gaps between the wires better and give lower risk of damage. And slight warming of the resin mix will make it easier to apply as drops.

[Gina]

Yesterday evening I scraped most of the remaining CFA off the sensor and today I've taken a couple of close up views of the sensor then stretched and sharpened them in Ps plus cropped to just the sensor area and resized for upload. This gives the maximum resolution possible in the posted images.

The first photo was taken with diffuse lighting (normal room daylight) and the second with a concentrated light from anglepoise lamp shining directly on the sensor and being reflected off it straight into the camera. The latter shows scratches up best. I'm a bit puzzled by the green appearance in the 2nd photo and wondering if this is just a spurious lighting effect or a thin layer of remaining CFA. Without a working sensor I can't tell if the scratches would affect the image or not. Sometimes you can have surface scratches the look awful under certain lighting conditions but which do not show in the image.

[Rottweiler]

I'm a bit puzzled by the green appearance in the 2nd photo and wondering if this is just a spurious lighting effect or a thin layer of remaining CFA.

Looks promising, I think you might be there if you can avoid the scratches . Can't really tell with the lighting, the photos I took also have a green hue and there is definitely no bayer left. The easiest way to see if the bayer still is there is to put it back to the camera and take a test shot. If there is any bayer left you will get a dim colour image. Or you can make a test swipe with a cotton bud and polish, you will get a green spot on the cotton bud. The colour in the matrix seems to be very powerful, even a trace of it will cause a visible stain.

[Gina]

Looks promising, I think you might be there if you can avoid the scratches . Can't really tell with the lighting, the photos I took also have a green hue and there is definitely no bayer left. The easiest way to see if the bayer still is there is to put it back to the camera and take a test shot. If there is any bayer left you will get a dim colour image. Or you can make a test swipe with a cotton bud and polish, you will get a green spot on the cotton bud. The colour in the matrix seems to be very powerful, even a trace of it will cause a visible stain.

Thanks I'm going to improve my copper "tool" - clean up the edge with fine emery and maybe then an oilstone (if I can find it). I tried a cotton bud and polish and it was clean - no green. I think it's a lighting effect. The green layer came off very positively with the scraper leaving the usual orangy/gold colour.

I have the working 1100D sensor off the camera and the sensor cover glass exposed ready for action but I think I'll practice on my test 350D first - £65 of risk rather than £200. I'm reasonably confident that if I take the heating process a bit more carefully I should be able to avoid cracking the glass.

[Rottweiler]

Very strange that the sensors in both our pictures appear green... Have you annealed the copper? It will make it softer. If you can't find your oilstone, get a diamond lapping tool. They are on ebay for under a tenner.

I also would like to apologise about the comment I made about avoiding the dremel to the OP, I didn't mean to offend. The dremel will work and is much faster than the hand methods, it's just very difficult to get a flat surface using a cylindrical tool. Especially as it is so quick to remove the bayer. Although doing the first 90% with the dremel and finishing by hand could offer the best of both worlds. We also tried the dremel way before we figured out to protect the gold wires, which ended badly for the sensors With them protected by epoxy you can also get right up to the edges too.

Here is some more motivation for people tempted to try this mod out.

[alistairsam]

Hi Rottweiler,

Could you give us details of the image, looks great btw. camera, scope, exposure, cooling...

sorry to repeat, any advise on an err70 on a 600D? i debayered it and as far as I can see, I haven't knicked the wires, but it gives an err 70. i've ordered a replacement, so will know for sure.

also with the spring loaded sensor, how critical is it to re-align perfectly?

Gina, I would suggest a plastic tool rather than a copper tool. those scratches don't look too good. can you see them when you see by eye?

This tupperware citrus peeler seems ideal as it has a flat end. just need to narrow it and it should be better than the wooden tool and softer than copper.

Cheers

Alistair

[Gina]

Very strange that the sensors in both our pictures appear green... Have you annealed the copper? It will make it softer. If you can't find your oilstone, get a diamond lapping tool. They are on ebay for under a tenner.

