Need a script to turn an RGB document with four greyscale layers(C, M, Y and K) into a CMYK image

Excellent work as always r-bin, that script basically automates the method that I proposed in post #9… However your script leaves the source files open and creates a new file, while my action “consumed” the open images. Sadly both of these methods remove the original base filename.

So all that is left is to script a solution that will cycle through a folder of images, taking a set of 4 or 5 images at a time with the same base filename and C|M|Y|K|A suffixes and combine the images just as your binary, then save out a file and repeat until no files were left.

If there was a source folder of say 500 images, a script would ideally output 100 combined images (presuming that all had an alpha mask).

Wow, thanks for all these options already! To clarify, here are two screenshots about where I start, and what I would like to achieve. This would be the "easy" solution, where I still have to manually put together my C, M, Y, K and mask layers beforehand.

So basically I start with an RGB Photoshop file consisting of 5 layers. Each layer is an 8-bit greyscale image. The way these are generated is a bit more complicated, but long story short, our client wants 3D renderings of his product packaging, with a CMYK file as an output, that exactly matches the printing colors of his CMYK Illustrator designs. Since 3D renderings are always done in RGB color space, the solution was to render the C, M, Y and K components separately as greyscale images and combine them afterwards, which works perfectly. It's just a lot of work to do this manually. Here's the simplified starting point:

The result should be the first four layers correctly put into the channels of a CMYK document, with the fifth layer applied as a layer mask:

The action you showed in your animated GIF and images above already takes a care of most of this, apart from the masking, but worst case would be to apply these masks manually. Could you share this action?

Also you are right, I did propose a second solution, which would need to be scripted, basically picking a directory. The script would automatically combine all the images with the same name + suffix (C, M, Y, K, MASK) into the final output like my second screenshot. This would work for batch processing, without having to put together my layers beforehand.