Debayering a DSLR's Bayer matrix.

[Stub Mandrel]

I have to wonder, if you have 8 sensors lined up, doesn't the total cost get rather close to buying a mono camera?

[JanH]

some of them cost as little as 20 pounds including postage :-)

[Gina]

Thank you Jan

[Astroboffin]

Saw this last night, superb way of protecting the little gold wires you are all talking about, using epoxy resin to encase the wires, which allows you to be a bit more heavy handed without fear of damaging them, would give a nice solid edge to scrape up too.

Of course you may already know about this idea, but I thought it was quite clever....

http://www.jtwastronomy.com/tutorials/debayer.html

AB

Edited January 22, 2016 by Astroboffin

[Gina]

Tried that but it wasn't successful even though I used slow setting epoxy to help avoid contraction and breaking the gold wires.  I'm using a 3D printed shield to protect the wires now and that seems to work well.  No broken wires

[Herra Kuulapaa]

There is a difference between bonding wires of Canon and Nikon. Nikon ones seem to be stronger and can take the epoxy better. But the cover glasses are much harder to remove, so they're no park walks either

[JanH]

I wanted to bring a good news tonight but business as usual. Still I made great progress.

Sensor no. 4: scraped with wooden stick - a lot of CFA left all over, but still working. I taped "no go zones" with office tape (transparent adhesive tape) and flooded active area with that plastX. CFA went of immediately (or within a minute) but only where I scraped away microlenses. There were some spots of sensor surface untouched and polish didn't manage to penetrate this top layer. I made a test with no. 5 with same result - Plast X doesn't get through microlenses and I have a feeling that I started removing CFA only after scraping off microlenses with polishing tool having trouble reaching the corners.

No. 4 was still working after polishing, resolution dropped from 14 to about 12.5Mpix. In the center of the sensor were some "shadows" of CFA (I didn't polish the center as much as the edges) and the very edges deserved some final touch so I repeated taping and flooding procedure while delicately scraping some more durable spots along the edge. Polishing went well, no stains left but sensor passed away. I probably scratched too close to the edge or something. doesn't give error, just white screen.

Good news before bedtime is that I succeeded with no. 6, at least sensor is glass-less and still working (cover glass shattered and I might have some bits in my eye but we are alive). Tomorrow I'll try and scrape it gently from edge to edge leaving CFA more or less intact and then making swimming pool with plastX disolving CFA. Plan sounds good, reality might be different.

[Gina]

For goodness sake use eye protection when working with glass!!  Your eyes are your most important asset - anything else is irrelevant.  Astronomy is useless if you damage your eyes.  ALSO - seek immediate medical attention if you get glass in an eye, it can do irreparable damage!!  And plese tell me you're alright - PLEASE.

[Herra Kuulapaa]

Yoy should be fine if you don't feel any pain or irritation today, but go to physician if you feel anything out of normal with it. Overall I strongly recommend proper eye protection due to the tempered nature of the cover glass -> It fragments badly and can shoot submillimeter shards with high velocity. Especially if you heat it with open flame.

I recall this was approx a 20 mm x 10 mm piece of cover glass, or what was left of it when I used the heat shock treatment on it. My point is that those smallest pieces are microscopic. You can barely see them with naked eye. They're sharp though, really sharp!

Edited January 28, 2016 by Herra Kuulapaa
image attachement issue..

[JanH]

taking dark frame now but sensor no.6 is working.

As i mentioned yesterday first I scraped some layers away but it was a mistake to start from the edges. I just didn't find the right angle and pressure to remove microlenses and to disturb CFA. I got it right just when I moved to the center.

brown layer around are smashed microlenses (or some hard layer beneath), dark brown in the middle is CFA alone and blue is clean sensor (sorry for colours, there was some custom white balance...).

 

Removing CFA with plastX went well in the center but brownish areas were tough and there are some stains left exactly in the same places.

And I can see a small scratch at the center-bottom. Loss in resolution is decent 100px in both directions (50px average per edge).

Dark doesn't look so good, making some more samples (could have been light leakage).
Eyes are fine btw.

