Transition pain from DSLR to CCD

P

primefocuser

Member
I have used Pixinsight for quite some time to process astro images taken with my Canon T3i and decided it was time to take the next logical step and buy an SBIG STF-8300C. I took some photos with this camera similarly to those I have taken with the Canon but I have run into problems in PI. To get started, I use the BatchPreprocessing script. However it fails with the error message <light file>_c_d_r.fit: Zero or insignificant signal detected. Empty Image?

When I examined the calibrated files, they looked totally bizarre so naturally the subsequent deBayered and registered images looked bad too. The question is why are the calibrated files messed up? When I look at a raw sub in PI and use STF AutoStretch to examine it, the image looks fine, not very different from those I have taken with the Canon. I don't know if the deBayering has anything to do with the problem but I doubt it since deBayering is part of the script. Other than that I would think everything else would be the same. The only other difference is that the CCD camera produces .fit files and the Canon produces CR2 files.

I have dropped some example files taken with the CCD camera here. There are no dark frames because I used the CCD camera's auto dark subtraction feature to save time and storage. http://tinyurl.com/hl9m3gr

I am using PI Core Version 1.8, Ripley on Win 10. I would appreciate any feedback on how to solve this problem.

Thanks,
John
 
if the camera did auto darks... maybe you've double subtracted the darks/bias? with that type of light frame the only calibration should be to use flats.

rob
 
Hi Rob,

The SBIG documentation recommended doing the auto dark subtraction specifically on the lights. It makes no mention of darks being subtracted from the bias or flat frames.

I followed the BatchPreprocessing script's steps manually. It quickly became evident that there is something wrong with my bias frames. By skipping those I was able to create a reasonable image. The script insisted on bias frames but I avoided that by calibrating manually. It remains to be seen what the problem is with the bias frames.

John
 
what i'm trying to say is that if the auto dark has been applied to the lights, then if you subsequently subtract a bias or dark from the light, you've double-subtracted the dark/bias. with an auto-dark subtracted light, the only thing left to do is to calibrate it with a flat.

what capture program are you using? CCDOPS?

IMO i'd just turn off the auto-dark and put everything into BPP. it should work out fine that way. i use an STT-8300M and i've never used the built-in dark subtraction (in fact i didn't know it existed - it's not a common way to calibrate images.) maybe there's a good reason to use it with the OSC version of the camera... i don't know.

rob

 
The auto dark subtraction is a feature of TheSkyX Professional which is what I am using. It even has a way to build and use a library of bias and flat frames to do all the noise reduction at one go. I haven't looked into that at all though.  Since I didn't take any additional darks there were no more to subtract but there is definitely something hinky about the bias frames. They seem to look reasonable enough when I look at them in PI and stretch them but that seems to be misleading. I'll contact the SBIG folks to see what they think is going on. I'm brand new to the cooled CCD world and I really liked the idea that I could take a single dark and apply it to each light frame but falling back to SOP won't be the worst thing that could happen.

BTW, what is BPP?

John
 
BPP = short for batchpreprocessing.

well the thing is, a dark frame contains the bias signal as well as the dark signal. so if you subtract a dark (in camera or with some other software) and then subtract a bias using ImageCalibration or BPP, you subtracted the bias twice.

however, that might not be the whole reason that the images were destroyed by the 2nd bias subtraction - in general mixing calibrated frames (your TSX-dark subtracted lights) with uncalibrated frames (the bias frames) in another application (PI) usually does not work. it boils down to what format TSX wrote the images out in. if they were not i16, then there's a good chance that things went wrong as whatever floating point format TSX used to write the lights is not necessarily the same as how PI employs floating point numbers. for proper results you should do all the reduction in a single program. so if you can convince TSX to also flatten your images, then just bring in the fully calibrated images into PI. but if not, you should just use raw lights, darks, bias and flats (meaning i16, just as they came off the camera) and put all that into BPP and let PI do it's thing.

anyway, most likely there's nothing wrong with the bias or with the camera... i have only used TSX in passing but PI's calibration routines are pretty top notch, so most folks here would probably say to just let PI handle calibration.

rob
 
John

If you used TSX to calibrate it "should" save two files, a calibrated .reduced.fit and uncalibrated version .fit (see bottom of page 460 in the current TSX user manual) though I am not too sure what happens with an Autodark subtraction (its really only meant for auto guider calibration). The TSX calibration is quite handy for doing a rough and ready stack as you progress through an imaging run but as Rob says, mixing calibration in TSX and PI is not a good idea (PI has a far more robust calibration workflow).

Chris
 
Thanks, Guys. That all makes sense. Going back to pure files makes the most sense -- at least until I'm familiar enough with the camera to want to explore its more sophisticated options. The TSX documentation recommended using the autodarks but all it did was open a can of worms.

p.s., As soon as I hit Send I realized what BPP stands for. Funny how often hitting Send jogs the memory  :p
 
You must log in or register to reply here.
Back
Top