Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Thinking about removing the CFA from my 20d... thoughts?

  • Please log in to reply
107 replies to this topic

#1 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 02 January 2015 - 01:45 PM

So, I've really been considering turning my modified (removed the IR filter) Canon 20d into a "debayered" b&w camera by removing the Color Filter Array.  I see where some have had great success, and some have had utter failure.  None seem to have ever touched a 20d though.  Has anyone reading this ever attempted this on a 20d?  I think the only way I'd try it is if I can get some sort of solvent to dissolve the array.  There are plenty out there that I have access to that might do it if I soak the sensor, so this is how I'd attempt it.

Then on the other hand, I don't know how much I could sell a modified 20d for, and put towards a CCD...  and, this takes real nice pictures just as it is, but it would be fun to have something to do monochrome with, I think.  Then again, if I do that, I'd have to invest in filters, wheels, and all kinds of other stuff to go along with that, and I'm not sure I'm ready to do that as of yet.
Thoughts?



#2 fetoma

fetoma

    Mercury-Atlas

  • *****
  • Posts: 2938
  • Joined: 26 Sep 2006
  • Loc: Southern NJ

Posted 02 January 2015 - 07:01 PM

The last 20D sold on A-mart was for $200. The last modified (Hutech) sold for $500 (2013) and the one before that for $350 (2012). Personally I wouldn't mess with it being an older camera without Live view or a newer Digic processor.



#3 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4950
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 02 January 2015 - 07:35 PM

The fact that it is an older camera is in my mind a reason FOR doing the mono mod!  Little less pressure on the wallet if the mod fails.  I attempted (and messed up) a mono mod on a 350D (the 20D's little brother) and while it was a disappointment it was not the disaster it would have been if I had damaged my 1000D or 650D.  In fact I still plan to re-attempt the 350D mono mod using a sacrificial 350D for parts...

 

Will a mono 20D or 350D be as easy to use as a more modern camera?  Of course not, but it will not be any *harder* to use then they already are and the mono sensor may give them a new lease on useable astro work!

 

I have no hints on the method though - I attempted the scrape-it-off method when I tried a year ago, but I have been out of the loop on the best ways to do the mono mod since then.  I will be interested to see if you find a solvent method that works.



#4 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 05 January 2015 - 12:55 PM

Yes - this is my thinking... I already have the camera, and it takes excellent pictures just as it is (IR filter removed).  Modding it for mono would be icing on the cake, but I do have to consider the value of it as it is.  If it were not modified, it would be virtually worthless, as I see them going for ~$75 all day long.  Modded makes it more valuable to someone that may want such a camera.  Taking the CFA off makes it valuable to me, and perhaps someone else further down the road.  The fact that it doesn't have live view is not that big of a deal; I've a genuine Canon intervalometer that I bought when I used to do film astrophotography.  The 20D will happilly snap pictures all night long with that, and focusing is a snap with a focus mask.

I may try to find a broken 20D just in case removing the CFA from this one fails.  The 90-some page thread on that other forum of removing the CFA from an 1100 is quite a read!  
I plan on putting a copper plate behind the imager and bring it out for cooling while I'm in there, so that will be another benefit.

I was just *really* hoping someone here had done this to a 20d and could share some points of interest ;)



#5 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 05 January 2015 - 05:22 PM

Attacked the sensor today...
The glass did not come off in one piece, unfortunately.  It did come off though, and looking through my loupe, I don't see any damaged wires.  There was some glass particles on the sensor, so I put a piece of scotch tape on the sensor to clean it up.  When I lifted the tape, the CFA just lifted off with it where the tape was.  At least, I think that's what happened... So I got the rest of what I think is the CFA.  I'm not sure if this camera has microlenses on it, and that's what I pulled off, but what is left is a shiny gold color with a bit of a vignette of reddish color.  I'm not sure if I overheated the sensor with the hot air gun in trying to remove the glass or not causing the reddish hue around the outside.  I don't have a lens with me, so I've not put the sensor back in to see what it looks like as of yet.  From my readings, there's supposed to be a microlens layer then the actual CFA, but the green layer that was on the imager is totally gone with no chemicals, scraping, or anything of the sort... just scotch tape.

I'll put a lens on this, put the sensor back in, and see what I can see.



#6 andysea

andysea

    Gemini

  • *****
  • Posts: 3480
  • Joined: 03 Sep 2010
  • Loc: Seattle, WA

Posted 05 January 2015 - 08:46 PM

Very interesting! I did that on an old 350D and had to scrape off the micro lenses and the bayer matrix. I got the job done but it took a lot of work. If the bayer matrix comes off with scotch tape and some heat I will redo the procedure on another sensor and replace it in the 350D.

