Jump to content

SkySurveyBanner.jpg.21855908fce40597655603b6c9af720d.jpg

Debayering a DSLR's Bayer matrix.


RAC

Recommended Posts

Hi Dave,

I'm pretty sure I removed the matrix not only the micro lenses, take a look at my gallery and check the sensors, I remove the "green" layer and now it just shows the golden underlayer, I think the color in the image has something to do with the cameras internal processing as it is from a color camera, or I'm doing something silly on IRIS, here's what I do: I open the "decode raw files" window, drag the CR2 image and then click "CFA" and done, this gives me a mono but with the matrix visible image, for the color image you saw i did the same but clicked the "color" instead of "CFA" with IRIS.

Maybe IRIS places the matix on images, not the camera...I dont know I'm confused... :confused:

Edited by Luis Campos
Link to comment
Share on other sites

If you have removed the Bayer matrix entirely there is no way you can get a proper colour image the data simply isn't there, sorry Luis.

Link to comment
Share on other sites

Dave and all,

First of all, BIG THANKS for helping with this, I'm getting a headache trying to understand why i I'm getting these results, I'm sure I'm goofing somewere...

Latter when I get home I'll try to attach a CR2 file (I'm not on my computer now) not sure if works but if not I'll try and use Drop Box

Thanks again guys, I'de love to see this working... :confused:

Bye,

Link to comment
Share on other sites

Bad news I'm sorry to say, the bayer filter is definitely still there.

The first image has definite colour to it, and in the seconds one with the H-alpha filter you can see that only one in four pixels has signal

first tests with a plastic needle resulted on sratches (bad ones...) on the underlayer, but with yout technic (woden paint brush) it came out easilly

I wonder if in fact with the plastic needle you were removing the filter matrix, but not with the wood paint brush. I tried a wooden tooth pick and it couldn't scrape away the edges of the filter, let alone break through the surface. Also I've never managed to scratch the layer below the filters using a plastic tool, only with a metal one.

post-970-0-83378200-1373924170_thumb.jpg

post-970-0-12848900-1373924180_thumb.jpg

Link to comment
Share on other sites

Dave, it looks like there are a couple scratches near the bottom of the frame (centre), on the lamp post. Maybe some of the CFA has been removed there? What does it look like on the Ha version, maybe all 4 pixels will show signal?

Link to comment
Share on other sites

The scratches on the lamp post still only show 1 in 4 pixels, they're just brighter. They could actually be a slither of microlenses that are still there.

Searching around though there are a couple of small spots where the filters have been removed.

post-970-0-88286400-1373925982.jpg

Link to comment
Share on other sites

Thank you for the precious help Dave,

So, looks that I've only removed the micro lenses at the end hu...!!?? So the green layer that was removed arent the filters after all, looks like I have to do some more homework next weekend, I'll try to scrape the sensor in one edge just to check if it works by removing the golden layer, I suppose that's the RGB array, I mean it has to be, there's no other explanation for that...

yes, theres some small scratches that show some pixels, maybe that's enough to show what's going on??

cheers,

Link to comment
Share on other sites

The scratches on the lamp post still only show 1 in 4 pixels, they're just brighter. They could actually be a slither of microlenses that are still there.

Searching around though there are a couple of small spots where the filters have been removed.

Ahh...ok, I'l get some tough plastic tool and try next weekend to kill it or succeed :)

Many thanks Guys!

Link to comment
Share on other sites

Hi luis

Well done

for testing, you can even set the format to jpg and open it. it will look reddish, that just needs a custom white balance. so open paint in your pc, set the background to white, full screen, take an image and use this in the camera's custom white balance. after that the jpg images will look b&w

btw that option in dss is for fits files, not dslr cr2's. its the first tab. try this option as attached

the cr2 files open as mono in maxim without any change

also, can you post pics of your cooling mod. how did you avoid sensor frosting issues?

do you have an Ha filter to test narrowband?

cheers

Alistair,

As promised here goes some images of my heater system on the sensor chamber, it consists of 8- 22 Ohm resistors welded togheter, I shaped them so they could fit inside the chamber under the sensor, a couple of glue drops keeps them in place, the two wires connect to the TEC power wire and that's it, just make sure you run the heating from the begining of the session with the cooler on, been using this for quite a long time now with zero dew problems even with extremelly damp conditions ;)

post-13017-0-10968900-1373927398_thumb.j

post-13017-0-62301200-1373927417_thumb.j

Link to comment
Share on other sites

Bad news I'm sorry to say, the bayer filter is definitely still there.

The first image has definite colour to it, and in the seconds one with the H-alpha filter you can see that only one in four pixels has signal

I wonder if in fact with the plastic needle you were removing the filter matrix, but not with the wood paint brush. I tried a wooden tooth pick and it couldn't scrape away the edges of the filter, let alone break through the surface. Also I've never managed to scratch the layer below the filters using a plastic tool, only with a metal one.

