Preview when shooting flats is completely yellow - actual files are correct

[Victor]

This issue happens with the latest NINA Nightly (#144) but I noticed it with previous builds too.

When taking flats (either with the flat wizard or just imaging manually) with an Artesky flat field generator, at the correct exposure the preview is completely overblown yellow. The actual files, if opened in another fits viewer, are good and calibrate well. Furthermore, if reducing the exposure time just a bit, the preview shown by NINA is correct. It’s like at a certain exposure time, the preview goes crazy and jumps to all yellow.

The camera is an ASI 2600MC and the issue is the same with either native and ASCOM drivers. Enabling disabling the unlinked stretch option makes no difference. When the previews are completely yellow, it’s noticeable in the preview window too even when it’s not stretched.

The two images attached have only the exposure changed slightly, the flat panel brightness is the same. This was the most consistent way to reproduce the issue:

• set the flat panel to a fixed brightness and take longer and longer exposure. Note: when using the exposure chosen by the flat wizard (with target ADU set to about 30k for this camera, the preview is already all yellow)
• the files themselves don’t seem to be affected. Both the ones where NINA preview works and those where it shows yellow, opened in another program are fine and calibrate well.

‌

[Dale Ghent]

First, nothing is overblown - the image statistics that are displayed for both examples indicate nothing of the sort. The means are < 30k and only two pixels are at maximum value; likely broken pixels that are permanently on.

There doesn’t seem to be an obvious reason for the difference in debayering. The debayering is only for visual use only in the image display screens and is a basic method provided by Windows itself. This is interesting in that it’s an unexpected visual artifact but it’s not something that impacts the actual saved image itself as that is still saved in the bayered, raw form as it ought to be.

[Victor]

Correct, the image itself is fine, the issue is just the preview. Not impacting actual imaging but it’s interesting why NINA is doing this.

[Stefan B]

When the image histogram is above 50% (first image) the stretch is inverted resulting in the observed behavior. It's intended and similar to how pixinsight stretch functions.

[Stefan B]

Removing version: 1.11 Nightly (automated comment)