Jump to content

sgl_imaging_challenge_2021_annual.thumb.jpg.3fc34f695a81b16210333189a3162ac7.jpg

Debayering a DSLR's Bayer matrix.


Recommended Posts

That's pretty impressive :)

Thanks Gina,

yeah results are promising

only thing I need to get a handle on is how to stretch these images cause I'm used to the unstretched ccd images.

skyglow and LP is also an important factor. I might stick with 400iso for anything over 2 mins

Alistair

  • Like 1
Link to post
Share on other sites
  • Replies 2.6k
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

I thought i would have a go at the impossible. I've got two Canon 1000d's so i took it upon myself to either break it or make it. I looked all over the net and found nothing at all about doing this ,

Failure is not always a total failure

Well.......  I guess I just couldn't leave well enough alone.  I picked up another sensor from eBay (my 4th one...) and went at it again.  This time, I'm really pleased with the results. My metho

Posted Images

Did you use a H-alpha filter for those images?

nope just an idas-lps-p2 light pollution filter. I'll use these as the L channel on rgb data from a qhy8l.

need to sort out my guiding and then I'll be able to take Ha images with a 2" baader 7nm Ha filter

challenge is using the oag with the mpcc and filter wheel

there's hardly any backfocus with dslr's

cheers

Link to post
Share on other sites

Hi Luis,

Those images are great all things considered!

I am planning an imaging trip for the weekend, and I will try to transplant debayered chip from my "regular" camera to my peltier cooled 450D.

I also have 1.25" filter wheel with RGBHa filters, those are too small but will be ok for a test.

  • Like 1
Link to post
Share on other sites

Hi Luis,

Those images are great all things considered!

I am planning an imaging trip for the weekend, and I will try to transplant debayered chip from my "regular" camera to my peltier cooled 450D.

I also have 1.25" filter wheel with RGBHa filters, those are too small but will be ok for a test.

hi

do you have any links to the cooling mod

Link to post
Share on other sites

Hi,

That is stunning. Amazing work. more or less like the Central DS Modification, but home built. brilliant.

I also like the silica gel ring idea.

I will attempt this a little later once I get my 600D b&W modded.

how difficult is it to design the shutter simulator? I work with picaxe microcontrollers so if you can share a few ideas on how you did it, that would be really helpful.

Look forward to seeing results from a cooled B&W 450D sensor.

Cheers

Alistair

Link to post
Share on other sites

Last night I made some imaging, I put this in the imaging forum, here goes some more good results in Ha:

http://stargazerslou...ono-canon-350d/

Cheers guys,

Luís

Hi Luis,

When I saw your eagle, I had to do a side by side comparison with an image on IIS taken with a QHY9 mono. I'm a member there as well.

I've asked his permission to use his image.

the results are stunning and shows how well the mono dslr stacks with a camera over 10 times its cost and all the other factors with cooled CCD's like much higher QE, 16 bits vs 12bits, cooling, dynamic range, pixel well depth, lack of microlenses etc.

this shows how narrowband imaging can be made extremely affordable.

the qhy9 image is 15x5 mins Ha with a 90mm AP, yours is 5x5 min with an 8inch F4.

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

Cheers

Alistair

  • Like 2
Link to post
Share on other sites

Hi Luis,

Those images are great all things considered!

I am planning an imaging trip for the weekend, and I will try to transplant debayered chip from my "regular" camera to my peltier cooled 450D.

I also have 1.25" filter wheel with RGBHa filters, those are too small but will be ok for a test.

Hi Filip,

Many thanks mate :)

Hope it all goed well with your modd, I'm curious to see the perfomance of the 450D sensor ;)

Let me tell you that you are very tallented, that's a smashing cooling modd you got there, very clean and beautifully built...if you were my neighboor I would ask you to make one for me :D LOL

Cheers,

Link to post
Share on other sites

Hi Luis,

When I saw your eagle, I had to do a side by side comparison with an image on IIS taken with a QHY9 mono. I'm a member there as well.

I've asked his permission to use his image.

the results are stunning and shows how well the mono dslr stacks with a camera over 10 times its cost and all the other factors with cooled CCD's like much higher QE, 16 bits vs 12bits, cooling, dynamic range, pixel well depth, lack of microlenses etc.

this shows how narrowband imaging can be made extremely affordable.

the qhy9 image is 15x5 mins Ha with a 90mm AP, yours is 5x5 min with an 8inch F4.

https://www.dropbox....9vsmono350d.jpg

Cheers

Alistair

Hi Alistair,

You are so right, and I used the 1.25" 12 Nm filter....I'm dying to get one 2" and 7 Nm filter...but I have to wait for a wile, it's too darn expensive :confused: :confused: :confused:

I think the 1.25" does cut light from the primary (acts as a diafragm) so this can only get MUCH better ;)

Cheers mate and keep on imaging,

I think this next weekend will be clear here in Portugal, so hopping to do a good imaging session, still ahve to decide what object I'm going to do...

Bye,

Edited by Luis Campos
Link to post
Share on other sites

Some great progress here guys :grin:

I finally managed to get a few test shots last night. Unfortunately I wasn't set up for bulb exposures so was limited to 30 seconds.

