I am attempting to do DSLR photometry at prime focus on my 12" SCT.
To avoid questions later I'll try and describe the full set up and procedure.
- 12" SCT on a eq wedge.
- Drift aligned such that a star stays centred in excess of 15 minutes with 9mm illuminated reticle.
- Pentax K10D DSLR, PEF RAW format images. 1600 ISO.
- Camera is on an off axis guider at prime focus on the scope.
- I use a mirror diagonal and eyepiece with similar FOV to the camera to frame the field then the camera on an off axis guider to do the imaging.
- I have made the 9mm illuminated reticle and camera parfocal on the off axis guider so I can just focus to the reticle and know the camera is in focus (or not).
- 30s exposures and then varies because I don't have the pentax disk with the SW on any more to set exposure times beyond what is on the camera. i.e. Manual >30 second exposures otherwise.
- I have taken bias, dark and flat images. I understand I need to take these each evening I'm imaging.
- The reason I'm using IRIS is that it handles the PEF RAW format and it's free.
-
The reason I'm using the scope and not the DSLR on a tripod is:
- Dimmer targets such as CVs, RCbs and BL Lacs interest me most.
- No spare money for new toys or SW means I have to use what I have to hand. I.e. Telescope and off axis guider.
The issue I have with IRIS depends upon what I attempt. I am using the guidelines at http://www.citizensky.org/content/iris-beginner
I have some images I've put up on boxnet along with the bias, dark and flats I've been playing with here: https://www.dropbox.com/sh/1sncy54f4vwz7eb/aMx5LNBMuN
Hopefully people can get to that.
Processing the master-offset, master-bias, master-dark and master-flat causes no issues.
If I process the 7 30 second images (see boxnet) I've taken, this is the result:
At step 4.1 in iris-beginner I set the output files type to Color. Thus later on when Iris complains about having incompatible colour and B&W images it's a total mystery as to how.
At step 4.2 stellar registration fails, less than two stars in common between the images apparently.
It seems reasonable then that step 4.3 would fail for some reason, but it complains about incompatibility between color and B&W images. Very confusing. So that's my give up point for processing multiple 30 second exposures.
If I process a single 300 second image (see boxnet), this is the result:
I get to step 5 in iris-beginner and then find that I can separate the channels but cannot adjust thresholds. The button on the Iris toolbar to do so is simply unavailable.
I've played around with just jumping to doing the photometry, but without the preparation working beforehand I don't have anything that I'd want to ever submit - which kind of defeats the object of the exercise.
If someone who knows Iris well could take a look at the images I've provided (hopefully - let me know if you can't download the images and I'll share them explicitly) and tell me what is wrong with the images or what I am doing wrong in Iris would be awesome.
Thanks.
- Carl Knight (KCD)
Hi Carl,
I just had a look to some of your images using IRIS. The only thing I got is a missfunction of the "global alignement" due to the fact there are very few stars in your 30s images. That option needs a good set of star at a good ADU count to work properly. Then I used the "one star" option of registration (4.2) and the following addition (4.3) worked properly. Then no problem upto photometry. My system is W7 x64, I also used XP and older systems, IRIS is 5.59 (It seems there are some problems with W8)
Note: As you used a driven equatorial you can use the "one star" option, you don't need the derotation and other corrections. To do that you should first point the star to be used for registration by drawing a rectangle around it ( mouse left button and drag ) That rectangle defines the search area, by the way it shall be large enough to include the positions of the star in all images. But it should avoid any other bright star that could create confusion.
Then I had a look at the 300s image and I had no problem upto making photometry.
Threshold: This is only for displaying the image and adjust the min and max levels to be vizualized. This doesn't affect the data in memory or files. The button is the one with two arrows. I didn't see any problem with, being able to switching it off and on any time.
IRIS is normally very reliable but its style of interface is special. It has been developped by CB during many years, with lot of evolution by addition of functions. The original was a console type of application and the GUI has been added later. Many functions are not under that GUI. Due to this not all functions follow the same process. For example after the addition of a sequence the result is in memory but not saved in a file, you shall do that using the "Save..." from the "file" menu.
Next the processing is made using a specific file format "xxx.pic" all images shall be converted to that format to work properly. This shall be set at step 1 by fixing the "file type" to "pic". Then you will find all those files used for processing into the folder defined under "Working path". Be sure to push the "CFA" button after dropping your RAW images into the "Decode RAW files" box and after having set a file name, wait for the processing being completed (could take time..) and push "Done". The result should be a set of "xxx.pic" files in the working folder. Don't use the "Load a RAW file... ", it doesn't generate a CFA.
