American Association of Variable Star Observers (AAVSO)
Wed, 01/16/2013 - 20:38
Looks like a Canon EOS series, 60D, or 450 -- unmodified cameras -- would be suitable for photometry, but not a Canon 60Da (IR filter modified) - correct?
There are two IR filters functions in the filter stack that is usually removed by the guys who provide the modification of regular DSLR. First is the IR cut (interferential) that makes a sharp cut about 700 nm. That function is needed for our photometry to avoid the pollution of all RGGB channels by the IR (the pigment filters deposited on pixels are transparent to IR).
The second IR filter function is an IR dye, kind of tinted glass. This filter reduces the IR but also reshapes the roll-off of the R channel. This is to make it compatible with the human vision response. This filter doesn't affect the B and affects a few the G channels. This is that last filter which is removed (by construction) from the 60Da, not the IR cut. By the way the camera remains able of regular photography.
Considering the related response curve, I think the photometry should be improved by removing the IR dye (only). This makes the R response somewhat nearer the Cousin's red and also should improve somewhat the G response toward the Johnson's V (on red side) (You can get some more from my paper:
The same result can be obtained from a regular DSLR by eliminating the stack (all filters: IR cut, IR dye, UV cut, anti-Moiré ) and then adding a removable IR+UV cut (like Astrodon).
Anyhow I confirm an unmodified 450D works very well in making photometry and more recent models have even less instrumental noise !
Is it save to leave the DSLR as it is (I use a EOS 1100D) and add a V-Johnson filter in the train and just do the magnitude messurment on the calibrated image? Is it not so that the V-filter will eliminate all wavelengts not needed and that the resulting combinated information from the R, G and B channels will give exact figures.
I have not tested the 1100D but it's very similar to the older 450D I know well. I think this is a very good candidate for photometry.
Our usual way is to use the output of the two G channels only. The pigment filter used on the G pixels is not far from the Johnson's V. There are just some difference (somewhat too much blue and some missing red) that are easily compensated with a classical transform or the VSF solution I describe in my paper. I think it's the best way as it provides a better signal to noise ratio than most other combinations.
We should also remember the V filter is not all what determines the pass-band, the sensor response combines with it. Even if the V filter is combined with an unfiltered astro CCD camera a transformation is usually needed. I should also say I am not very convinced by the response curve of the commercial V filters, they are generally significantly different from the best present definition of the overall V Johnson response curve.
If I understand well, your idea is to use the three channels of the DSLR combined into a kind of luminance output (L= a.R+b.GG+c.B) and have a V filter in front of the DSLR ? The result will be the math product of the two transmissions, wavelength by wavelength. I think it's possible to make it with IRIS using its luminance function. But the question is: is it possible to get the right curve with such combination ? Maybe, maybe not, I will make the simulation using my software and let you know. The issue is probably the loss of signal and a reduced SNR, the combination of the transmission of filters is in any case a lower overall transmission.
There are other possible combinations. One very good is to cut the extra blue of the green channels using a yellow filter. With a 450D the Hoya Y50 filter makes a near perfect V Johnson blue roll-off. Then adding a little of the R channel makes also a near perfect V red roll-off. The result is near perfect in term of filter response shape but the SNR is somewhat affected.
Well you have a point in avoiding as mutch as possible in the lighttrail to keep a high signal and so a good S/N.
I'm developing a lightsource for taking flats based on a OLED panel who gives a nice homogenous white glow. The OLED is good controllable with a pulse wid controller so that I gonna be in the possibility to take linearity curves of the sensor. In combination with the spectum i'll take of the light it must be perferct posible to compare the behaveyour of the tree channels from my sensor and using the combination of the logic in 'the handbook of AIP' from Richard Berry and you papers to calculate a scale factor to emulate the V-Johnson curve. I have an original V-Johnson here so I can compare my results on a artificial star in different spectra.
If it seems to have a stable and relayable behaviour, it must be the best way to mesure. If my experiments leads to a good result, I'll publish them on my site and here on aavso so that others having a 1100D can proffit from it.
Looks to me like the responses to my posts have headed off into a slightly different direction than what I wanted to know. Let's try again:
Looking only at the Canon family of DSLR's,
1) Would any stock / off-the-shelf model be suitable or perhaps even preferred for stellar photometry?
2) Because of the high prices for 1D/5D/6D/7D series models I would have to take them off my wish list. At several thousands of dollars for just one camera body and one lens in this series of cameras a person might as well buy a mid level CCD with appropriate filters and software and skip the DSLR's all together.
3) Would a 60D or 60 Da or Rebel T4i or Rebel T3i, Etc. -- all well under $2K -- be acceptable for stellar photometry? For each model what modifications, if any, would be needed to be made to the actual camera / len / filter system for this application. Filter recommendations, software, Etc.?
I think my answer was clear: You were thinking the 60Da was improper to photometry. I explained in what it is different from regular DSLR and why it's certainly even better than regular DSLR for photometry !
I also mentioned an alternative exists: this is to remove/change the filter stack from lower cost DSLR that opens several possible configurations interesting for photometry, spectrometry, astro imaging. The purpose of it is to extend the response to the deep red.