No, I haven't annealed the copper - I could do that. I'll take a look at those lapping tools

The dremel will work and is much faster than the hand methods, it's just very difficult to get a flat surface using a cylindrical tool. Especially as it is so quick to remove the bayer. Although doing the first 90% with the dremel and finishing by hand could offer the best of both worlds. We also tried the dremel way before we figured out to protect the gold wires, which ended badly for the sensors With them protected by epoxy you can also get right up to the edges too.

Yes, I found the Dremel method was quicker on my 1100D but yes, impossible to get a flat surface and I think it took off more than the CFA in places. OTOH I found the CFA came off really easily with the copper tool and apart from some scratches, didn't appear to touch the underlying layer.

And yes to the gold wire protection scheme - all too easy to catch them however careful you are.

[RAC]

I also would like to apologise about the comment I made about avoiding the dremel to the OP, I didn't mean to offend. The dremel will work and is much faster than the hand methods, it's just very difficult to get a flat surface using a cylindrical tool. Especially as it is so quick to remove the bayer. Although doing the first 90% with the dremel and finishing by hand could offer the best of both worlds. We also tried the dremel way before we figured out to protect the gold wires, which ended badly for the sensors With them protected by epoxy you can also get right up to the edges too.

Haha no need to apologise, don't worry I never took it that way at all.

I love the look of that scope you guys are making, I liked your facebook page a long time ago so have been watching with interest all the cool things you have been upto.

[Gina]

Gina, I would suggest a plastic tool rather than a copper tool. those scratches don't look too good. can you see them when you see by eye?

This tupperware citrus peeler seems ideal as it has a flat end. just need to narrow it and it should be better than the wooden tool and softer than copper.

Cheers

Alistair

The scratches are almost invisible by eye even with a magnifier. I had to stretch and sharpen the image a lot in Ps to get them to show up like that.

I tried variuos plastics from various items and couldn't find anything hard enough to remove the CFA - just slid over the surface.

[flolic]

I think that only plastics hard enough to remove CFA is fiber reinforced construction plastics, just the kind the Canon camera bodies (mirror box) are made of. But I think that's too hard and will scratch layers below CFA...

[Rottweiler]

Any error code generated from the debayering is pretty much terminal, you can't undo going too deep or breaking a wire. Err 70 is "Malfunctions related to images have been detected.", if something breaks, it is usually the last thing you did to it... RAC, great to read As for realigning the sensor, it is pretty critical but you can realign it yourself with a bit of patience.

Luis, could you share what type of wood you used for your successful attempt? I had a play with a broken sensor with wood from a pallet, wood from a pencil and also some dry wood from the car park floor(lol!). No success at all for us, couldn't even get it started. Just seemed like the wood was too soft and wouldn't hold a point or edge.

Info for that image - JTW Ultimate Mono 600D, F6/460 Telescope Service ED, 6nm Astronomik h-alpha clip in h-alpha filter. No dark, no flat and no bias calibration. Image rights - Dr F Hemmerich.

SH2-129 - 2h40m of 20m exposures @ ISO1600.

IC1396 - 2x1200sec ISO1600, 2x600sec ISO3200, 8x300sec ISO 6400 (running out of night and also experimenting)

Here is another one Dr Hemmerich sent in (a very productive guy)

JTW Ultimate Mono Canon 600D, Sigma 70-200 telelens at F2.8, 6nm Astronomik h-alpha clip in h-alpha filter. No dark, flat or bias calibration.

16x 600s ISO1600

[Luis Campos]

Ok,

Yesterday I made my second debayer on a 350D sensor, I used heat to remove the cover glass, I made a cardboard mask to protect the metal frame and connector and left a "window" for the cover glass only, applyed heat in short periods (I have a big heat gun...) just half a second or so every 5 seconds interval, after a bit the clear glass just pop out, it was instantly, the glue turned white very fast almost all around the frame and the glass just popped, unfortunelly not ALL the way around but say 70% of the area...I then let the sensor cool to ambient temperature, still the remaining glue on the glass was enough for me top brake the darn glass into pieces...oh boy I was so near this time... So I ended up putting some tape over it and remove it by pieces, not an easy task by all mean, bits of glass all over the place this time, but upon inspection I couldn't see any damaged wire

So off I went to scraping the CFA, I openned a small hole with the needle and strated removing the layer with the wooden tool, all good but this time it was harder to remove and I made some more scratches (very small), when finished I reassembled everything again the camera connected (hurra) and when taking the first shot...a black image!!!

