Debayering a DSLR's Bayer matrix.

[GTom]

Anything less than 50% (1EV) is a positive figure for narrowband imaging... H-alpha is crippled 2EV, O-III is around 1.5EV using the Bayer array.

 I just checked I CAN USE filters with my d5100 AND the Rriccardi reducer (reducer needs 76mm, -Nikon flange: 46.5mm -Baader UFC: 13mm leaves 16mm for adaptors, GREAT, I can sell an unnecessary camera now ) .

I'll collect some courage (+spare sensor...) and try your mod, including the cold finger. I was hoping JTW accept sensor assemblies for conversion but they need the whole camera - which of course involves more costs (work & shipping), thus commercial
solution seems to be a no go, I am not shelling out the price of an asi1600mm-cool on a dslr-mod... Free modding my d5100 at home is a different story:)

Edited November 20, 2016 by GTom

[Adam J]

I would strongly recommend you by a cheap busted camera off e-bay and practice on it first. Broken DSLR's can be picked up for less than 30 pounds. 

[GTom]

I'll probably buy just a sensor... 

[GTom]

In the meantime ZWO is not stopping in the production of newer and newer CMOS cameras. If their newest APSc model gets a mono buddy, I wont bother/invest time&money in debayered stuff. They effectively halved the cost of entry to the "ccd-world".

[Adam J]

Yes, given 12 months as a minimum ill be joining you, but most likely with something along the lines of the QHY163m, if nothing else the competition is likely to lower cold moss prices across the board over the next year. But until then I have what I have.

 

[GTom]

BTW, has anyone tried to deBayer the sony A7? Being fullframe, it's quite an affordable camera. However, looking at replacement sensor prices I'd probably check if a pro could do the monochrome mod...

[Adam J]

Ok manage to get some comparison shots of the pelican nebula. Seeing was not the best last night with lots of moisture about.

1 x 300 second ISO 800 H-A frame with the RGB 1000D and one frame with the Mono 1000D.

Why only a single frame? Because I wanted to take each image as close to the other as possible to mitigate against changing conditions. For this reason I am also not cooling as it would take too long to transfare the camera in and out of the cooler and achieve the desired temperature. Mono camera 13c, RGB camera 12c. So about 10 mins between the images being taken.

I have processed these in Photoshop, DSS and IRIS. Only the levels have been manipulated in each case with a slight curve adjustment with the aim of getting the best image possible out of the data. No noise reduction of other techniques. The RGB image is Red channel only converted to 16 bit gray.

The red channel level in the RGB image was a long way above the Mono camera with the histogram of the mono being fully to the left and within the read noise still.

Image 1: Full resolution Mono Image cropped and converted from CR2 to fits using IRIS then run through DSS without debrayering.

Image 2: RGB camera image, run through DSS using bi-linear debrayering. Subsequently having Blue and Green channels removed in Photoshop.

Image 3: Mono CR2 file run through DSS as super pixel debayering.

Image 4: RGB super pixel debayering in DSS, Blue and Green channels removed in Photoshop.

Problem is that with the cooling the noise drops massively....I am not sure which camera that will help more. I actually suspect that cooling will have a greater effect on the RGB camera due to the higher signal level in the red pixels.

So I have an opinion based on this data, am interested to see what other people think though before I reveal it.

Image 5: A previous picture taken with the RGB 1000D cooled to -5c. 4 x 600s exposures using a CLS filter.

Edited November 26, 2016 by Adam J

[D4N]

I think the mono camera will require a longer exposure to get out of the read noise due to the loss of the micro lenses, with four times as many photo sites picking up on the Ha light it should still have a higher total sensitivity but individual pixels are less sensitive so more prone to being swamped in noise.

I dare say that the mono camera may fare better when cooled as the noise level should be so much lower.

 

I'm a big fan of cooled cameras DIY or otherwise the difference in the noise is astounding.

[Xplode]

5 hours ago, Adam J said:

Image 2: RGB camera image, run through DSS using bi-linear debrayering. Subsequently having Blue and Green channels removed in Photoshop.

Not sure if this is what you always do with Ha taken with a RGB camera, but it's not the right way to do it.

The best way is to split the CFA 's R, G and B channels into separate images, that way you're left with half resolution mono images and there's no way the B and G noise can pollute the R channel.

[Adam J]

7 hours ago, Xplode said:

Not sure if this is what you always do with Ha taken with a RGB camera, but it's not the right way to do it.

The best way is to split the CFA 's R, G and B channels into separate images, that way you're left with half resolution mono images and there's no way the B and G noise can pollute the R channel.

Here it is your way.

Edited November 27, 2016 by Adam J

[GTom]

Thank you for the effort! I compiled a quick 100% crop from each of your shots, trouble is, that the focusing seem to went a bit rouge:(

What I see, is that the Mono image is great, sharp and the stars should be there are there. The nebula interestingly fades in the full-resolution image "thanks" to the loss of microlenses.

Binning - here deBayering OTOH bring the missed intensity back, even adds a tiny bit to the intensity. Unfortunately the focus issues didn't allow for limiting magnitude check :(.