Anyhow, on my knowledge, ALL Canon DSLR are well suited for Johnson's V photometry. It's also possible to get a reasonably good B-V . No need of any extra filter for V and B-V or change of the internal filters. But regular DSLRs are unable to make U,R and I photometry. It's maybe possible to extend the photometry toward R ( I ? ) using DSLR having all IR filters removed. But this remains to be experimented, I have never seen any report about that.
Software working with DSLR raw image files: IRIS is free, AIP4WIN comes with a book, to see Citizen Sky site. This is not limitative but documented at CS.
I fully agree with your point 2 ! I even think the 60D, and specially the 60Da are too expensive for just making photometry (and even astro imaging). I think the Canon's choice of the 60D to make an "a" version is a mistake. The functions that makes a DSLR an "expert class" one and its value are not useful for astro applications. For the same price there are astro camera with less spectral limitations.
Personally, if I had to replace my 450D I will just take an 1100D = larger pixels and less expensive than all others. ( "Rebel" is a specific brand name used by Canon in the USA, unknown here ).
This is a fascinating and timely topic. I have previously used a DSI Pro III/V filter combination for my photometry, but this camera is often tied up as a guide camera. I was wondering if I could use my 550D's green data for this purpose and I'm excited about giving it a try. When reporting the data does you enter TG as the filter?
As a relative newcomer to DSLR photometry I found myself in your boat.
A fewf observations.
1. My researches before I bought my camera indicated most of the Canons are suitable for basic photometry and astrophotography but there are some limitations. You are ok as long as you stay within them. Roger has pretty well addressed this. I use a Canon T3i (EOS 600D).
2. Bear in mind what you want to do. For me the price was prohibitive but CCDs are the better choice if you will be doing Cousin's or Johnson filters. That's just the nature of the beast. Astronomical CCDs are designed for astronomy. Commercial DSLRs aren't.
3. At my level of experience modification of the camera is a bad idea not worth consideration.
4. Going beyond which camera for the moment. The real key to good photometry is in the taking and processing of the images- both the science frames and the calibration frames. If you haven't already done so- do the DSLR photometry tutorial on the Citizen Sky side. I did this before buying my camera. AAVSO is developing a DSLR manual but in the meantime this will give you an intro.
There are two IR filters functions in the filter stack that is usually removed by the guys who provide the modification of regular DSLR. First is the IR cut (interferential) that makes a sharp cut about 700 nm. That function is needed for our photometry to avoid the pollution of all RGGB channels by the IR (the pigment filters deposited on pixels are transparent to IR).
The second IR filter function is an IR dye, kind of tinted glass. This filter reduces the IR but also reshapes the roll-off of the R channel. This is to make it compatible with the human vision response. This filter doesn't affect the B and affects a few the G channels. This is that last filter which is removed (by construction) from the 60Da, not the IR cut. By the way the camera remains able of regular photography.
Considering the related response curve, I think the photometry should be improved by removing the IR dye (only). This makes the R response somewhat nearer the Cousin's red and also should improve somewhat the G response toward the Johnson's V (on red side) (You can get some more from my paper:
http://www.aavso.org/ejaavso402834 and also http://www.aavso.org/ejaavso402815 ).
The same result can be obtained from a regular DSLR by eliminating the stack (all filters: IR cut, IR dye, UV cut, anti-Moiré ) and then adding a removable IR+UV cut (like Astrodon).
Anyhow I confirm an unmodified 450D works very well in making photometry and more recent models have even less instrumental noise !
Clear Skies !
Roger
Roger,
Is it save to leave the DSLR as it is (I use a EOS 1100D) and add a V-Johnson filter in the train and just do the magnitude messurment on the calibrated image? Is it not so that the V-filter will eliminate all wavelengts not needed and that the resulting combinated information from the R, G and B channels will give exact figures.
Clear sky's
Jean Paul
Jean-Paul,
I have not tested the 1100D but it's very similar to the older 450D I know well. I think this is a very good candidate for photometry.
Our usual way is to use the output of the two G channels only. The pigment filter used on the G pixels is not far from the Johnson's V. There are just some difference (somewhat too much blue and some missing red) that are easily compensated with a classical transform or the VSF solution I describe in my paper. I think it's the best way as it provides a better signal to noise ratio than most other combinations.
We should also remember the V filter is not all what determines the pass-band, the sensor response combines with it. Even if the V filter is combined with an unfiltered astro CCD camera a transformation is usually needed. I should also say I am not very convinced by the response curve of the commercial V filters, they are generally significantly different from the best present definition of the overall V Johnson response curve.
If I understand well, your idea is to use the three channels of the DSLR combined into a kind of luminance output (L= a.R+b.GG+c.B) and have a V filter in front of the DSLR ? The result will be the math product of the two transmissions, wavelength by wavelength. I think it's possible to make it with IRIS using its luminance function. But the question is: is it possible to get the right curve with such combination ? Maybe, maybe not, I will make the simulation using my software and let you know. The issue is probably the loss of signal and a reduced SNR, the combination of the transmission of filters is in any case a lower overall transmission.