Here's a comparison of M57 with a stock 350D, the 'normal' IR filter mod (Baader ACF replacement), and then the sensor with CFA removed.

Shot with an Astronomik 1.25" 12nm H-alpha filter on an f/10 LX200, 4 x 30s, no calibration frames.

m57comparison.jpg

Also got a quick M31, 10 x 30s with 4 darks, using a 2" Baader neodymium filter:

m31debayered.jpg

Looks like no flat issues here. I still haven't finished cleaning up the sensor and it's looking pretty good at f/10

  • Like 2
Link to post
Share on other sites

Some great progress here guys :grin:

I finally managed to get a few test shots last night. Unfortunately I wasn't set up for bulb exposures so was limited to 30 seconds.

Here's a comparison of M57 with a stock 350D, the 'normal' IR filter mod (Baader ACF replacement), and then the sensor with CFA removed.

Shot with an Astronomik 1.25" 12nm H-alpha filter on an f/10 LX200, 4 x 30s, no calibration frames.

m57comparison.jpg

Also got a quick M31, 10 x 30s with 4 darks, using a 2" Baader neodymium filter:

m31debayered.jpg

Looks like no flat issues here. I still haven't finished cleaning up the sensor and it's looking pretty good at f/10

nice work, that m31 looks awesome.

but I think the stock 350d image might have some artefact cause it shouldn't be too different from the one with the filter cause data would only be in the red channel and it'll need to be separated. I'm guessing you already did this?

I got the parts for the serial port mod using the opto coupler and will be making a cable. but I got a timer remote for the interim from ebay and that works great.

but you really need control from the pc for focussing and centering.

look forward to more

cheers

alistair

Link to post
Share on other sites

Those are good news Dave :)

The "debayred sensor brotherhood" is growing...:D

It certainlly show improved resolution and sensitivity on the debayred sensor, the IR mod 350 shows pixelated data because of the 1/4th resolution compared to the mono sensor, put some more time on it when you can.

Congrats, keep us updated plz

Luís

Link to post
Share on other sites

but I think the stock 350d image might have some artefact cause it shouldn't be too different from the one with the filter cause data would only be in the red channel and it'll need to be separated. I'm guessing you already did this?

I should have mentioned I stretched the images seperately so that the background and nebula were around the same brightness, perhaps not the best way of doing it but it shows the signal strength above the noise. The stock camera should be around 16x less sensitive to H-alpha than the debayered one, as the IR filter block 75% of that wavelength, and the CFA blocks another 75%.

Of course the dark and bias noise are not affected by these mods so should be pretty much the same. The one with just the IR filter changed might actually be less noisy than the others as it has a cold finger attached (cooler not on, but would still draw the heat away). The stock and debayered one are actually the same camera, I just swaps the sensors and IR filter during the run.

Link to post
Share on other sites

Those are good news Dave :)

The "debayred sensor brotherhood" is growing... :D

It certainlly show improved resolution and sensitivity on the debayred sensor, the IR mod 350 shows pixelated data because of the 1/4th resolution compared to the mono sensor, put some more time on it when you can.

Congrats, keep us updated plz

Luís

Maybe a new section should be created here on SLG for mono converted DSLRs? So everyone with this mod can post images and discuss? How do we get in touch with SGL to request this to be taken into consideration?

Link to post
Share on other sites

Well, I've been busy today removing as much of the Bayer array in the1000D as I could. This is what I found out today; first the positive bits:

-The scratches that inevitably (I think) are left on the sensor don't show up in the flat frames (or at least 99% of them won't come up if they aren't too bad). That's good news. Here you can see a picture of a typical scratch I left on the sensor (I left many like that one). I also include a flat frame at F36... yes F36 ! for you to inspect it.

-The glass covering the sensor in the 1000D is very, very easy to remove intact. I did it with a xacto razor very carefully and it took me 10 minutes.

-I checked for dead pixels and the camera is fine so I don't thing those scratches will have an impact on the quality of the photos.

Now the negatives:

I didn't manage to clean completely the sensor and when I went back to the bayer array leftovers, nothing seemed to work. The only thing that would remove them was a needle used very, very gently without much pressure applied. Now, some bigger scratches were left with the needle. I can tell how bad those scratches were or whether they would show up because I only tried this method with my canon 450D sensor (the practice one).

You can see in the flat that the sensor isn't very clean where in some areas I have been more successful than others. I thought that it would be a piece of cake to remove those tiny bits with the microscope but nothing seems to remove them apart from metallic things which would surely damage the sensor. Has anybody here been successful at this? Can somebody hep with this? Any ideas? I even used acetone with the practice sensor but it only removed the mocrolenses and left the surface uneven so I'd try to avoid using any chemicals.

Finally, How do you process the raw file guys? The images are appearing with some kind of artifacts/interpolation? They appear with a red tinge. Are those raw files not real, completely raws?

