AutoStretch view is different in both Siril & PixInsight

[RichStevenson]

I stacked in Siril and then straight to GraXpert running AI BGE and AI Denoise, save processed as 32bit fits, then open in that fits in Siril or PixInsight, the image looks a lot different than what is shown in GraXpert. I think it may just be the autostretch percentage, but that can't be changed at least in Siril. Have not seen this issue in previous versions of GraXpert, so I'm not real sure if it's the AutoStretch or something else?

EDIT: Just discovered you can see the same issue within GraXpert.
-Open stacked fit from Siril in GraXpert (15% Autostretch). No processing was done in Siril.
-Crop
-Run AI BGE & Denoise
-Save as 32bit fits
-Now load that saved fits file again into GraXpert, and the same results shown in the images are seen. It's a completely different looking image.

Opened in GraXpert

Opened in Siril

Thanks!

[B1Gt0ne]

I am getting exactly the same issue. When I re-open the file in Siril I have to flip the horizontal geometry

[tbgh166]

Confirmed, same issue:
-Open stacked fit from Siril in GraXpert 3.0.2 (15% Autostretch). No processing was done in Siril.
-Crop
-Run AI BGE & Denoise - the image, as viewed in GraXpert:

-Save the image as 32bit fits
-Now load that saved 32bits fits file back again into GraXpert, - it's a completely different looking image:

[ajeb78]

@B1Gt0ne The orientation issue is probably due to the row order that many cameras store data in. The FITS standard says FITS image data should be stored bottom up, whereas many CMOS cameras produce data top down. There are some attempts to standardize this by introducing the ROWORDER FITS header, but it's not done entirely consistently between applications and therefore fixing it is really awkward (mainly because of images with encoded Bayer patterns). Happily, fixing the orientation is always a horizontal mirror (or plate solve) away :)