I am curious to see how your mod turns out.



#7 fetoma

fetoma

    Mercury-Atlas

  • *****
  • Posts: 2938
  • Joined: 26 Sep 2006
  • Loc: Southern NJ

Posted 05 January 2015 - 09:57 PM

Any photos Matt?



#8 shawnhar

shawnhar

    Cosmos

  • *****
  • Posts: 9393
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 06 January 2015 - 12:32 AM

I would be VERY surprised if you actually got the bayer matrix off with scotch tape.

 Good luck though, looking forward to your results!



#9 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 06 January 2015 - 01:19 PM

Well, it looks to me that the CFA came right off with the sticky tape!  I don't think that the heat applied with the heat gun had anything to do with the ease of removal, as there was no time that the chip was too hot to not be able to hold it with my fingers.
Here's the picture of a ChromaDuMonde chart straight from the camera.  The only thing done here was to convert the raw image (CR2) to jpeg and crop a bit.
 

normal_IMG_8387.jpg

 

 

Here's the image white balanced with Adobe's raw converter.  It's at least as close to white balanced as I could get it...  No chroma processing (other than balance) was performed.  I can crank up the saturation and not see anything but monochrome.

 

normal_IMG_8387a.jpg
 

 

For reference, here's what the chart looks like in the world of color (at least close to what it looks like... this is a newer version of my chart)...

 

CDM12-300w2.jpg

 

I guess that speaks volumes for the old 20d when it comes to our purposes.

Things I have yet to do:  

-Clean the sensor

-Get some sort of glass to put over the chip

-Purge with nitrogen when gluing glass back on

-Re-assemble and wait for the moon and clouds to go away :)

Of course the autofocus is non-functional like this...  I think that I will likely see if I can remove the IR blocking coating from the original filter, and source some glass that is the same thickness as the glass that was on the chip so that I can use this handheld with an AF lens as well.

 

Any suggestions on a piece of glass to go over the chip?  I was thinking about using some kind of glass from an old camera lens UV filter or something of the sort.

 

I also have to figure out what to process the image with to actually verify for 100% sure that the bayer mask is gone.


Edited by kd4pbs, 06 January 2015 - 01:20 PM.


#10 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 06 January 2015 - 01:40 PM

Here's what seems to be the CFA stuck to the sticky tape.  I should have taken a picture before I folded the tape up on itself...

IMG_20150106_130450.jpg

 

Here's a close up of the imager, with the light from the phone camera illuminating it.  I guess the RGB diffraction either means the CFA is still there, or the small pattern of the imager is diffracting the light to split it up into primary colors.  Anyone that removed the CFA ever get a shot like this?  This is the image that makes me wonder if indeed the CFA came off, but the image from the camera seems to tell me it did... I can't see any different colors at all.  Then again, if the CFA remains, and that's what's causing the RGB pattern here, why would it be printed all the way to the edge of the silicon?

IMG_20150106_131518.jpg



#11 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 06 January 2015 - 05:07 PM

Looking at the raw image in photoshop, converting with dcraw, and a few other tools, I can indeed see that the image is certainly monochrome.  There is no CFA left!  All from sticky tape.

 

I'm thinking that eBayers will be selling a few 20d cameras now... ;)

The hardest part was removing the glass from the sensor.  All it took was some heat and an Xacto knife.  I could have probably done it without breaking the glass, but it would have been pretty slow going.

 

Now to figure out where to snake the copper through the case for cooling...



#12 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 06 January 2015 - 08:38 PM

Please, y'all... tell me I didn't accidentally discover how to remove these dang stubborn CFA layers by sticking a piece of scotch tape to it...  Please tell me that this is a freak incident, and that for some reason, my camera was the only one that can be de-CFA'd like this... ;)

 

I found an old UV block filter and removed the glass from it.  Can anyone give me any pointers for cutting this to fit?  I'm guessing I just use the glass cutter to score it, crack it along the line, sand the edges with something like 500-grit paper, and glue it on?



#13 andysea

andysea

    Gemini

  • *****
  • Posts: 3480
  • Joined: 03 Sep 2010
  • Loc: Seattle, WA

Posted 06 January 2015 - 09:09 PM

I think if you still had the CFA on the sensor you would see it in the raw image. The only thing I notice is that the sensor my debayered 350D is a gold-ish color with no prism effect.

