ImageIntegration evaluates noise even when unchecked

[pfile]

maybe this is not a bug, but it's kind of annoying when integrating flats or bias or any other frames for which you don't necessarily want to evaluate noise:

even if weighting by some other criterion than noise, and even if "evaluate noise" is unchecked, ImageIntegration still seems to do MRS noise evaluation on every sub. this kind of slows things down a bit.

of course the "evaluate noise" checkbox probably refers to the result image of ImageIntegration, not the subs.

expected behavior?

[zvrastil]

I noticed that yesterday as well. At first I considered it a bug. But on the second thought, I believe computed value is stored in the file cache for further integrations of the same set of files with different parameters. I did not tried to uncheck "Use file cache" though, to see if it changes anything.

regards, Zbynek

[pfile]

yes just now i was thinking the same thing, but for some types of frame i think it does not make sense to do a noise-weighting. so while on some lights you might go back and change your mind, on flats or something like that you'd probably never switch to noise weighting.

i should try unchecking the "use file cache" and see what happens. i'm in the middle of some OIII flats.

[georg.viehoever]

I noticed this behaviour as well. I find it a bit annoying because MRS noise evaluation never succeeds with my DSLR images, so this (from my point of view) is just  a waster of CPU time. Anyway, it should be possible not store the information "No noise evaluation" in the cache file, and compute them only when required.

Georg

[NKV]

I noticed this behaviour as well. I find it a bit annoying because MRS noise evaluation never succeeds with my DSLR images, so this (from my point of view) is just  a waster of CPU time. Anyway, it should be possible not store the information "No noise evaluation" in the cache file, and compute them only when required.

+1

[Juan Conejero]

Noise evaluation can be prevented if you disable the file cache feature when you integrate a particular set of images. If the cache is enabled, noise statistics are necessary because they are part of each cache record and there are other data that depend on them.

I find it a bit annoying because MRS noise evaluation never succeeds with my DSLR images

As noted above, noise evaluation and file caching should be disabled when integrating master calibration frames. In this way no noise evaluation routine will be applied to your CFA/Bayer frames.

On the other hand it would be possible to automatically disable noise evaluation for calibration frames. This implies that we should have a reliable way to identify the type of frames that we are reading. It seems that the FITS header keyword IMAGETYP is being used for this purpose by most image acquisition applications, so I'll add this feature in the next version of the ImageIntegration tool. However, I can't see a way to identify the type of frame in DSLR images (if anybody knows one, please let me know), so disabling file cache and noise evaluation seems your only resort.

[georg.viehoever]

Why dont you store some value "no noise evaluated" in the cache, and update records when required. I find this more intuitive that some obscure automatism that may or may not work.
Georg

[Juan Conejero]

Why dont you store some value "no noise evaluated" in the cache, and update records when required

But, to achieve this "no noise evaluated" cache state we should have a way to tell II something like: "please don't evaluate noise for these images". We already have two checkboxes to say: "don't use the file cache and don't evaluate noise for the integrated image".

I think the idea here is that II should be able to circumvent noise evaluation automatically for those images that don't require it, i.e. calibration frames. I was thinking on adding a checkbox such as "Don't evaluate noise for calibration frames" (enabled by default). It should work fine with the IMAGETYP keyword for most CCD images, but I see no way to implement this automatic feature for DSLR images...

[georg.viehoever]

Well, "don't use the file cache" has a "dont evaluate noise" as a rather surprising side effect. It was my impression that you would use the file cache not only for noise numbers, but also for other data. If this is not your intention, rename "dont use file cache" to "dont cache noise data in files".

Georg

[pfile]

this shouldnt be so hard. the proposed solution does not work for DSLR users, so it's worthless to me. how about, as georg suggests, fixing the semantics of the check boxes and possibly adding one more?