Based on this result I'd say the conversion doesn't really boosts the sensitivity of the 1000D - which is not surprising if we look at the Canon vs Nikon (=Sony) pixel structure.

 

[Adam J]

Its quite hard to focus a narrow band filter on a DSLR, especially as I was trying to do it fast. But I went back and forth between the two cameras multiple times ill see if any of the other mono / rgb pairs have better focus.

[Adam J]

Cooled mono 1000D M45.

All in all i think that the camera breaks out about even following the mod. Still it was an interesting project

Edited November 30, 2016 by Adam J

[GTom]

On 11/30/2016 at 18:55, Adam J said:

Cooled mono 1000D M45.

All in all i think that the camera breaks out about even following the mod. Still it was an interesting project

Absolutely! Moreover, I linked above, that Nikons might have an advantage here, because of their smaller diodes (that might be the case for modern Canons too)

[Adam J]

10 hours ago, GTom said:

Absolutely! Moreover, I linked above, that Nikons might have an advantage here, because of their smaller diodes (that might be the case for modern Canons too)

Yes I agree, however...QE is a measure of the percentage of the incident photons converted into electrons and that includes the effective aperture of the pixel. So in terms of signal to noise per pixel although the Nikon cameras start out with a higher QE ~50% as opposed to Canon 40%. You have to also consider the smaller pixels in the newer cameras. I calculated it and the light collected per pixel in my 1000D at 35% QE is still more (per pixel) than the D5100 with its QE of 50%. However the D5100 has better noise so it gains quite a bit of performance from that. 

Incidentally in your link they compare the original 5D with the newer Sony sensor...not a fair comparison. The 5D had horrific QE (26%) and the 5D MK2 (31%) even in comparison to the 1000D (35%) so while the 1000D has poor QE its not so poor as the two sensors pictured in your link.

I did think about the 550D (same sensor as 600D) with QE (41%). However my logic is that if the 1000D is breaking even using super pixel, (but worse when processed for full resolution) then the 550D might do a little better than breaking even and the D5100 might do better still....but even if the D5100 is seeing a performance advantage post mono mod due to its sensor construction I would not expect it to be a huge one, i.e. it might not be sufficient performance increase to make it worth the effort and the D5100 is more expensive.

In terms of risk vs reward, independent of the camera / sensor a larger performance boost is obtained through cooling than via a mono mod in my experience.

If you get a D5100 and have a go with it, please please post some analysis here. I would be very interested.

Edited December 5, 2016 by Adam J

[GTom]

20 minutes ago, Adam J said:

Yes I agree, however...QE is a measure of the percentage of the incident photons converted into electrons and that includes the effective aperture of the pixel. So in terms of signal to noise per pixel although the Nikon cameras start out with a higher QE ~50% as opposed to Canon 40%. You have to also consider the smaller pixels in the newer cameras. I calculated it and the light collected per pixel in my 1000D at 35% QE is still more (per pixel) than the D5100 with its QE of 50%. However the D5100 has better noise so it gains quite a bit of performance from that. 

Incidentally in your link they compare the original 5D with the newer Sony sensor...not a fair comparison. The 5D had horrific QE (26%) and the 5D MK2 (31%) even in comparison to the 1000D (35%) so while the 1000D has poor QE its not so poor as the two sensors pictured in your link.

I did think about the 550D (same sensor as 600D) with QE (41%). However my logic is that if the 1000D is breaking even using super pixel, (but worse when processed for full resolution) then the 550D might do a little better than breaking even and the D5100 might do better still....but even if the D5100 is seeing a performance advantage post mono mod due to its sensor construction I would not expect it to be a huge one, i.e. it might not be sufficient performance increase to make it worth the effort and the D5100 is more expensive.

In terms of risk vs reward, independent of the camera / sensor a larger performance boost is obtained through cooling than via a mono mod in my experience.

If you get a D5100 and have a go with it, please please post some analysis here. I would be very interested.

Plans changed, I am going Full Frame. I just learned that the 6D performs the same as the D610 and sony a7 (apparently canon learn't how to properly fab sensors even with their ancient 500nm tech), thus I am decided now, I'll grab a 6D for monochrome and keep the d5100 for color. But for sure, I'll take standard shots of all kinds for a meaningful comparison.

[Adam J]

20 minutes ago, GTom said:

Plans changed, I am going Full Frame. I just learned that the 6D performs the same as the D610 and sony a7 (apparently canon learn't how to properly fab sensors even with their ancient 500nm tech), thus I am decided now, I'll grab a 6D for monochrome and keep the d5100 for color. But for sure, I'll take standard shots of all kinds for a meaningful comparison.

6D is an awesome sensor, but its also expensive. You will be a brave man to try that camera as your first mono mod. How much are the replacement sensors. I would be going for the ASI1600 at that price point myself. Also dont underestimate the effect of cooling....and that is much harder on most full frame cameras not least because of their weight. 