I would also like to know what is the actual final gain in sensitivity as there is some loss due to the removal of the micro lenses.

There is definitely the advantage of NB imaging and higher real resolution. 

Keep us posted on your further findings!



#14 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 06 January 2015 - 11:42 PM

The only way I saw this color was with the point light source of the smartphone camera.  Looking at it under fluorescent tube lighting I see no rainbow effect at all.  I'm guessing it would look like a rainbow in direct sunlight too.

So, it looks as if it's likely that any camera that uses this same sensor (704W) will be able to have it's CFA removed simply by smoothing sticky tape onto the sensor (I first used my finger then got more fancy with a small flat-blade screwdriver) and peeling.  Can't be much simpler than this.

I don't know how much sensitivity will be gained, but it should definitely be quite a bit when using filters.  From what others have discovered, it's a wash if not slightly better when shooting full spectrum, but of course putting photons on 2 or 3 more photon counters at the same time when doing narrow band will have no choice but to increase the sensitivity.

What I'd like to know is how to convert from the 4x4 bayer encoded image that comes from the camera raw into an image that is simply 4x greater in spatial resolution.  Unless I'm totally off bat here...  someone help my thinking...

In a normal color image from a bayer matrix single-chip camera, we have a 2x2 pixel arrangement where there are four pixels that make up one color pixel, right?  So, if my color sensor takes say, 1000x1500 pixel color images, from what I understand, that is created by 2000x3000 pixels (that are all sensitive to the same spectrum) arranged under color filters (that help them distinguish from each of the red [1 pixel], green [two pixel], and blue [one pixel] spectrum), right? 

If so, it would be kinda fun to take advantage of that higher resolution image that can be created.


Edited by kd4pbs, 06 January 2015 - 11:43 PM.


#15 Campos

Campos

    Apollo

  • *****
  • Posts: 1266
  • Joined: 26 Dec 2012
  • Loc: Portugal

Posted 07 January 2015 - 04:33 AM

Whaaaaaaaaaaaaaaaaaaaaaaattttttt??????!!!!!! :O

 

Scotch tape?????

 

I don't know if the CFA on the 20D is softer as on the newer generation sensors, but the first experiences I did were on a 350D wich I guess should be near identical, but heck...I even tried a thin layer of epoxy and then tried to peel it off with no success, the only way I could get it out was by polishing (didn't like that method due to uneven flats) and finally settled on scraping the CFA off with a wooden tool, it works but it's really hard and time consuming work, not to mention dangerous...one deeper scratch and bye bye sensor.

 

You definately removed the microlenses and CFA, the green color layer are the color filters, you see green because there are more green filters then the others (RGGB). There are some left overs though but I wouldn't worry too much about that now as a propper flat callibration will handle that nicely.

 

To verify you true RAW I can point you some directions using the French program "IRIS" by Christian Buil, that's what I use and it doesn't get simplier then that ;)

 

Dowload it here and then I'll point it out to you on a next msg: http://www.astrosurf...s/iris/iris.htm

 

Cheers, 


  • kd4pbs likes this

#16 Campos

Campos

    Apollo

  • *****
  • Posts: 1266
  • Joined: 26 Dec 2012
  • Loc: Portugal

Posted 07 January 2015 - 04:38 AM

Talking about increased resolution, see here some examples of a debayered Canon 450D sensor using a simple 18-55mm lens:

 

https://www.flickr.c...016475/sizes/o/

 

https://www.flickr.c...086476/sizes/o/

 

https://www.flickr.c...859256/sizes/o/

 

https://www.flickr.c...244040/sizes/o/

 

https://www.flickr.c...233172/sizes/o/

 

Also note the noise levels, no "blotchy" noise anymore, images now ressemble those taken with very fine grain BW film ;)

 

Cheers,


  • kd4pbs likes this

#17 pmesquita

pmesquita

    Vendor-MonoMods

  • *****
  • Vendors
  • Posts: 505
  • Joined: 04 Apr 2009
  • Loc: Lisbon, Portugal

Posted 07 January 2015 - 05:26 AM

well, there are very interesting developments and my dear friend Luis Campos is definitely going to buy a truckload of tape tomorrow:-)

 

As for you expectancy on NB results with a monochromatic canon camera, well, I was Luis's 1st client and i have the 1st MonoMod made by Luis. A TEC Peltier cooled canon 450d @minus 15ºC from ambient.

