ANIMATIC
Rough first pass on Animatic
Overall Idea: Transforming myself into a toy collectable/action figure that escapes my packaging and driving a toy car while I explore the room I am in.
Goals:
- Develop a ComfyUI to Houdini Pipeline for speeding up time.
- Localize all AI generation offline and learn the ways around the models: WAN VACE, QWEN-Image, HiDream, Hunyuan3d and Flux.
- Create a fun visually beautiful short film that has the audience and myself smiling constantly.
Action Figure Concept
The Concept art was generated based off an actual photo of me using QWEN Image edit.
The character sheet allows me to use Hunyuan3D or any other image/multi-image to 3D for quick initial asset creation and then will refine with better topology in Houdini and Maya.
Modeling the plastic joints will be done in refinement manually and the materials will be done either procedurally or in Substance. This is a Hero Asset I want to keep this as high fidelity as possible.
These are the possible cars I am thinking about playing with. I want to still stylize and play with proportions more but I am leaning the most towards the Porsche 911.
These are the Cars after a restyle in ComfyUI using bytedance USO with Flux. Really happy with these stylized results.
Turned the cars into convertibles using QWEN image edit.
Converted Image to 3D model using Hunyuan 3d 2.0 directly in ComfyUI (Proxy Geo – Still needs proper cleanup and attention but speeds up the process tremendously)
