Dumb post about a dumb process -- PhotometricColorCalibration in PI

So...   I was going through my old files and I thought I'd get reacquainted with PhotometricColorCalibration in PixInsight.   Well, well, well.  Error after error after error, I did the obligatory search online and looked at forums and I didn't really see a consistent solution to the problems which seem related to platesolving.  I now remember why I don't use it. :)

But after almost giving up, I finally got it to work... sort of.   What I figured out was that there were SIX (!!!) different settings that you could "wiggle" that would sometimes make it work... eventually.  To me, there is no logic to it.   But I thought I'd write a few notes here so that when I actually work on a data set, I don't pull my remaining hairs out of my proverbial head.  

#1:  The first time I got a successful solve was by changing the White Reference setting to O9V star.   I was trying to solve the Flaming Star Nebula and I looked up AE Auriga (the brightest star in the frame) on wiki and found it was a type O9.   After the successful solve, I tried other "white references" and found that G2V and Average Spiral Galaxy worked as well.  So...  that actually didn't give me much confidence.

#1A:   I tried a bunch of other images and I found that one time I could get a solve by changing the Declination by a few minutes.   Weird.   Again, no logic to it.

#2:  After a bunch of failed solves, I noticed in the process console that the scale was oftentimes wrong.   It would show something like 17.x arc/pixel.   Which is WAY OFF by a factor of 10-ish.  So, I entered .43 for the pixel size when in fact it was 4.3.   That would do the trick sometimes.   BUT get this....  one time, it didn't solve and I changed .43 to .42 and it solved.   WTF!?!?!

#3:  The thing that made things work the most consistently however was changing the Astrometry catalog from either UCAC3 to Gaia or vice-versa.   When you change the catalog, a new download occurs.  For some reason, this can make everything suddenly work without changing another setting.    Again:  WTF!?!?

#4: A couple people online have said that unchecking the 'Automatic limit magnitude' and then going to 13 can help.   Well, I don't think that ever was a solution for me, but it might be something to try if everything else doesn't work.  As a default I just use the 13 mag setting.

[Update:  I had a situation where I did the opposite (ie changing the magnitude BACK to 'Automatic limit' and it worked.  So, toggle, toggle...]

#5: I've tried using Noise reduction at a '1' or '2' setting.  I don't know if it does anything really.   [Later: I was having problems with a solve with the Noise reduction set to 0, changed it 3 and then got the solve.  So, it can work, but I was working on individual (not stacked nor calibrated) frames which tend to have some noise.]

#6: I also keep the automatic limit magnitude unchecked and enter 13 for the limit.

# Bonus:  Lastly, I make a small preview window in the same way you might for Background Neutralization and click the "From Preview" button and select the preview for the ROI.   I'm assuming this just saves a step.

Why all this effort?   Well, I normally just use Colorcalibration and it works great.   But sometimes, I feel like my colors are off and by using PCC (ugh), I'll be starting with a clean slate.

Another example where I didn't get a solve with spiral galaxies as the "white reference".   But I changed 3 things (I know, I know): 1) Set the white reference to G2V; and 2) since arc seconds/pix was at 17.7x, I switched the pixel size to .43; and 3) I switched the Astrometry catalog to Gaia.   The image was a little green so I did a pass of SCNR for good measure.

Hm...

So, I tried solving the Flame again and no dice.  :/  I played with settings and this time the thing that worked was setting the pixel size to the erroneous .37 value.  Look at those stars match the expected line on the graph.