STEP BY STEP POSE-BASED TEXTURE BLENDING FOR ALITA

01 ESTABLISH NEUTRAL POSE

This will be your rest state, the base pose for your look development. The mesh topology and UV layout needs to be the same across all of your poses or blendshapes.

02 ALIGN EXTREME POSES WITH NEUTRAL

If you are working from scan data the alignment may take some work here, and grid warping is probably your best bet for that. If you are working from sculpted poses or blendshapes the displacement bake will be the same process you go through to get from sculpt to render geometry for your neutral mesh, but the albedo work will probably be more subjective – collect reference for the pose you are re-creating, not just the final result. Be sure to also collect information like thermal images, which will give you more insight into what’s happening under the skin and guide where you need to make changes.

03 DEFINE THE MAXIMUM EFFECT

Again, if you are working from sculpted poses rather than scan data, reference is your friend here. How much should the forehead and cheek blood flow increase when your character is angry? How much blood should drain when they are scared? How furrowed should their brow be? The key is that your character is in pose, so any displacement and albedo adjustments are done completely in context.

04 DEFINE THE TRIGGER MECHANISM

The last part of this is to define when, and by how much, these maps should be mixed in. This could be as simple as keyframing a variable or user attribute that your shader is looking for, or for a more procedural approach you could associate (for example) a ‘flushed’ amount with relevant face shapes in your data set, so as those shapes are triggered your shader will automatically receive the associated ‘flushed’ value.

For example, the above image is a combination of many shapes, but the shapes ‘Cheekraiserl’ and ‘Cheekraiserr’ have a ‘flushed’ value of 0.9 associated with them. At this particular frame they are at 0.87 extent, so the final ‘flushed’ variable on Alita’s skin geometry at this frame would be 0.783, and so the shader would mix 78.3% towards the fully flushed albedo map. Easy! Tip of the hat to the incredible Olivier Lesaint who implemented this for us on