Subject-driven text-to-video model that keeps multiple reference characters and objects consistent across art styles.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
DomainShuttle is a research model for freeform, open-domain, subject-driven text-to-video generation. Given reference images of subjects (characters or objects) and a text prompt, it generates video in which those subjects stay consistent, including when references come from mixed styles such as realistic, 2D and 3D.
According to a creator review, code has been released and the model is built on a 1.2 base model, with roughly 35 GB of weights and a need for a high-end GPU. The official page is a demo page and, in the material provided, gives little detail beyond the title. Vendor, license, exact release date and paper link could not be confirmed.
It is aimed at researchers and technically capable users who can run the model locally rather than through a hosted product.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Reference-to-video generator that keeps multiple characters and objects consistent, even across mixed art styles (realistic, 2D, 3D). Code released; built on a 1.2 base model, 35 GB, needs a high-end GPU.
Closest neighbours by meaning, not just by category.
Model · ByteDance
ByteDance research model unifying image and video understanding, generation and editing in a 3B active-parameter network.
Research
Research framework for long, consistent video generation using a persistent 3D world state as memory.
Model
Rank-asymmetric flow method for pixel-space image generation, incl. AsymFLUX.2 klein finetuned from FLUX.2 klein 9B.
Model
Research project for real-time autoregressive generation of multi-shot video narratives from shot-level prompts.
Research
Research memory router that lets a frozen autoregressive video generator reuse selected past KV entries for consistency.
Research · ByteDance
ByteDance and Texas A&M research method distilling image and video diffusion models into few-step generators adversarially.
