Motion Control Through Seed And Stochasticity Parameters

An image-to-video and text-to-video model developed by Niobotics ByteDance.

Unique: Exposes seed and stochasticity parameters at the diffusion sampling level, allowing users to control the randomness of the noise injection process and achieve reproducible or varied results without modifying the underlying model weights

vs others: Provides more granular control than simple 'deterministic vs random' toggles because it allows continuous adjustment of stochasticity levels, enabling users to find the right balance between reproducibility and creative variation