Saving Combined Files

[Martin]

Hello Group.
When I go to save a combined image as a FIT, I have several options.

If I save the combined image as a 16bit FIT (signed) the resultant image is fine, but if I want to save my image in a larger space, and select 32bit IEEE 754 Floating Point, PI strips the background out.  I mean, the resultant ADU value of the background is zero....so you have nothing to work with.  In the 16bit image, the background is in tact.  Why does PI do this?

cheers
Martin

[Carlos Milovic]

Hmmm. When PixInsight saves at 32bits fp, everything is rescaled to the normalized dynamic range. That means, 0 is black, and 1 is white. Data should not be destroyed in any way. If a STF or the Histogram shows data clipping in this process, then something very wrong happened. Please confirm that.

[Philip de Louraille]

I am not seeing what you are seeing. I am on OSX 10.6 using the latest PI. I opened an unsigned 16-bit FIT and saved it as a 32-bit FP and was able to reopen it with all its details.
Is the problem you are seeing due to opening the saved 32-bit FP file with another program or within PI?

[Niall Saunders]

Hi Martin,

If your original data (prior to 32-bit Float saving) was OUTWITH the [0.0, 1.0] 'native' range of PixInsight (which I suspect is maybe the case), then the way PI will interpret the 32-bitF data on re-load will vary depending on how you have configured the FITS format in the Format Explorer window.

Can you tell us what the current settings are for your PI implementation?

Cheers,

[Martin]

Gents
thanks for your replies, and sorry for the belated response.

So, my images are first calibrated in Maxim DL and saved as a 16bit signed FITS image.
I then open these calibrated images in PI, and perform alignment, and the alignment saves those aligned images also in 16bit signed FITS format.  If I then open those aligned images individually, they are fine, background in tact.

It is when I then take those PI aligned and saved images and combine them in PI that this 'error' occurs i.e when using 32bit FP format.

I am using PI versin 1.06.592 and I am using the Image Integration settings pretty much as per Harry's tutorial.

thanks
Martin

[Niall Saunders]

Hi Martin,

Please feel free to email me THREE of your images.

The first should be an example of a file 'straight out of' Maxim, after it has been calibrated.

The second should be one of your aligned images after this stage has been implemented in PixInsight

The third should be your 'ImageIntegrated' result

I need these in FITS format - so they might end up being huge files, perhaps too big for email. If that is the case, I would be happy to do a TeamViewer link-up to your PC, and 'examine' the files directly. Have a think and get back to me.

Cheers,

[Jack Harvey]

Back when I used Maxim to cal my images I think I had this issue but solved it by having Maxim save the calibrated, or aligned or integrated images in 32 bit floating point and not 16 bit unsigned.  So you might try that?

[Martin]

Hi Niall...I will get those images to you.

Jack, I did as you suggested.  Took my PI aligned images, saved them all as IEEE Float, combined them in PI, saved the resultant file as 32bit float but every single background value was zero, and star values were extremely low...so you have nothing to work with.

Martin