subframe selector and DSLR

[Torsinadoc]

Im attempting to use subframe selector in addition to blink to select best frames.  I searched PI and looked at documentation but I am at a loss on how to set up the parameters.  How can I determine just basic set up information  DSLR (canon T3i)?

Thanks
Alan

[mschuster]

Hi Alan,

Please try the defaults first. The numbers will be displayed in units of pixels and DN data numbers.

If you want to see star size FWHM in units of arcseconds, in System Parameters set Subframe scale to your image scale and choose Arcseconds as the Scale unit. Similarily, you can display image statistics in e- electron units rather then DN by setting Camera gain and Data unit.

Once measurements are made, you can see a map of stars detected by clicking Output Maps and opening the resulting s_m.fit files. If the set of stars detected does not seem quite right, try modifying Log(detection sensitivity) and Upper limit in Star Detection and Fitting. You can also try a different Point spread function. The other parameters likely don't require modification.

Thanks,
Mike

[Torsinadoc]

Thanks.  I converted from raw to .fit
loaded subs into subframe selector and left it default.

It gives a list of numbers and possible graphs. How do I utilize those to determine best frames?

Thanks for the help

[Torsinadoc]

Indiv frame and map

[mschuster]

Alan, I look at FWHM, Eccentricity, and SNRWeight mostly. FWHM provides info on seeing and focus. Eccentricity tracking and optical quality. SNRWeight camera noise, exposure, and observing conditions. Check out the documentation. I have not used a DLSR, so can't say much about results. Comparing frames of the same target and filter, I toss outliers, and also try to figure out want went wrong and if possible fix it so it does not reoccur. In addition to checking star detector parameter setup, the maps provide info across the frame, helpful for checking optical quality, for example.

Mike

[MikeOates]

I am not sure if what I am saying is correct (can someone who does know confirm?) but the example image you show has not been debayered, I think the OSC (i.e. DSLR) images should be debayered before using the subframe selector script.

Looking at the documentation it says:

In the Target Subframes section, add all target subframes. Subframes may be raw, calibrated or registered, but all should be the same type, have their overscan regions (if any) applied and cropped, be compatible for registration (if unregistered) and be compatible for integration

see: http://pixinsight.com/doc/scripts/SubframeSelector/SubframeSelector.html#description_001

You can't register or integrate un-debayered frames.

Mike

[mschuster]

Yes, thanks, OSC raws must be debayered first, then make separate script runs for each filter, sorry I did not say so. You may also calibrate the frames prior to running the script, I do so, but don't register them.

Mike

[sctall]

Mike-

I have tried to get the multi image when I create the star maps using "Output Maps", but all I ever get is the image of the fitted stars.
Not the FHWM or Eccentricity maps.
Is there something else I need to install to get these?

Scott

[mschuster]

Scott, The script should be writing three images into the s_m.fit file, as a multiple-image fit file. When opening the file in PI, the three images should open automatically in separate windows. Not sure what is going wrong. If you can, please post a dropbox link to one of the files, I will double check it. It may be your copy of PI is configured not to read or write multiple-image fits. Please double check that Format Explorer > FITS > capabilities > multiple-images is true. It may be that this is a script bug on your system. Not sure. Sorry for the trouble.

Mike

[sctall]

Mike
Turns out the issue is that the process console says:

** Warning: insufficient number of fitted stars in at least one subregion of the subframe to output FWHM and Eccentricity maps.
I was able to get one image to output all 3 , but the rest would not.  I did put 2 files up on dropbox, where L_0004 worked and L_0005 did not.
These are with converted to .fit and debayered.  I get a little better SNR weight with debayered.
When I use them as .CR2, I don't even get file L_0004 to work.
Here is the link.  I left the files as 212MB as I use them when they are .fits.
 
https://www.dropbox.com/sh/kawfzgw9c8uoz5e/AADJGxl3Lx1ozzY8BBinTMP9a

Could you help me know why all my other sub (subs) do not have enough fitted stars.

I have a lot of LP I deal with here, and I can only image for around 5 min before I saturate.
I certainly do not get as many stars as I would like, but

I do have 400+ Star Support stars in the images though.

I am just learning how to use this script.

Thanks
Scott

[mschuster]

Scott,

When the script runs on an debayered RGB image, it first converts the image to monochrome, and then proceeds with the measurements. Another way to run the script is to first run Image > Extract > Split RGB Channels to separate the data, and run the script three times once each with all frames of a particular channel. This gives information about each filter, which might be helpful. Running the script either way is OK.

Also, please apply an STF to the star map image. This will make information more visible. Doing so will show detected stars (marked with a cross) as an overlay on top of the original image.

The stars near the edges are a bit defocused and elongated. This fact plus the LP accounts for the low star detection, basically SNR is fairly low. The script's star detector is not well suited for very low-SNR cases.  And the script needs to detect a reasonable number stars distributed across the frame to make maps that are useful.

Please try adjusting the Star Detection and Fitting parameters to see if you can get things to work. First decrease Log(detection sensitivity), try -1.5 or maybe -2. You might also try increasing Star peak response to say 0.9 and/or decreasing Maximum star distortion to say 0.3. Unfortunately, in the worst case of low-SNR, none of these tweaks are going to solve the problem.

If you can, please place the original debayered RGB fits for 0004 and 0005 in dropbox and I will try tweaking the parameters myself to see if I can get something to work better.

Mike

[sctall]

Thanks

OK

I don't have the best of conditions, but I still get pretty decent images. They would be better with a cooled CCD camera.
I am saving for a Flattener, which I hope will help the edges.

I like the script because I can really sort out the outliers. I need to keep as many as possible with my conditions.

I have put the 2 orig debayered images in the same dropbox.

I appreciate your help

I just thought of one thing

If I debayer the light subs first, and I use the BPP script for integration and all, can I still leave the Dark, Bias, and Flat subs as RAW and still load all into BPP?. Or do I need to convert and debayer those as well?

Scott

[mschuster]

Thanks Scott.

Try setting the parameter Star peak response to 0.9. This will nearly double the number of stars detected, and the maps should generate OK. The other parameters don't seem to have much impact.

In addition, for measurement purposes, you may wish to crop off the edges where star detection density is lower due to lower SNR. The script will then give more representative statistics on the central image area, and further improve the likelihood that maps get generated. Try setting the parameter Subframe region to 500, 500, 4208, 2476 (left, top, width, and height). This will crop off 500 pixels on each side.

I don't know the answer to your question about leaving as RAW versus debayering darks, biases, and flats for BPP. Maybe someone else knows?

Mike

[sctall]

Excellent

Thank you again

Scott