Poser could be the ‘killer app’ in creative AI, in terms of usable graphics production for storytelling.
Imagine an AI that takes what you see in the Poser viewport, and works on that, giving you 98% consistent character renders which would allow the creation of graphic novels etc.
The aim would be to keep all character details consistent and stable, while also ‘AI rendering’ the viewport into a consistent professional ‘art style’. Auto-analysis of a quick real-time render from the Viewport might be needed (already here, elsewhere) and auto-prompt construction (already here). Perhaps there might be some on-the-fly LoRA training going on too, behind the scenes. Then, the AI image generation would be done.
You can kind of do all this now, outside of Poser, using Poser renders. But what if it was all neatly integrated into Poser, and ran on SDXL? All those royalty-free runtime assets then become super-valuable, since with their aid you can easily get the AI to do exactly what you want. Face, expression, pose, clothes, camera-angle. Hands. All output by the AI to the usual masked .PNG file, ready to drop over a 2D backplate in Photoshop. In three clicks. And all consistent between images, enough to satisfy even the most fersnickety regular comics reader.
The aim would not be to go wild, but to get something very close to the arrangement and content seen in the real-time viewport. It doesn’t necessarily have to be done by Bondware/Reallusion either, since Poser is Python 3 friendly and very extensible. All that would be needed, perhaps, would be to open up PostFX to be able to run a Python plugin that applies its own FX on real-time renders.
So imagine Poser’s Comic Book Preview or Sketch rendering, but done by an AI on a purpose-built ‘AI-native’ PC (coming in 2025 in retail, if not before). With No Drawing Required™ and Character Consistency.™ Let’s call it SuperFlying Dreamland.™ 😉