There are other possible combinations. One very good is to cut the extra blue of the green channels using a yellow filter. With a 450D the Hoya Y50 filter makes a near perfect V Johnson blue roll-off. Then adding a little of the R channel makes also a near perfect V red roll-off. The result is near perfect in term of filter response shape but the SNR is somewhat affected.
Clear Skies !
Roger
Well you have a
Roger,
Well you have a point in avoiding as mutch as possible in the lighttrail to keep a high signal and so a good S/N.
I'm developing a lightsource for taking flats based on a OLED panel who gives a nice homogenous white glow. The OLED is good controllable with a pulse wid controller so that I gonna be in the possibility to take linearity curves of the sensor. In combination with the spectum i'll take of the light it must be perferct posible to compare the behaveyour of the tree channels from my sensor and using the combination of the logic in 'the handbook of AIP' from Richard Berry and you papers to calculate a scale factor to emulate the V-Johnson curve. I have an original V-Johnson here so I can compare my results on a artificial star in different spectra.
If it seems to have a stable and relayable behaviour, it must be the best way to mesure. If my experiments leads to a good result, I'll publish them on my site and here on aavso so that others having a 1100D can proffit from it.
We'll keep in touch,
clear skies (we can use them in Belgium),
Jean Paul
Looks to me like the responses to my posts have headed off into a slightly different direction than what I wanted to know. Let's try again:
Looking only at the Canon family of DSLR's,
1) Would any stock / off-the-shelf model be suitable or perhaps even preferred for stellar photometry?
2) Because of the high prices for 1D/5D/6D/7D series models I would have to take them off my wish list. At several thousands of dollars for just one camera body and one lens in this series of cameras a person might as well buy a mid level CCD with appropriate filters and software and skip the DSLR's all together.
3) Would a 60D or 60 Da or Rebel T4i or Rebel T3i, Etc. -- all well under $2K -- be acceptable for stellar photometry? For each model what modifications, if any, would be needed to be made to the actual camera / len / filter system for this application. Filter recommendations, software, Etc.?
Regards,
Gary Wood
Hi Gary,
I think my answer was clear: You were thinking the 60Da was improper to photometry. I explained in what it is different from regular DSLR and why it's certainly even better than regular DSLR for photometry !
I also mentioned an alternative exists: this is to remove/change the filter stack from lower cost DSLR that opens several possible configurations interesting for photometry, spectrometry, astro imaging. The purpose of it is to extend the response to the deep red.
Anyhow, on my knowledge, ALL Canon DSLR are well suited for Johnson's V photometry. It's also possible to get a reasonably good B-V . No need of any extra filter for V and B-V or change of the internal filters. But regular DSLRs are unable to make U,R and I photometry. It's maybe possible to extend the photometry toward R ( I ? ) using DSLR having all IR filters removed. But this remains to be experimented, I have never seen any report about that.
Software working with DSLR raw image files: IRIS is free, AIP4WIN comes with a book, to see Citizen Sky site. This is not limitative but documented at CS.
I fully agree with your point 2 ! I even think the 60D, and specially the 60Da are too expensive for just making photometry (and even astro imaging). I think the Canon's choice of the 60D to make an "a" version is a mistake. The functions that makes a DSLR an "expert class" one and its value are not useful for astro applications. For the same price there are astro camera with less spectral limitations.
Personally, if I had to replace my 450D I will just take an 1100D = larger pixels and less expensive than all others. ( "Rebel" is a specific brand name used by Canon in the USA, unknown here ).
Clear Skies !
Roger
This is a fascinating and timely topic. I have previously used a DSI Pro III/V filter combination for my photometry, but this camera is often tied up as a guide camera. I was wondering if I could use my 550D's green data for this purpose and I'm excited about giving it a try. When reporting the data does you enter TG as the filter?
-John
Yes, if it's untransformed magnitude we report TG as the filter.
Gary-
As a relative newcomer to DSLR photometry I found myself in your boat.
A fewf observations.
1. My researches before I bought my camera indicated most of the Canons are suitable for basic photometry and astrophotography but there are some limitations. You are ok as long as you stay within them. Roger has pretty well addressed this. I use a Canon T3i (EOS 600D).
2. Bear in mind what you want to do. For me the price was prohibitive but CCDs are the better choice if you will be doing Cousin's or Johnson filters. That's just the nature of the beast. Astronomical CCDs are designed for astronomy. Commercial DSLRs aren't.
3. At my level of experience modification of the camera is a bad idea not worth consideration.
4. Going beyond which camera for the moment. The real key to good photometry is in the taking and processing of the images- both the science frames and the calibration frames. If you haven't already done so- do the DSLR photometry tutorial on the Citizen Sky side. I did this before buying my camera. AAVSO is developing a DSLR manual but in the meantime this will give you an intro.
http://www.citizensky.org/content/dslr-documentation-and-reduction
5. Don't get discouraged at what you don't know. I nearly intimidated myself out of starting but am glad now I didn't.