Hot Stars after recombining with starless

I’d play with the linear coefficients between stars and nebula when combining them in PixelMath. In your case you can start with a 0.5 reduction factor for the RGB stars to see if they are correctly exposed and then adjust further.

Another idea is to punch some “holes” in the starless image where there should be stars prior to recombination. This would prevent the color of surrounding background to be blended into the colors of stars, and to potentially cause overexposure. One easy way to punch these holes is to apply the mask produced by StarNet2, and dim the passthrough areas to near zero.

Nick Grundy:

Yuxuan:
dim the passthrough areas to near zero.

very interesting. That's what I thought was happening. I'll try that out

Let me know how it works out.

In the past I had the same issue and I had to simply dim the stars in PixelMath. It avoided overexposure but gave the stars an underwhelming look:

https://www.astrobin.com/wgzrza/C/ (those are SHO stars though)

I’d be happy to learn about other techniques.

Did you forget the ~ in the PM

~((~Starless)*(~Stars))

You can also create a luma mask from your starless image, put that inverted over your stars only image. then pull down the gamma curve to reduce the brightness of stars were your starless image is so bright it saturates the stars when recombining.