I attach another version of the tutorial, it has been revised, some error corrected and additions have been made to show various tools available in IRIS. They are useful to check images and set parameters like the photometric aperure size.
My last point is about your optical setup. It's very surprizing you got so few stars with a large aperture scope, 30s exposure and 1600 ISO. Normally you should get a lot of stars. We have often several hundred stars in one image when conditions are ok. In your 30s images one star is strongly saturated (not usable) two others are at a good level and about 8 at very low level.
My most usual setup is a telelens (200mm F4 that's a 50 mm aperture only, yours being 40 times that surface ! and photon flux too..), a 450D and a small equatorial. At ISO 400 and 60s exposure my limit magnitude is 14 (pushing ISO further doesn't bring any more electrons to the 14 bits sensor... just an ADU multiplication without bringing any more information... ) With your scope it should be at least mag 17 at 30s exposure, a lot of stars !
Then I don't understand what you mean in " ...the camera on an off axis guider to do the imaging..". An "off axis port" used for guiding usually capture only a small section of the beam and its F# is very high (corresponding to a very small aperture). Then the FOV should be very small too ? Is this the reason for which you got so few stars ? If yes, this is not a good solution for reaching faint stars for photometry, the photon flux would be very low through that off-axis port. You could make much better with a telephoto lens and attaching the DSLR "piggy back" onto the scope enabling you to use the drive and long exposure. Then why not attach the camera to the normal port instead the off-axis one ?
Clear Skies !
Roger (PROC)
Hi Roger,
Thanks for your time and effort. It's much appreciated. I'll try processing the images per your example.
To clarify the optical set up...
The camera is at prime focus (straight through the off axis guider - normal port I think you called it). I have an illuminated reticle in the off axis guider port. To ensure understanding have a look here: http://ngileahtelescope.blogspot.co.nz/2013/04/the-sun-as-variable-star.html - the bottom image is pretty much my set up with two exceptions:
Regarding the lack of stars you identify, I have also been somewhat surpised as the DSLR is not reaching much further than I can with the naked eye and my favourite eyepiece. I wonder if the camera's noise reduction is getting in the way? I'll turn it off and see what the result is.
Regarding a telephoto lens piggy backed on the scope... that's precluded by item 10b i.e. the lack of any spare money at present for doing anything but use what I have unfortunately.
Thanks again regardless.
- Carl.
Hi Carl,
Normally the DSLR internal processing should not apply to the RAW image file, but ok, I have no experience with Pentax and it's possible they do apply it (Nikon do some... ). By the way it's better to switch off all noise and other processing internal to the DSLR. Maybe the small number of star comes from the narrow FOV of the configuration, the focal length of your 12" LX90 should be about 3 m that's about 0.4 x 0.26 deg only for an APS-C sensor. Then this location in the sky could be poor in star. To make a test you could target a star cluster, M67 should be well documented, but it's low at your latitude.
Clear Skies !
Roger (PROC)
Hi Roger,
Funnily enough the images I have taken of XMMU J115113.3-623730 are full of stars. The field around 1RXS J105010.3-140431 is very sparse anyway.
When the weather and phase of the Moon next allow these are my next steps:
1. Make sure the camera is not doing any noise reduction.
2. Take darks and bias frames with the camera at the same temperature images were made at.
3. Make master twilight flats (now going to have to wait until the moon is gone from the evening/morning sky) taken opposite the sun position.
4. Separate the flats/bias/darks by channel. This also implies that I should separate the RGB channels of my images and process each separately.
5. Determine DSLR saturation behaviour - next step in the Citizen Sky tutorial.
6. ...if I get something where my comparison stars V mags are close enough to the catalogue magnitude to be confident in my transformed variable V mag I'll submit everything for review and if people are happy, I'll make my first DSLR photometry submission to the AAVSO.
Clear skies.
- Carl.
Hi Carl,
I don't understand why you want to make the preprocessing at level of separated R,G,B images ? The functions under "digital image" work well under the CFA format of image (step 4.5 of my tutorial version) But as you would probably use only the green channel (classical transformation to Johnson V) it's true you could separate the images before preprocessing and continue only with the green... Anyhow the DSLR R is not usable and there are no published ways (if I know) to use the DSLR B to generate Johnson B (I do it with my own technique, but there are problems...)
