Solving issue PNG vs. FITS

[Warren Schultz]

Came across something odd, while trying to figure out why an image wasn't solving.

As a test, I took a FITS image and ran it through the solver. This worked correctly, and allowed me to set the parameters as shown in the image.
I then converted the image to a PNG, and attempted to solve, which failed.
I generated the quads for both images, and you can see they're significantly different.

Any insight into this? I've attached the FITS and PNG in question.

[han.k]

Hello Warren,

ASTAP can SAVE or EXPORT as PNG. The save option takes the data from the screen including annotations. This is always a 3x8 bit colour operation. The export signal is very depending on the sliders. You get what you see.

The export PNG is a full 16 bit or 3 x 16 bit operation. The routine has some extra options like save as stretched or unstretched. If you export it as unstreched PNG 16 bit (see save as type) , then the tetrahedrons shaped quads are the same. The data in the PNG and FITS is fully identical.

This indicate again that 8 bit and stretching is less suitable for solving.

ASTAP version 0.9.421 will be released next hour. I overhauled the mosaic (image stitching method) option. Feedback and bug finding is most welcome. :)

Clear skies, Han

[Warren Schultz]

Ah, that makes sense. I'll check out the new version, and try with an unstreteched image.

Thanks!

[Warren Schultz]

Yup, works perfectly with the unstretched PNG. Thanks again.