Edited December 5, 2016 by Adam J

[GTom]

I probably won't try it myself, I am looking for a business doing it. Hypercams - if still active - would do it for ~320 $/ a bit more than the price of 1 sensor/, but for the time being I asked europeans (astroghost & jtw) first, to keep shipping costs down. Btw, new sensors go around 260-300 $, that's at stake if I start experimenting, but I won't, unless I see a fool-proof 6d debayering guide:D

 

Edit: running circles to find a conversion service but other than CentralDS (900$!! for the conversion = that's more than the price of 3 new sensors) no luck so far...  Apparently I have to go the hard way...

Edited December 6, 2016 by GTom

[Xplode]

19 hours ago, GTom said:

I probably won't try it myself, I am looking for a business doing it. Hypercams - if still active - would do it for ~320 $/ a bit more than the price of 1 sensor/, but for the time being I asked europeans (astroghost & jtw) first, to keep shipping costs down. Btw, new sensors go around 260-300 $, that's at stake if I start experimenting, but I won't, unless I see a fool-proof 6d debayering guide:D

 

Edit: running circles to find a conversion service but other than CentralDS (900$!! for the conversion = that's more than the price of 3 new sensors) no luck so far...  Apparently I have to go the hard way...

Not sure where you got your price from, but it has to be way off...

From the Hypercams price list i got 6D mono mod is $895 + filter

That's also what i payed to have the mod done in 2015.

[GTom]

The centralDS website states the modification prices. 6D conversion costs 900$, APSc cameras cost 600$. For a comparison, JTW sells a ready cooled mono 700D for 660€.

BTW I really like the cooling concept of CentralDS, if I can get a longer sensor cable, I might do something similar.

 

It seems, just as my first IR-mod, I have to do this myself too (that time ~5 years ago the ICF removal was just as big bang as a mono conversion today. Everybody sent their cameras to MaxMax and Spencer's for the price of a new camera - on a conversion that takes you 1 hour with practically no risk).

 

 

Collecting the critical points where things can go south for myself, please correct me if wrong/key elements are missing:

- static electricity->how to ground the sensor? 

- Breaking the sensor glass / damage to the silicon by the fragments -> efficient heating or chemical removal of the glue. Any efficient chemicals out there, that doesn't damage the sensor? What about acetone, dichloromethane? 

- protecting the gold wires -> mask or glue?

- "not to touch" area on the sensor's borders. How much to spare from the process?

- polishing techniques: which materials to use? How much force to apply? How to avoid killing pixels/entire rows?

- how to seal the sensor properly once the job is done?

+Joker: is there an antireflex paint one could use on the bare sensor? I see shiny silicon definitely causes problems with reducers (sorry, german).

Edited December 6, 2016 by GTom

[Xplode]

I meant the 6D mono mod price from Hypercams that you wrote is $320...Where did you get that price?

 

The JTW site has the same price for both mono and color cooled 700D? Weird...either it's the wrong price or it's a really really good price so if you want a mono camera i suggest you get one from there...Remember it's a kit thou since they won't sell finished cameras anymore.

 

[GTom]

9 minutes ago, Xplode said:

I meant the 6D mono mod price from Hypercams that you wrote is $320...Where did you get that price?

I Just realized that was a mistake, I mixed up the mono and ICF-prices .

I don't understand JTW either, anyway they don't take 6d, so they are out of business for me. However at the CentralDS/Hypercam price point I want to learn how to do the removal properly, that I have a 90+% survival rate for my next camera...

 

BTW, you are the best person to ask regarding the 6D: how much difference do you see in narrowband images? (trying to estimate the microlens loss)

Edited December 6, 2016 by GTom

[GTom]

I was wondering, has anyone tried to put some kind of AR-coating on the mono-modded sensor? The AR-coated cover glass is of course a good thing, but it doesn't help reflection loss. Bare silicon has a reflectance of up to 30%(!), even the simplest coating would help...

[Xplode]

22 hours ago, GTom said:

I Just realized that was a mistake, I mixed up the mono and ICF-prices .

I don't understand JTW either, anyway they don't take 6d, so they are out of business for me. However at the CentralDS/Hypercam price point I want to learn how to do the removal properly, that I have a 90+% survival rate for my next camera...

 

BTW, you are the best person to ask regarding the 6D: how much difference do you see in narrowband images? (trying to estimate the microlens loss)

Not really sure about the loss because i don't have data from the same target for both cameras.

I didn't do any testing with the camera really...it was busy being stuck to my scope 

There's one thing about microlenses i'd like to mention...they give more and more benefit the smaller pixels are so with larger pixels like on the 6D the loss is probably less than on a camera with smaller pixels.

[GTom]

Other thing I just mentioned above is reflection. The microlens+Bayer array acts as an AR coating, without that, the bare silicon reflects 20+% of the incoming light! This you can clearly see, because the modded sensors are shiny. Added, that large pixels are less affected, a rudimentary polymeric AR-coating could probably curb the losses.

To my knowledge, neither of the modders (Brent and CentralDS) coat the sensor with an AR-resign. That's one more reason to learn the process.

OTOH, as I read, CentralDS claims their work is scratch-free: is there a chemical that dissolves a 6D CFA?

Edited December 7, 2016 by GTom