With DSLR, master_bias are not needed as the bias (if bias defects exist ! ) is included in the dark. Adding a bias from the DSLR will just add extra random noise, better to replace it by a dummy file (there is no option to skeep it under IRIS. Point 4.1.2).
With recent CMOS DSLR dark are not needed upto 30 sec of exposure, but I would recommand to apply the hot pixel process, even be aggressive ! (point 4.4 in my version) But I don't know if this Pentax is a CMOS or a CCD, I saw hot pixels in your images. From your pictures it's a 12 bits ADC with a saturation level at 4095 (probably some less at ISO 100 ~3500), there is no coding shift of the ADC zero (like Nikon) that means the black level is 0 ADU. Canon uses a shift to avoid clipping in deep dark image. A CMOS is normally perfectly linear up to saturation. CCD could show a problem above 50% of the range due to the anti-blooming.
The flat is critical to correct the vignetting of the optics (and maybe dust on the lowpass filter as your F# is high, but better to avoid dust ! The flat process being not very effective for this). It's difficult to generate a precise enough flat with wide angle lens, it's probably easier with the long focal of a telescope but frankly I doubt the twilight method would be ok. It would be better to use a diffuser at the top of the telescope (the uniformity of it is not much critical as at this point as you are near the pupil of the optics). Then illuminate the diffuseur with a lamp well on the axis. I use a non-glossy, fine grain photopaper as a strong diffuseur and an halogen lamp. The light attenuation is strong but it's ok. You can use a short exposure (but not too short: ~1 sec...) for the flat and then have no need for a flat-dark. If the optics configuration is stable the flat can be reused for many sessions (case when lenses are used, less obvious for a telescope...)
Clear Skies !
Roger (PROC)
Hi Roger,
I have been discussing with Mark Blackford how he processes his images (using AIP4WIN) I was considering separating the images into their respective channels sooner in the processing. Certainly not doing so makes for simpler processing. I have talked to Mark about whether the bayer filter B & R can be similarly transformed to Johnson B & R and yes, it sounds a good deal more complicated.
The K10D has a Sony ICX493AQA CCD sensor.
My experimentation with twilight flats, just eyeballing the flats, whilst the colour is fairly uniform, there are stars (donuts out of focus). That in itself makes me think that you'd need to take a good number of images and make sure the scope is not in the same star field or tracking otherwise I can only imagine that will make for rubbish flats. In fact, what I did was slew the scope a small distance when I took the flats. Then there's the fact there is colour in the flats too.
I'll try your suggestion for getting better flats.
I just need those clear skies you mentioned to get all my ducks in a row and get everything done and achieve some actual photometry.
Cheers again and thanks for the advice.
- Carl.
Hi Carl,
The IRIS "Grey_Flat" function under 4.3 normalizes the flat between the color plans. Then R,G,G,B are flat corrected without affecting their ratios.
The color index B-V can be extraced direcly from the B/G ratio (ADU, photon count level) It's easier than trying to extract Bj direcly. It works reasonnably well for most star spectral types except M type at luminosity I to III and possibly a few late K (this is due to their blue "excess" between 390 and 440 nm that is not well seen through the band-pass of the B channel of the DSLR)
I = -2.5 log(B/G) then classical transformation:
B-V = zp + k I
The DSLR R is too far from the Cousins Rc to work properly (missing a lot of deep red and NIR of Rc, critical as there are a lot of molecular bands at those wavelength).
You are right, our most serious problem is we don't see the sky ! Here (France, western Europe) it's an astronomy disaster for a year now. Clouds, clouds and clouds ! Now cold, a lot of rain... I am afraid this is a climate cycle that could be linked to a 10~20 years cycle of the ocean.
Clear Skies ! I hope...
Roger
Hi Roger,
Rest assured I will post back here any results when I get cooperative weather. It is likely to be in a couple of weeks as the Moon is going to be too bright for a bit.
I took a chance on DT Oct last night, opened up the observatory, got set up. Cloud arrived, finished having to close up hurriedly to avoid rain. Didn't clear again until midnight by which time my bed was too warm and my brain was reminding me I had work in the morning.
We had drought through the summer. Unfortunately (or not) I spent it building the observatory. We're still well behind on our annual rainfall but we're getting plenty of cloud. It's almost enough to make one take up radio astronomy!
Cheers.
- Carl.