I had to exert a bit of pressure and sort of dig in with my wooden tool as otherwise it would just slide off.

luis, when you were scraping it off, were the scraped off bits white or dark?

the microlenses are whitish and the cfa is a darker colour. have a look at my video again

sharpened the tool not to a point but to a wedge so it could dig in

Dave, a woooden toothpick did nothing for me either

good luck luis, your cooling mod looks more complex than this

are you using a microscope? it definitely helps

cheers

Alistair

Link to comment
Share on other sites

Hi Luis,

I opened your raw files in Maxim and as Dave and Gina mentioned, the matrix is definitely still there.

i've split your Ha file into RGB and attached. as expected, the red is a lot stronger than other channels.

here's a link to one of my tests, see how this opens in IRIS. you shouldn't see the matrix and if you split it into RGB after debayering, all three channels should be very similar with the green slightly stronger.

its still cloudy out here so no chance to test my 350D on a galaxy yet. waiting...

it might be a lot more worth if we could debayer a model with a high QE.

the models we've been looking at are between 26 and 32% QE.

the nikon 5200, 7000 have around 60% QE. if I'm brave enough, I might have a crack at them. but not just yet.

this is from my debayered 350D

https://www.dropbox....gf/IMG_1225.CR2

Cheers

post-12882-0-37241700-1373940504_thumb.j

post-12882-0-72658700-1373940512_thumb.j

post-12882-0-25025500-1373940516_thumb.j

post-12882-0-61491500-1373940519_thumb.j

Link to comment
Share on other sites

I had to exert a bit of pressure and sort of dig in with my wooden tool as otherwise it would just slide off.

luis, when you were scraping it off, were the scraped off bits white or dark?

the microlenses are whitish and the cfa is a darker colour. have a look at my video again

sharpened the tool not to a point but to a wedge so it could dig in

Dave, a woooden toothpick did nothing for me either

good luck luis, your cooling mod looks more complex than this

are you using a microscope? it definitely helps

cheers

Alistair

Hi Alistair,

I didn´t have to apply much pressure to remove the green layer, it came out very easily just by scraping, and I was convinced this was the RGB array, on you video it too seems that the green layer is the only thing removed, so I assumed that was it, and yes this was too easy to be true :) the cooling modd is much more complex.

When scraping the layer there was this darkish "powder" comming out, I made a few passes and cleaned this residue with a qtip embeded in acetone, worked fine.

Then when taking the first test shots and I still saw color I was so confused...how could this be even possible??? i though it has something to do with the cameras internal processing or something, boy was I wrong...

Bottom line is, I still have to try and dig on the sensor at a corner and try to see what happens, hope to be abble to remove the darn layer, if not I give up...glad i got now a 450D to do the cooling modd too :D and then maybe I'll get a damaged 350D to salvage the sensor and put it back on line :)

Cheers

Link to comment
Share on other sites

Hi Luis,

I opened your raw files in Maxim and as Dave and Gina mentioned, the matrix is definitely still there.

i've split your Ha file into RGB and attached. as expected, the red is a lot stronger than other channels.

here's a link to one of my tests, see how this opens in IRIS. you shouldn't see the matrix and if you split it into RGB after debayering, all three channels should be very similar with the green slightly stronger.

its still cloudy out here so no chance to test my 350D on a galaxy yet. waiting...

it might be a lot more worth if we could debayer a model with a high QE.

the models we've been looking at are between 26 and 32% QE.

the nikon 5200, 7000 have around 60% QE. if I'm brave enough, I might have a crack at them. but not just yet.

this is from my debayered 350D

https://www.dropbox....gf/IMG_1225.CR2

Cheers

Thanks Alistair,

Latter I'll open this CR2 file to see what I should expect, very helpfull info ;)

So you confirm that with the wooden tip you could remove the matrix right?

Cheers,

Link to comment
Share on other sites

I've uploaded an image (please check in full size) that shows my training sensor, please is that area in white the "naked" pixels? So I have to remove the yellowish layer to get to remove the RGB array, correct? There's also some red circles that show a bit of the green layer, that was what I removed from the working sensor, so these should be only the microlenses? I thought the micro lenses were a very thin allmost skyn like aperance at the top of this green layer, and these are very easily removed, looks like gel or something.

http://www.flickr.com/photos/37419943@N08/9300505884/

Thanks

Link to comment
Share on other sites

Hi Luis,

its a bit hard to make out from the photo cause it looks like the cfa has been removed. one indication is when you scape the blue corners, it'll show up yellow or orange underneath. so looks like in the practise sensor, you have removed it but its hard to say.

best is to use a microscope cause under the microscope you can actually see the matrix.

i got a USB off ebay which did the job.

for my sensor, the cfa was a greenish layer but was quite dark when scraped off.

yeah, the whitish gel like layer is the microlense which comes off easily. the greener one was the cfa and here's a screenshot from the microscope that shows the matrix pattern and where its been removed there's no pattern.

