Lights Integration Failure no M106

[niteman1946]

Hi group.

This morning I followed the same routine that has been successful for three previous attempts. 
1.  Bring in all of the lights from previous sessions and calibrate them
2.  Use previously created Bias, Flat(uncalibrated) and Dark(uncalibrated)
3.  Check the "calibrate" box for each of the Master Flat and Dark
4.  Debayer all files once successfully calibrated
5.  Star Align after successfully Debayering
6.  Execute the Image Integration after aligning.
         a)  Combination -- Average
         b)  Normalization -- Additive +Scaling
         c)  Weights -- Noise Evaluation
         d)  Rejection Algorithm -- Linear Fit Clipping
         e)  Normalization -- Scale +Zero Offset
         f)  All other settings -- default
 I had no problem with sessions of 19 lights, then later 31 (19+12), and then 47 (19+12+16) lights.  I have now added 17 more lights to the process total of (64).  All goes well through the steps until I get to the light integration.  The process fails due to "...unable to compute noise estimate".  And this occurs early on, at about the 7th sub, which had no problem the previous three tries. Removal of the offending sub does no good.

Appreciate any help on this.

Mark
 

[Cleon_Wells]

Mark, did your Integration Reference image/file change after you added the last set of subs?
A note on the Reference image, while still trying to figure out why my MonoChrome Calibration files work the way they do, I noticed that the Reference image I chose for Registration, which was open during Integration is used for the Reference image for Integration. Learned something new, with PI I think you never stop learning, of course with my “CRS”, this learning process will last a long time.. 8>)
Cleon

[niteman1946]

Cleon.
Yep, the reference image probably did change from earlier successful registrations and integrations to the later problem one.  I did not make a note of which one I chose, so the likelihood is great.

You've got me a little confused about a reference image for integration.  I don't see that in the integration window.

I am also convinced that my computer is right on the ragged edge of managing PI.  I was able to successfully integrate (after my last post) by rebooting (i.e. clearing all memory) after each operation.

So, I'll be adding memory and disk space, or buying new.

Thanks for your suggestion. We'll get there.

Mark

 

[Cleon_Wells]

Mark, take a look at the right side of my screen shoot, note the second button in the Image Integration window, (Set  Reference), if you don’t select a Reference sub, the first image is used or if you have a Reference Image that has been selected for Star Registration and it is open, this open image will be the Reference image used for Image Integration. (on my Mac).
On the possible computer problem, you could give your wife your old computer and use this good will to justify buying a new Cray. 8>)  I did something like this when I gave my wife my 2 year old 17” MacBookPro, replacing her old Dell Inspiron  that had every kind of -Ware program, Mal , Ad, Spy, Under.  I bought a new 17”  i7 MacBookPro  for PI. What a considerate thing to do; am I a good husband or what? It seems I still have to remind her now and then of this, I think see could have “CRS”.
On the other hand, for the computer problem , if it’s a Windows Computer and has a motherboard that can hold the extra memory,  you could load it up with ram and make a ram disk , check out this web site.

http://www.tomshardware.com/reviews/ram-memory-upgrade,2778-4.html

Another thing to do for the computer issue is check the  (Region of Interest) box located at the bottom of the Integration window and set the top left corner point and then the width and height of this region , this reduces the size of the image the program  uses for its calculations.

On a completely different point  see the image, is a Hot Pixel at 2522x1573 on my DSLR, note that  when I convert this sub using the MonoChrome setting, I have one hot pixel. When I covert to RGB and then Debayer this same image I get one Hot Pixel at 2522x1573 that equals a level of 0.24 and four not so hot pixels at a 0.016 level surrounding the Hot Pixel . The camera is a Canon T1i that creates a 14bit file so clipped level is 0.25 in theory.
Cleon