Ison through sickening LP

[timtrice]

I'm at the end of my wits with this thing. I shot Ison Sat morning through street lights and the LP dome of Houston which has also resulted in significant color - well, crap. And while I think there is a good image there, the LP itself is just destroying me. I've seen hints of bowing during wavelet transformations. But I don't progress too far before trying to get a grip on the LP. And I just can't. If I extract too much I have terrible noise and still have leftover gradients in the corners.

https://www.dropbox.com/s/x33ba7gxejr7hbo/ISON_2013-11-16-star_alignment.fit

https://www.dropbox.com/s/al0x5zap1244wzl/ISON_2013-11-16-comet_alignment.fit

[pfile]

i tried processing it a little bit, first cropping the comet only picture so that the nucleus was in the lower left hand corner, then running DBE.

i had to tweak some of the DBE parameters to avoid things getting too lumpy in the background.

the main issue seems to be a circular gradient that looks like either no flats or bad flats. it cuts right thru the comet's tail…

let me post the crop and DBE icons and maybe it will be a start for you

rob

[timtrice]

That's not what I was expecting to hear. I did 50 using AV Flats in BYEOS on a lightbox. I really tried to spend a lot of time making sure I got all the calibration files right.

My run w/ the DBE really made it very noisy and that still wasn't enough as I had additional gradients in the bottom corners. I think the major issue was having a street light off to the left when i started on the comet around 4:30. Of course as it moved up the gradient shifted. That combined with the light of a near full moon I knew it wasn't going to be fun but had no idea it'd be this freaking difficult.

[pfile]

well i don't know for a fact that the flats are bad but there is a circular gradient going on… it's not terrible however and may just be the result of trying to stretch too aggressively.

anyway here are the process icons. i can see that the background is still kind of colorful and lumpy due to this dbe. one way to get rid of the color is to make a mask which strongly protects the comet and then do atrous to the background, setting the layers to maybe 6 and then removing the last layer in "Chrominance (restore Y)" mode.

rob

[pfile]

not my best work…

[timtrice]

Rob, thanks for the zip. I'll take a look at it and see what I'm doing differently.

I did take dozens of lights at different ISO's - 400, 800, 1000, 1250 (all 10 secs plus an additional batch of 20s 400ISO - 92 lights total). I know the higher ISO's are contributing but I was aiming to try and get the streamers and bowing which judging by your pic at least some of it is there.

So, that being said - is it possible I got similar detail on the lower ISO's? I got 150 bias, 50 darks/flats for each ISO/TV. But I'm thinking I may need to just cut out the 1000 and 1250 ISO's to at least get a better BG. Sound about right?

[timtrice]

The zip is an xpsm file PI won't let me open. What am I missing?

[pfile]

oh right, well you need to do the following - Process > Process Icons > Merge Process Icons … and navigate to and open the file. the icons will appear at the bottom of workspace 4.

this is one area where PI can be improved; i think that if you did Load Process Icons i think it would overwrite all of your process icons! also you do the merge or load and if you are on the wrong workspace, you're left wondering where the icons are…

the streams are definitely there and i think it's a pretty good image. you might experiment with different rejection parameters for the starless image since the stars are faintly visible still. one technique for handling this is to stack all the odd images together and all the evens together so that the separation between stars is greater and it's easier to reject them. then stack the stacks. ison is moving pretty fast so this might not even be necessary.

as for the different ISOs… i don't know. PI is going to weight each image by the SNR of the sub so in theory each one is contributing exactly the right amount to the final image. having said that the "signal" here is mostly sky gradient so i'm not sure if the right thing is happening.

there is a pretty good post by juan about comet processing in PI here:

http://pixinsight.com/forum/index.php?topic=2008.msg13040#msg13040


rob

[timtrice]

Thanks again, Rob. I'll take a look at it tonight. I saved my process icons as I was running through pre processing and now PI won't load them; it freezes and forces a restart on me.

I'm going to implement your suggestion and even break it down further just by ISO/TV and see what I can get. Maybe if I can isolate the light on a smaller signal-scale, so to speak, I can pull it out. If not well then I just process what I got...

...and never try to shot an object through Houston LP ever again...

[jkmorse]

Tim,

I am off in the deserts of Oman and the UAE (light dome in Dubai stinks here too).  But I used to shoot in Houston all the time.  Are you aware that the Houston Astronomical Society has a really nice dark sky site just south of Columbus out on I10.  For only being 75 miles away it gets you out of the glow in a big way.  You need to join, pay a small fee and sit through an orientation session, but it is well worth the effort.

Best,

Jim

[timtrice]

Hi Jim. Yes, I am a member and have been out there before. Problem is the gear I'm using now is too big to fit comfortably in my Camry. I am considering taking a smaller scope out there this weekend but weather isn't looking good. It's about an hour drive but between weather and work things just haven't worked out for me. Saturday worked somewhat well despite some cloudiness but I also had to make a road trip shortly after my session was done.