I'm trying to create a motion study by stacking three-hundred images of someone dancing. Is there any way to automate the selection of the dancer in each image by giving photoshop the background, and telling it to select the difference between the subject image and the background image?
I've already tried stacking the images as a smart object and using the various "stack modes" and I've also tried "Auto Blend- Layers".
You can have blending mode difference. This show differences between pixels so if no difference, then pixel is black. Difference is per channel so you will need convert it into grayscale, apply inversion, and then create mask from this which will be applied on same layer which was duplicated before.