[Gina]

I'm dying to get back to this but there are higher priorities ATM.  Much more "cutting edge" than making up electronics and writing Arduino sketches

[JanH]

dark didn't look good at all so I turned to next sensor. This one will be for normal shooting in daylight.

Same approach as earlier. Looks usable. 9min dark frame at room temperature doesn't show any major inconsistencies but it's a pity I didn't take a dark frame before I started.

after scraping:

flat (not exactly, fridge door and opened camera body = light leaking all over)

and dark, 520s, ISO 1600, (bed)room temperature, no heatsink, no cooling.

[Herra Kuulapaa]

Hmmm., funny thing that those borders are visible in dark shot. Dark frame shouldn't show anything else than noise - there may have been some light leakage through the viewfinder or something

[JanH]

it is stray light. I sealed the sensor with electricians tape (black), it's not totally light tight  but in total darkness the result is cleaner without any patterns. There is a red LED on card reader glowing constantly during exposure. I suspected that before (together with AF light and IR remote control), now confirmed.
Gotta work now.

[JanH]

It showed up that there was similar issue with my other cameras too thus I was so amazed with results from Nex 5N - LED is there too but it's not glowing during exposure, just when writing to memory card (or during start up...). This internal light pollution is not a problem when camera isn't moded - AA/IR cut filters and piezo element protect the sensor from this light. All these things are gone leaving empty space.  Apply to both Sony nex 5 and Nex C3. With this diode blackened out the dark frames are really dark (and it gets better outside or with cooling).

Nex 5 mono, 520s, ISO 1600, room temperature, colour image, WB "daylight".

[Gina]

I have removed the card reader from my modified DSLRs so the problem doesn't arise.

[JanH]

I see trouble.

Composition frame at ISO 6400, 10s. Processed from raw to mono in rawtherapee, some curves applied. Even background illumination.

 

Flat I took after some time this evening. Compare with scraping pattern from 28th of January (red, second sensor of that day).

 

Second sensor (first on 28th January, those blue-brown images):

ISO 1600, 520s, filter Sii, some clouds obscured the view for most of the time - there is not much to see (but it's easy to find M42 on the right, 300mm FL, first is 800mm). Scaled JPG, no post process.

Flat, same data (iso 1600, t=520s), it's not flat-flat but light frame with clouds (only brightest stars are visible).

And when I combined these two frames in rawtherapee I got this:

I had to set  "blur" to 0 for flat filed. As default is set 32px - just moving the pattern and not eliminating it. Looks brighter but that's because rawtherapee applied some exposure correction I didn't notice.

I don't understand why this pattern didn't show up on short exposure at high ISO (both cameras, marked S-silver, B-black in image names). Does DSS or other stacking software blur flat fields? Scraping wasn't probably the best option after all...

Edited February 2, 2016 by JanH
misleading description

[JanH]

It is in red only. I just made a test showing that with no filter, CLS filter and OIII the pattern is unnoticeable or extremely weak. in Ha and SII - bad. That very first composition frame yesterday was made with CLS.

Edited February 3, 2016 by JanH
spelling

[Icosahedron]

This is not something I'll ever attempt. I've read the first few pages of this thread and hope someone suggested later on to gently grind the sensor window off with fine grit and a little water on a flat piece of glass. Safe and easy.

 

Edited February 3, 2016 by Icosahedron
Spelling.

[JanH]

I tried to re-polish the sensor but with little success. It might be better in some places but I added some scratches too so it's time to leave it.

 

Next time I'll try polishing&rubbing method alone without scraping. Sensor from Sony Nex 5N is ready. 16.1Mpix CMOS sensor, ISO up to 25600, it's a good sensor, better than old 14mpix in original Nex 5.

Nex 5N seems to have much weaker epoxy - first time I got the cover glass off in one piece, it's chipped in several places though but only at the very edges so I believe it's reusable. I didn't have to force it and heating with soldering iron went fast and smooth. Strange as others report that newer sensor = stronger bound.

[Kokatha man]

You folks are nothing if not persistent..!

