Following on from my recent “How to extract skin and eyes for colour blending” post, I think I’ve now successfully cracked the ‘skin and eyes-only render’ problem for Poser 11. The extreme use-case is: how is your graphics software going to select just the skin on this render, by colour, without complex fiddling around in Poser with setting Toon_IDs for each material, setting up a Firefly render and masking?
Instead of trying to render just the skin from Poser, which does not seem possible in a form partly masked by clothes and hair, there’s another option. I instead have Poser render everything except the skin as black or dark grey.
To do this my Python script scampers through the Poser scene in a few seconds and looks for hair, props, conforming clothing. For anything that isn’t the figure itself. For what it finds, the script sets the diffuse colour nodes to black. A preview render is then made. As this is obviously a destructive script, after running the script the user is prompted to revert the scene to the last saved state.
The resulting real-time Preview render then has skin that can be easily selected and masked in a graphics programme, even if in the original scene the figure had red hair and was wearing a red outfit. Which would have confused the heck out of the software’s skintone selection process.
In PhotoLine (which I now prefer to Photoshop) an Action can automate the ‘Channel to Selection’ process and in a microsecond it has whisked out the skintones and eyewhites to their own layer…
Her eyes have also been selected, but that was because they were a soft skin-like hazel colour to match the outfit. This layer can then be used to restore colour to the skin after a render has been run through several filters and plugins to make it look hand-painted.
Photo.Net can also extract the skintones, using the Color to Alpha v2.2 plugin. Which is probably better if you need fine control over the eye-white selection. It’s very likely Photoshop also has such selection capabilities in its newer subscription version, though readers will have to discover those elsewhere.
Your (only) choice for a third-party Photoshop plugin appears to be Imagenomic Portraiture 3, which has automask of skintones, but which doesn’t return a selection mask (it all happens inside the plugin). It also fails on dramatically-lit pictures such as this…
The advantage of a scripted Poser render is that even if you have a dramatically lit ‘sunset forest’ scene like this, you’re still going to get a relatively easy mask. Because you’d just tweak the colour in the script from black to bright green, and render against a bright green plain background (hide all other elements). The masking out of the bright green should then be very easy, leaving only the skin and eyes.