Alistair

post-12882-0-67835900-1374041926_thumb.j

Link to comment
Share on other sites

Hi Alistair,

Thanks mate, well I guess I have no choice but ti disassemble the camera again (I'm getting an expert now : :) and see waht's going on, as I don't have a microscope, I'll try to build one with a web cam and an inverted eyepiece, I think I can get a good amplification with that.

I oppened your CR2 file and yep, that's what I was expecting to see, you can even see a rest of the cfa on the upper left corner, with green color :)

Thanks again and please, do keep us informed of your efforts, don't hesitate to ask if you need some help with the cooling ;)

Cheers,

Edited by Luis Campos
Link to comment
Share on other sites

Hi,

I debayered a 1000D sensor last night. it was a lot more difficult than the 350d. it was just hard to get under the cfa layer with the wooden tool. not sure if there is a difference in thickness of the cfa between sensors.

anyway here's a pic of the cfa partly removed Luis. when held at an angle, the cfa would look dark green and the layer underneath a golden yellow.

and those wires on the sides are so fragile. i must've accidentally passed my tool over the edge, but I ended up breaking 11 of them in a row. luckily that was a practise sensor.

i'll be a lot more careful once i get the 1000d delivered.

also with the 350d, those brown clips for all the ribbon connectors are so darn delicate, i popped one off and just couldn't get it back in. it was the tiny one at the top left.

so yeah, if you're going to keep opening the 350D, be wary of those clips.

here's a pic of the 1000d partly debayered and those 11 wires at the bottom knocked off. it was barely a touch.

still cloudy here so can't do any Ha tests.

Cheers

Alistair

post-12882-0-20800500-1374110287_thumb.j

  • Like 1
Link to comment
Share on other sites

Hi,

I debayered a 1000D sensor last night. it was a lot more difficult than the 350d. it was just hard to get under the cfa layer with the wooden tool. not sure if there is a difference in thickness of the cfa between sensors.

anyway here's a pic of the cfa partly removed Luis. when held at an angle, the cfa would look dark green and the layer underneath a golden yellow.

and those wires on the sides are so fragile. i must've accidentally passed my tool over the edge, but I ended up breaking 11 of them in a row. luckily that was a practise sensor.

i'll be a lot more careful once i get the 1000d delivered.

also with the 350d, those brown clips for all the ribbon connectors are so darn delicate, i popped one off and just couldn't get it back in. it was the tiny one at the top left.

so yeah, if you're going to keep opening the 350D, be wary of those clips.

here's a pic of the 1000d partly debayered and those 11 wires at the bottom knocked off. it was barely a touch.

still cloudy here so can't do any Ha tests.

Cheers

Alistair

Hi,

Oh man, I'm sorry to hear that :(

Yeah, those delicate golden wires are way to easy to destroy, we have to be extremelly concentrated and carefull not to touch them, glad this was the practice sensor :)

I'm starting to think that there's some variations not only amongst different models but also from the same models, by what I've seen some layers are easy to remove, and other people using the same methods just can't get it out, same cover glasses are easy to remove, others come out in pieces, on my practice sensor I removed the cover glass easilly, on the working one though it was a nightmare even using a hot air gun, I just couldn't get it out, I had to cover it with tape and brake it appart to remove it...

Those ribbon clips are brittle yes, just don't bend them at all, I use a wooden tothpick to remove them and insert them in place and a small flat screwdriver to open/close the plastic levers, I also modded my "new" 450D, much easier this one, next month I'll do the cooling modd on it, I'll be on vacation soon :):p

Well, cloudy here too this northern summer is really nuts this year, 42ºC one day and the next day it's raining... :confused::eek:

Cheers,

Luís

Edited by Luis Campos
Link to comment
Share on other sites

SUCCEESS THIS TIME!!!! HURRRRAAAA!!!! :grin::p:laugh::shocked:

Finally I removed the CFA, this time was for good, you can see rests of it in the full size image, very little remained but I'll leave like that for now, man can't wait for the night to arrive and shoot the firts testes in Ha, sadlly I only have a 1.2" filter, but that will do for now ;)

I'm a happy man, I'll get drunk tonight...LOL!

Cheers Rac, Gina, Dave, Allistair and all the buddys that made this possible, BIG KUDOS!

Luís

P.S. I used the wooden tool method as discribed by fellow Alistair, worked great :p

http://www.flickr.com/photos/37419943@N08/

Link to comment
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.