I'll admit I thought about this with a planetary camera where the area is much smaller...but this was with the latest colour cam & after some research re the need to remove the micro-lenses as well as CFA decided it was too damaging a project with more downsides than upsides...even if the camera was free!

Personally I can't see the benefit of using up a dozen or more cams even if they are only $30-$40 each because that adds up to more than a good mono camera...not that I think their is any current mono better than the ASI224MC for planetary imaging...

But I started reading & after a few pages realised that there are currently 91 pages to this thread..!!!

So I skipped to this last page...sensors dying constantly...dangerous projectiles penetrating eyes (that sound stupid, I'm sorry to say!)...& I'm still none-the-wiser as to if anyone has made a successful conversion where a significant area of the sensor is now operating in mono mode & tests (both using a light screen & of astro-targets) have shown tangible benefits...don't think I want to wade through another 80-odd pages in case I missed this vital piece of information, so if anyone could inform me or point to the relevant posts I'd be grateful..!

"Onwards & upwards" I say...except on this I'll definitely pass..!

 

Visit my website http://momilika.net/

[Thalestris24]

6 minutes ago, Kokatha man said:

You folks are nothing if not persistent..!

I'll admit I thought about this with a planetary camera where the area is much smaller...but this was with the latest colour cam & after some research re the need to remove the micro-lenses as well as CFA decided it was too damaging a project with more downsides than upsides...even if the camera was free!

Personally I can't see the benefit of using up a dozen or more cams even if they are only $30-$40 each because that adds up to more than a good mono camera...not that I think their is any current mono better than the ASI224MC for planetary imaging...

But I started reading & after a few pages realised that there are currently 91 pages to this thread..!!!

So I skipped to this last page...sensors dying constantly...dangerous projectiles penetrating eyes (that sound stupid, I'm sorry to say!)...& I'm still none-the-wiser as to if anyone has made a successful conversion where a significant area of the sensor is now operating in mono mode & tests (both using a light screen & of astro-targets) have shown tangible benefits...don't think I want to wade through another 80-odd pages in case I missed this vital piece of information, so if anyone could inform me or point to the relevant posts I'd be grateful..!

"Onwards & upwards" I say...except on this I'll definitely pass..!

 

Visit my website http://momilika.net/

I suppose it's really only worth attempting for a dslr (large sensor) where the equivalent mono camera is rather expensive. I wouldn't consider doing it myself but would pay someone with expertise to do it for me  

Louise

[JanH]

It's nothing else but success! Despite one broken foot of sensor housing (camera had impact damage), in one corner I scraped too far over the edge when polishing and plastX paste flooded some golden wires in quite many places. But sensor still works, area loss is about 150x100px leaving about 15Mpix and no stains visible!!!

This is "flat" taken with coma corrector and Ha 7nm Baader filter (I didn't bother to point the camera against evenly illuminated area thus there are some inconsistencies but nothing to worry about). Non-debayered in Rawtherapee choosing "mono". Once I close the camera dark frame will follow.

[JanH]

Mono astro camera with sensor this big costs far more than most of us can afford to spend. Cheapest Mono astrocamera I know SBIG STF-8300M with large sensor costs 2500eur and yet the sensor is only 18x14mm in size having 8Mpix resolution.  No doubt it has better parameters for astroimaging (it's cooled CCD after all).

When considering this mod keep in mind that resolution will be reduced by about 0.5mm on each site (I don't dare to go closer and risk destroyed sensor for gaining 5 more pixels). This loss is bearable with aps-c sensor - it can be as little as 50px per site. Do this with VGA camera or even with HD the loss would be too great not talking about practical issues like how to get there (on the other hand for planets you don't need the corners, just center...). On aps-c I can do long strokes, circles.

Here goes dark, Nex 5N, ISO 1600, 520s, room temp, color image from raw, just cropped (for some reason camera generates JPGs in reduced panoramic resolution, old firmware??).

 

with some luck I give it a try tonight (after midnight might be 2hour window).

 

 

[Icosahedron]

Here is another crowd working on this. They report some success with chloroform and two matchsticks I suppose .