Combining images with different exposure times

S

sreilly

Well-known member
I started an imaging project last week and had set my luminance images for 15 minute subs. I later changed this to 20 minute subs and would like to be able to use all the data collected. What I initially did was combine all the 15 minute subs and process. Then repeat with the 20 minute data. I know what I would do outside of PI but feel there has to be a good way to combine these in PI to get a good image. Any suggestions as how to procede? I have 12 - 20 minute subs and 13 - 15 minute subs.

Thanks,

Steve
 
Hi Steve,

Just combine all of them in a single operation. Use ImageIntegration and be sure that the Weights parameter is set to Noise evaluation (this is the default setting). Our noise evaluation image weighting algorithm will automatically multiply each of your images by the optimal scaling factor to yield the highest SNR improvement in the final result.

If you have 25 images in total, I'd suggest the new linear fit clipping rejection algorithm. See this post for detailed information.
 
Hi Steve,

I believe the PI image combine process can deal with this because it estimates the noise in each image and applies weights accordingly. I never switch sub lengths so I haven't tried this myself yet, sorry.
 
I usually don't change exposure times but with the ST10XME and the blooming, I sometimes need to try to tame the blooms as best as possible. It really gets ugly when on a bright open cluster like M44 where the blooms actually dissect stars above and below. The normal bloom removal processes I've used will usually destroy the star or wipe it out completely. Going too short of an exposure leads to way more noise. My thoughts often wander to the likes of a non blooming camera like the STL-11000.......

Thanks all for the responses. I'll give the new linear fit clipping rejection algorithm a go.

Steve
 
Imaging with a NABG camera has its advantages and disadvantages, like most things :)

Let us know how it works out. I recommend creating 3 stacks, one for each exposure and one for the total pile. Then compare the tree. I've had long discussions about this topic before and I'm not convinced yet that you truly can combine dissimilar exposures. It seems that when the two stacks have different SNRs the combined SNR will be lower than the best of the two. As lights get scaled up to match signal levels the noise also gets scaled after all. Anyway, let us know how it works. Perhaps PI has some magic pixy-anti-noise dust.
 
Juan Conejero said:
Hi Steve,

Just combine all of them in a single operation. Use ImageIntegration and be sure that the Weights parameter is set to Noise evaluation (this is the default setting). Our noise evaluation image weighting algorithm will automatically multiply each of your images by the optimal scaling factor to yield the highest SNR improvement in the final result.

If you have 25 images in total, I'd suggest the new linear fit clipping rejection algorithm. See this post for detailed information.
Click to expand...

Juan - does this imply that using different length exposure subs with a weighting from sub frame selector that includes fwhm and eccentricity will not work and that unless noise evaluation is used only equal length subs should be used? 
 
Hi,

I use a DSLR and pre-process each batch of images (different ISO/exposures) to the debayer point.  I then do a StarAlignment of all the batches and continue processing.  I guess for CCD images this is different.

Look up
John
 
russp said:
Juan - does this imply that using different length exposure subs with a weighting from sub frame selector that includes fwhm and eccentricity will not work and that unless noise evaluation is used only equal length subs should be used?
Click to expand...

The Integration will still work but you'll get reduced SNR if short subs are weighted highly because of good FWHM and eccentricity.  I'd be inclined to use a weighting expression that includes SNRWeight as well as the other parameters.

Cheers,
Rick.
 
You must log in or register to reply here.
Back
Top