 

On the  link here to my Flickr album you can see the results my 1st Tricolor NB image - The Rosette nebula (4 color tests) - with the monochrome 450d done a couple of weeks ago, with the Baader NB filters and an Orio Optics CT CF newtonian mounted on a SW AZ-EQ6.

 

Basically the session was done the following way:

 

-10 x 600s lights for each NB filter

-20 Flats

-10 Darks

-50 Bias

 

Capturing and processing: BYE, PHD, PixInsight and PS CS6

 

https://www.flickr.c...otos/pmesquita/

 

PS: the Crab nebula next to the Rosette was my 1st Bicolor image too, also done with the 450d MonoMod.

 

....NOW GO AND BUY YOUR STICKY  TAPES, GENTLEMEN :-)

 

best regards,

Paulo Mesquita

sintra

Portugal



#18 pmesquita

pmesquita

    Vendor-MonoMods

  • *****
  • Vendors
  • Posts: 505
  • Joined: 04 Apr 2009
  • Loc: Lisbon, Portugal

Posted 07 January 2015 - 05:30 AM

...forgot to say that all the photos on my Flickr 2nd and 3rd year album (with the exception of my crappy andromeda) have been combinations of images from the cooled RGB IR modded 450d...with the cooled MonoMod monochrome 450d using another Ha filter, the 12nm Astronomik.



#19 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 07 January 2015 - 02:03 PM

Thanks, Luis and Paulo!  Yes, I had seen your images before.  Beautiful!
Luis, this is certainly strange...  I couldn't believe it myself.  Sure enough though, I couldn't believe it either.  What's still on the sensor is some fingerprints, dust, and just the blue border outside of the sensor area.  It all cleanly and easily peeled right off with the tape everywhere I pressed the tape to the CFA.
I did play with Iris some yesterday after I put the chip back in the camera, and have it installed on my computer here.  Not the most intuitive program by any stretch, though.

Now I have my glass, and hopefully this evening I will be able to cut it to fit.  It's quite thicker than the original glass, but since there is no IR filter, it should still fit easily... I'll wait to see on that.  If not, I don't know what kind of inexpensive option I have for replacing the glass on the chip.

All said, I can't wait to hear of anyone else that has a 20d that wants to modify for monochrome to verify this isn't some crazy thing that happened because someone at the factory forgot to put the "make it tough to remove the CFA" layer on this sensor.


Edited by kd4pbs, 07 January 2015 - 02:07 PM.


#20 andysea

andysea

    Gemini

  • *****
  • Posts: 3480
  • Joined: 03 Sep 2010
  • Loc: Seattle, WA

Posted 07 January 2015 - 03:50 PM

ooookay I bought a D20 with defective autofocus, I figured a $30 investment was worth the fun. I will experiment and report back!


  • kd4pbs likes this

#21 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 07 January 2015 - 04:09 PM

Andy, that's a great investment for sure... even if only to remove the IR filter, it takes excellent pictures once modified.  An intervalometer is all that's needed.  I can't wait to see how it looks and how sensitive it is with the CFA removed!  Of course, it looks like I'll have to invest in filters now...



#22 andysea

andysea

    Gemini

  • *****
  • Posts: 3480
  • Joined: 03 Sep 2010
  • Loc: Seattle, WA

Posted 07 January 2015 - 04:14 PM

Or what you could do is shoot luminance with the mono DSLR and color with an astro modded one, then combine the data.


  • pmesquita likes this

#23 kd4pbs

kd4pbs

    Apollo

  • *****
  • topic starter
  • Posts: 1139
  • Joined: 13 Feb 2013

Posted 07 January 2015 - 04:29 PM

Great minds think alike, so I don't know why I came up with your idea as well  ;)


Edited by kd4pbs, 07 January 2015 - 04:29 PM.


#24 pmesquita

pmesquita

    Vendor-MonoMods

  • *****
  • Vendors
  • Posts: 505
  • Joined: 04 Apr 2009
  • Loc: Lisbon, Portugal

Posted 07 January 2015 - 05:13 PM

Or what you could do is shoot luminance with the mono DSLR and color with an astro modded one, then combine the data.

Even better, do as  i do: ha on the mono and then combine with the RGB. I only do luminance on the mono when I' trying to work some galaxies like the M33


  • kd4pbs likes this

#25 andysea

andysea

    Gemini

  • *****
  • Posts: 3480
  • Joined: 03 Sep 2010
  • Loc: Seattle, WA

Posted 07 January 2015 - 05:51 PM

Oh boy...the possibilities are endless...:)


  • kd4pbs likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics