16bit and 32bit files

Discussions of Electronically Assisted Astronomy using the Live Stacking feature.
Post Reply
rockenrock
Posts: 68
Joined: Tue May 03, 2022 12:53 am

16bit and 32bit files

#1

Post by rockenrock »

Hi Robin,

I activated my license tonight and have just live stacked with both display screen stretch reset and the large window histogram reset. I am using a 14bit QHY168C without debayer turned on. I am continuously live stacking 20 subs (for now). It is not by sequencer but the Save and Reset tick box in the live stack dialogue box.

After the 20 subs done are done SharpCap saves the image in both 16bit and 32 bit. The 16 bit file image has a background of approximately .01 Pixinsight. The 32 bit file has background of.5, and does not look nice when I apply a screen stretch it. The 16 bit file has a background of 0.015.

I would like to know why the 32 bit file looks so bad (poor contrast), and why the background is poor contrast? In Pixinsight I can change the readout integer range to 32 bit (4G), and the background level reading drops from .5 to .21, but that is far from .015.

The live stack bit depth starts at 14 bit, then after the 19th sub the bit depth is 18.2. So does this mean the 16 bit image file will not be a good choice?

Thanks,
Roger
User avatar
admin
Site Admin
Posts: 13177
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: 16bit and 32bit files

#2

Post by admin »

Hi Roger,

there are two types of 32 bit FITS files, integer and floating point. SharpCap creates integer files by default, which can have values from about -2 billion to +2 billion, SharpCap uses 0 to +2 billion to avoid negative numbers, but it sounds as if PixInsight misinterprets this as starting at 0.5 brightness.

For the files you have, you should be able to subtract 0.5 using PixelMath in PixInsight. Do make sure that PI loads them as integer, as I think it can sometimes incorrectly load them as floating point and make a right mess of things.

I think you should probably set the option to save 'floating point' 32 bit fits (in the saving settings), which should resolve the issue for future captures.

cheers,

Robin

PS... Although your final stack has more than 16 bits of data, in practice the lowest bits will be pretty much all noise, so for a fairly short stack of just 10 or 20 frames, the practical loss of info in 16 bit rather than 32 is pretty much zero.
rockenrock
Posts: 68
Joined: Tue May 03, 2022 12:53 am

Re: 16bit and 32bit files

#3

Post by rockenrock »

Hi Robin,
Thanks for your reply. I always try to search the forum before I post my question. But when my search term had "32bit" the returns were all about 64bit vs 32bit installs... By accident I found your May 26 post this morning on the same subject. So I will change to 32bit floating point and try it.

Also thanks for explaining the minimal loss of data if saving 16 bit, especially for low count live stacks.

We appreciate your patiently replying our posts!

Roger
Post Reply