Interesting.

I come from a 3D rendering background besides photography. So I am used to think as separated layers. My approach would be:

• Cut the object in one image without the shadow.

• Depending on the background, using all the image with shadow as a separate layer. Probably convert it to grayscale if the background is not neutral white.

We can go several ways for the shadows.

1. Using this grayscale image with a blending mode multiply and using it below the object layer. This is faster but only if you work with a layered method.

2. Using this image as a transparency mask for a total black layer. You need to invert the image. This second option is the one that can give you a single PNG with shadows included.

You can play with the levels of this mask and the curves, to adjust the intensity of the shadow and to clean the background.

You probably need to paint a bit on the borders of this mask so it does not show behind the clipped object.

---

An additional explanation for The comment about the mask.

1. Open your image, and make it a new layer.

2. Add a layer mask

3. Alt+click on it, and paste the image again inside. You have now a transparent image based on the information on the image. But the information is backwards, you need to invert it in order to work.

Apply this method to a black plate as described in the steps above.

Your shadow layer should look like this:

I have not masked the red dot layer, I only did the shadow part.