I have problem with flipping in KStars/EKOS

Left is Siril, right FITS viewer, it is M81 as you can see from plate solving, but it is flipped.

Camera is placed Up when in parking position.

In my touptek IMX585 cooled camera there are no FLIP options in INDI, but I think that Siril have a correct FLIPPING (it is not a rotation problem).

This photo was taken yesterday in Italy, in the FITS header I can see ROWORDER Top-Down.

Maybe that FITS viewer ignore FLIPPING ?

As normally a "flip" is called the maneuver of the mount (or the scope) when changing pier side (and is basically a rotation) I prefer to say "reflected" (just what happens with a mirror) than "flipped". Now because of this reflection in the two images I think it has to be some issue regarding row order. Indeed EKOS does save images in top-down order, as you said, but normally does NOT display them reflected. After first load SIRIL on the other hand does.
For analyzing the situation could you please
- run the image plate solver in SIRIL to see what the result is?
- compare the PA's of the image in FITSViewer and in SIRIL respective (after plate solving)?

Ok, on mirroring and rotaion.

Using Stellarium I can see true Up from my location and time (2023-12-15 4:54 AM from Italy)

Images from stellarium and Siril are not mirrored (I can see using inclination of M82 and rotation of the branch of M81), plate solving on Siril is wrong.

In FITS viewer image is incorrectly Mirrored. (and rotated but this is not so important).

FITS Viewer read images top-bottom instead of bottom-up like some apps do it. You'll find the FITS Viewer images matches with what you see in KStars sky map.

... plate solving on Siril is wrong...
That is what I supposed! Plate solving in SIRIL shows wrong parity, because there is something wrong with row order in the image.
Just try the following:
Start simulators and take an analogical picture of M81 (galaxies do not appear, but star patterns are recognizable). Open this image in FITSViewer and see the correct alignment. Now open the same image with SIRIL: It's mirrored! Only after plate solving in SIRIL the image is correct, as it should be. SIRIL recognizes the row order and reflects ("flippes") the image.

... In FITS viewer image is incorrectly Mirrored. (and rotated but this is not so important)...
No, that's wrong: In FITSViewer the image is only mirrored!

BTW, did you analyze the PA's of the images? I bet these are inverted.

Images that came from Touptek camera in FITS viewer is mirrored Top-Down, I can't mirror it in INDI properties (there aren't any MIRROR VERT or HORIZ like ZWO), but I don't know why Siril remirror it, so Image is correct in that program.

Simple KStars show FITS images from top down while Siril show them from bottom up.

So software doens't know if the image if mirrored? Touptek driver doen't specify the correct way to display image?
In FITS file there is a ROWORDER parameter.

> Siril will always read and display images in the bottom-up order, however if the top-down information is specified in the keyword, then Siril will demosaic the image with the corrected pattern.

siril.readthedocs.io/en/latest/file-formats/FITS.html