BatchPreprocessing issue.

[whitenerj]

Ok, I just started using PI a couple of days ago and cannot get a decent stacked final image out of the BatchPreprocessing script. I'm using the defaults for everything for now and the images were taken with a STF-8300c OSC camera. I have CFA set with my bayer pattern of BGGR as well. Tried with and without optimize dark frames. The set of lights I'm using are 32 300" frames. When stacking in DSS, the stack looks great. But the one from PI looks pale and almost like a single sub. Also when stacking in PI the image is all green until I decouple the channels in STF. Below is my stack from PI and DSS.

PI:

https://drive.google.com/file/d/0B1ae8BkkZsmrQTM0QWRLb3lxb0k/view?usp=sharing

DSS:

https://drive.google.com/file/d/0B1ae8BkkZsmrelpzdThRcEI3Snc/view?usp=sharing

Thanks for the help!!

Jason

[jkmorse]

Jason,

What capture software are you using?  Also, try calibrating your lights without using a bias frame just to see what happens.

Let me know if that helps. 

Best,

Jim

[whitenerj]

I will try the bias frames next time.

I'm using Nebulosity for capture.

Jason

[oldwexi]

Hi Jason!

Your PI Fits file is still linear and in perfect shape and much sharper than the DSS version.
You have to unlock in STF the channels, this
gets you the image in the right color and similar stretched on screen.

Anyhow, both images look like as you did not use flats, because the corners are to dark...
And,  9 Images to stack is not enough for deeper Images... Check the FITS Header in PI it says you only used 9 Images for Integration-

Here your PI Image simply stretchted with STF>HT in PI.


Gerald

[whitenerj]

DSS stretches the image? I never knew that. I was sure it only stacks the image

The images both consist of 32 stacked lights. Each at 5 minutes. Not sure where you got the 9 from.

You are correct that no flats were taken. But DBE takes care of that fairly well.

Jason

[oldwexi]

Hi Jason!
DSS did not stretch but did some "Color Align" means a Change to the channels.
DSS kills data because you saved with TIFF 16 Bit. Check the filesize of the FITS file
and the TIF File the FITS file is 32-Bit and also double the size which includes more dynamic.

Yes, DBE can remove the dark corners but in your specific case, because of so much nebulosity around,
it also removes in the upper right Corner some HA.
Deep Image processing without Flats is useless in my experience, because you cant get deep really.

Check your FITS File in PI with "FILE  - Fits Header" - it says only 9 Images were used for Integration.

Gerald

[whitenerj]

Interesting, I guess it threw some of the lights out.

I do wanna stick with the 32 bit. How can I get rid of the green cast in my linear file and make it look like it does with the color channels unlinked?

Jason

[oldwexi]

There are more ways to keep the image linear and get the channels right.
First :
Extract R G B
use LinearFit    the G Channel as reference      on R and B
than Channel Combine.

Second possibilitiy :
open Histogramm Transfer, it should be open during all processing anyhow (!!!!)
for better control whats happening,
move the left slider of the green channel to the right and bring the G bump to cover the R bump
move the left slider of the blue channel to the right and bring the B bump to cover the R bump.
Pull the triangle of the HT process over the Image and you can afterwards use STF
with linked channels.

Or Third:
Create a Preview somewhere in the middle of the Image,
Use the Process" Background Neutralisation" using the preview as reference
gets you also the wished result.ö

There are also other possibilities how to get the channels adapted,
but using one of the first three would be o.k for your Intention.

Gerald

[whitenerj]

Ok, I got the channels coupled back with the colors right but once I preform DBE on the 32 bit PI stacked image, it looks no where near as good as the DSS stacked image. BTW I re-stacked in PI and had an integration of 31 lights on the new stack. The stack on the left is from DSS and the one of the right is from PI. I tried a saturation curve to bring in those colors but wouldn't help. I want the PI image to look as good or better than the DSS image in linear state.

Jason

[oldwexi]

Jason!
Let me have the 31 Image PI stack. So i can try to push the PI stack, 
i have only the 9-image stack of PI compared to the 32 image stack of DSS.

If you compare both histograms (zoom in deep),  the DSS version has a much wider red channel.
And the stars have a more red halo.
The question is: is this correct? Because if i unlock the STF channels und do autostretch for both images
from the color point of view both images look the same except the image of PI is sharper.

Your comparison of the last post therefore compares a slightly stretched DSS Red channel with
a smaller red channel of the background neutralized PI image.
Stretching the red channel in the PI Image gets you the same or more Color in PI.

Zoom into the histogram and compare - DSS did some preprocessing of the stack.

Gerald
P.S.:
here also a  comparison from the PI Point of view. Left the PI right the DSS image

 

[whitenerj]

Ok, here is the stack of 31 lights.

https://drive.google.com/file/d/0B1ae8BkkZsmrYlM1aHhfREdmdDQ/view?usp=sharing

Take a look at it and if you can show me how you achieve your results, I would be grateful.

Thanks

Jason

[oldwexi]

Hi Jason!
Find atttached a version of your image processed.
Cant go much deeper without "selective painting steps..." because of missing flat.
Whenever i stretch the image in a subsequent step the bright center appears again,
which should haved been reduced at the beginning during calibration with flats.
So, the result is not really true color, tried to use the data for stretching and did no
manual selective processing.


Have saved all steps and data as project (3.2GB).

Gerald

[whitenerj]

Thanks, I'm not seeing the project files? Did you have a link to those?

Jason

[oldwexi]

Hi Jason!
As mentioned the Project File is 3.21 GB large.
Do you have some space whre i can upload to?

Gerald