Thank you all for your help.

post-18331-0-75583800-1375502614_thumb.j

post-18331-0-96610500-1375502643_thumb.j

Edited by pixueto
  • Like 2
Link to post
Share on other sites

hi

well done

to fix the red tinge, bring up a whitescreen on your laptop or monitor, take an image of that and choose that image for a custom white balance.

I'll suggest saving it as raw+jpg.

I open the raw files in maxim and save as fits to stack in dss cause dss will always try to debayer cr2 files but can read fits files as mono.

as for the small bits, only way I've gotten rid of them is by repeatedly scraping on top of it with a wooden or plastic tool under a microscope until you see it dissappear. its like wearing it down.

the 350d is brittle and comes off easily but the 1000d, 600d and 450d are similar and leave these small bits behind.

  • Like 2
Link to post
Share on other sites

Been giving this some more thought with the results of debayering looking so impressive :) I love being on the "I'm doing something different from the mob" bandwagon :D There are several areas that would need careful thought. Aside from that there are two ways to go as I see it :-

  1. Go with a known solution with a 1000D body available at around £100. Cons :- A slightly lower spec camera with fewer data bits. It would need all the work to strip it down and fit cold finger and Peltier cooling which is already done with the 1100D.
  2. Wait until I find a way of removing the cover glass from the 1100D. I think I still have the dead sensors put away in a box somewhere. I have a hot air gun and could try the "heat treatment" :D It would be fun to try removing the CFA with a pointed stick too.

With either option there remains the need to provide big enough filters. I have a 5 hole home made filter wheel which takes 36mm unmounted filters which are big enough to cover the APS-C sensor. I also have Baader Ha and SII NB filters (I was unable to obtain an OIII 36mm filter). There remains a BIG problem with a filter wheel of spacing - I can only just about get enough with CCD cameras with 13mm back focus so a Canon EOS would need a lot more stripping down than I've done as yet. An alternative might be a combined EOS camera and filter wheel in the same casing.

Perhaps those who have tried 1.25" filters can say how bad the vignetting is - maybe we could have an uncropped (just resized for posting) image please? :)

Edited by Gina
  • Like 2
Link to post
Share on other sites
There remains a BIG problem with a filter wheel of spacing - I can only just about get enough with CCD cameras with 13mm back focus so a Canon EOS would need a lot more stripping down than I've done as yet. An alternative might be a combined EOS camera and filter wheel in the same casing.

Would fitting an EOS bayonet fitting straight to the back of the filter wheel get you anywhere, Gina, or is that still not close enough?

James

Link to post
Share on other sites

Gina, forgive me if I'm saying something stupid (probably) but, wouldn't re-shimming the camera sensor solve the backfocus problem? The sensor can be moved forward easily another 0.33mm in most cameras.

My intention is to use a H-alpha clip filter only and add colour from another DSLR. No filter wheel.

The problem with passing the wooden tool repeatedly over the leftovers is that you eventually risk damaging the sensor. In my case it's worst as it doesn't seem to do anything at all. It would be ideal if we could find a tool to achieve that.

I've found an easy solution to avoid going over the edges risking damaging the gold connectors. It's as simple as using a couple of straight thin lines of masking tape at the back of the sensor plastic cover; the bit that attaches to the metal housing of the sensor. The receptacle for the LP2 filter has some in this way anyway. I'll do it if I manage to clean the bits of bayer array.

Everything else failing, maybe the best bet would be be to stick to the 350D for monochrome conversion since, apparently, there is a fundamental design difference in the way the array was glued and there is a better chance of removing the array completely whilst not damaging the sensor. The 350D still is an awesome camera -pity the lack of live view! Problem is, nobody seems to have been able to remove the glass over the sensor in the 350D without breaking it, right?

Edited by pixueto
Link to post
Share on other sites

Gina, forgive me if I'm saying something stupid (probably) but, wouldn't re-shimming the camera sensor solve the backfocus problem? The sensor can be moved forward easily another 0.33mm in most cameras.

My intention is to use a H-alpha clip filter only and add colour from another DSLR. No filter wheel.

Unfortunately we're talking of 10-20mm to be saved. I sold my Ha clip filter :D Anyway, I have a Baader 7nm Ha 36mm.
Link to post
Share on other sites

hi

well done

to fix the red tinge, bring up a whitescreen on your laptop or monitor, take an image of that and choose that image for a custom white balance.

I'll suggest saving it as raw+jpg.

I open the raw files in maxim and save as fits to stack in dss cause dss will always try to debayer cr2 files but can read fits files as mono.

as for the small bits, only way I've gotten rid of them is by repeatedly scraping on top of it with a wooden or plastic tool under a microscope until you see it dissappear. its like wearing it down.

the 350d is brittle and comes off easily but the 1000d, 600d and 450d are similar and leave these small bits behind.

Thank you Alistair. Are there any free options to Maxim where I can treat the raw files?

Thanks

Link to post
Share on other sites

Nearly bought a 1000D off ebay today until I noticed it was collection only. Missed another at just over a ton. Been looking at prices of sensors as spare part - equal to or more than a new camera!! If another 1000D comes up at around £100 I might see if I can win it - just to have a play. It would be a pretty careful play though as I wouldn't want to blow a hundred quid toooo soooon :D

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.