DNA Linear Fit on multiple panels

[Ryderscope]

Hi, I have a couple of questions regarding DNA Linear Fit. I am working on a 6 panel mosaic and it is not behaving in the way that I was expecting.

My first question is: Can DNA Linear Fit be used on multiple (6 in this case) panels?
If yes to the above question, what is the recommended sequence to use. For example, to use the script it is necessary to select the reference frame and the target frame. For a six panel mosaic, should we select one of the six frames and use this as the reference for all of the others? This would seem logical but it is not working for me. The symptoms that I am getting are that after the script is applied to the third frame it clips all pixels to zero. For example, If frame 6 is the reference, DNA Linear Fit frame 5 to frame 6 (ref) - all ok. DNA Linear Fit frame 4 to frame 6 (ref) - all pixels in frame 4 clipped to zero.


Most odd and I must be doing something silly so any help would be appreciated.

Thanks in advance.
Rodney.

Edit: I should add that the panels are all linear, RGB and have had DBE and PCC applied.

[pfile]

the frames must have some overlap for DNALinearFit to do it's thing. so you need to do it piecewise - say for instance you have 4 frames and they all overlap. in theory you can then select the top left as the reference and run DNALF on the other 3 in turn. or if you had 9 frames and all the outer ring panes overlapped the center, then you can use the center as a reference and run DNALF on the 8 surrounding.

if you have frames with no overlap, i think it's safe to switch references midway to a pane that's been fixed with DNALF. so you just have to choose a sequence of runs that covers all the panes and make sure of course that the reference is either the original, or a pane that's had DNALF run on it.

rob

[stevek]

Interesting.  When I used MergeMosaic I could still see where the joins we despite using the top left panel on my mosaic of 4 with a 20% overlap as the DNALF reference
I need to look into this more.

[pfile]

well sometimes the seams can be obvious due to SNR differences between the panes. and there's always the chance of bright artifacts from stars near the seams...

rob

[stevek]

Yes that's true.  Closer inspection reveals that my integration times were different on the panels.  I will revisit this when I get more data.
steve

[Ryderscope]

"the frames must have some overlap for DNALinearFit to do it's thing."

....

rob

That was the key, Rob. As long as I make sure that there is an overlap between the reference and the target I avoid the problem that I was seeing. I still have some colour gradients but I have a good base to work from now. Thanks again.

Rodney.

[pfile]

good to hear, thanks for confirming it worked.

rob