Bias PreProcessing Issues

I’m having some odd discrepancies between preprocessed and non-preprocessed stack results. My only preprocessing is a bias correction. For some reason, while the non-preprocessed stack ends up exactly as expected, the bias corrected one results in an image that has blown out highlights and zero shadow detail. In AutoStretch/using the Histogram Transformation, all the pixels across the scene are either on or off, with each color channel having slightly different values. That specifically seems like it could be a 16/32 bit conversion or aliasing thing, because when I run an intense Asinh Transformation I can recover a very small amount of highlight detail, but the shadows are still all obliterated. I’ve processed both sequences the same in multiple different ways, with the only difference being the bias adjustment. What could be the cause of this? Is the bias overcorrecting and removing good data? Any help is greatly appreciated, thank you!

Hello, to make sure I understand correctly: in one case you just stack all light images, in the other you calibrate them by removing the bias then stack, and the result is not better? It should be very close, but if it’s bad it can be because the bias were not taken with the same offset and gain settings as the lights.

I triple checked, they’re all at the same settings. What’s especially weird is that the preprocessed stack gets worse with the more frames I add; a stack with just 5% of the total data yields a noisy result but some shadow data is clearly visible, but above 10 or 20% it gets completely destroyed. The shadow detail is getting averaged out for seemingly no reason, even though it’s visible in each frame individually.