What a frustration....I disassembled the camera to the bone again, checked all connectors and ribon cables, reinspected the sensor and every gold wire (all connected and good) reassembled the camera again, took a shot...black image....ok, I think the sensor is dead, the only thing I did different this time was to apply more heat to the cover glass, I think that killed the sensor, I don't know exactly how but it did I still could touch the sensor with my bare fingers for a little time so it was not scorching hot...I just don't get it.

Any ideas guys...?

Thanks,

Edited August 12, 2013 by Luis Campos

[Luis Campos]

Any error code generated from the debayering is pretty much terminal, you can't undo going too deep or breaking a wire. Err 70 is "Malfunctions related to images have been detected.", if something breaks, it is usually the last thing you did to it... RAC, great to read As for realigning the sensor, it is pretty critical but you can realign it yourself with a bit of patience.

Luis, could you share what type of wood you used for your successful attempt? I had a play with a broken sensor with wood from a pallet, wood from a pencil and also some dry wood from the car park floor(lol!). No success at all for us, couldn't even get it started. Just seemed like the wood was too soft and wouldn't hold a point or edge.

Info for that image - JTW Ultimate Mono 600D, F6/460 Telescope Service ED, 6nm Astronomik h-alpha clip in h-alpha filter. No dark, no flat and no bias calibration. Image rights - Dr F Hemmerich.

SH2-129 - 2h40m of 20m exposures @ ISO1600.

IC1396 - 2x1200sec ISO1600, 2x600sec ISO3200, 8x300sec ISO 6400 (running out of night and also experimenting)

Here is another one Dr Hemmerich sent in (a very productive guy)

JTW Ultimate Mono Canon 600D, Sigma 70-200 telelens at F2.8, 6nm Astronomik h-alpha clip in h-alpha filter. No dark, flat or bias calibration.

16x 600s ISO1600

Hello,

My "tool" is made out of an old small paintbrush, nothing special really but it's probably a bit harder wood, on my first attempts it did absolutelly nothing about the CFA but then I remembered to crack open a small area with the tip of a hipodermic needle, (just touching the surface will crack the CFA open) and begun from there, after a couple passes the CFA begun to come out, really easy this time

[Gina]

Oh dear Luis And seemingly so near!

I've started stripping down my 350D. Got as far as trying to remove the main board but it's "putting up a fight" It seems to be loose but just won't come out. I've taken some other bits off to see if I could see what's holding it. Removed viewfinder eyepiece adjuster - took some screws out and "boing" a spring, some gears and other bits sprang off Just as well I won't be wanting the viewfinder. Also removed the base plate with the tripod bush. I see lots of soldered on screens any advice most welcome

Edited August 12, 2013 by Gina

[Rottweiler]

Gina, you will have to desolder these shields, some of them have screws under them. Luis, black image is a dead sensor, sorry buddy Seen it all too often... Usually from going too deep or a broken wire. You can get some pretty cool modern art if you go too deep but not enough to get a black image. Lots of colours! Also debayering can introduce defects like dead columns or traps.

It's pretty annoying that we have to debayer these cameras, because adding the bayer matrix is an extra step for the manufacturer. Such is the power of mass production I guess.

[Luis Campos]

Oh dear Luis And seemingly so near!

I've started stripping down my 350D. Got as far as trying to remove the main board but it's "putting up a fight" It seems to be loose but just won't come out. I've taken some other bits off to see if I could see what's holding it. Removed viewfinder eyepiece adjuster - took some screws out and "boing" a spring, some gears and other bits sprang off Just as well I won't be wanting the viewfinder. Also removed the base plate with the tripod bush. I see lots of soldered on screens any advice most welcome

Yep...seems like I lost the fight this time, oh well I knew what to expect

Gina, you mean the main board were all the connector attach? On a 350D there is a very small screw on the "left side" of the chasis near the USB port, have you removed this? You need to remove also the metal plate where the CF card sits, there 3 small screws in total, one at each corner of the main board and one at the base of this plate (Cf card).

Hope it helps,

See more info here:

http://www.lifepixel.com/tutorials/infrared-diy-tutorials/canon-rebel-xt-350d