WAN 2.2

151 Models

4.5K Posts

WAN 2.2 is the community tag for Alibaba's WAN 2.2 open video model family and everything built on it. The series is the current workhorse for AI video on Tensor.Art: text-to-video for clips written from scratch, and image-to-video (I2V) for animating a still frame — including the popular I2V High Noise checkpoints that push motion strength. Posts tagged WAN 2.2 are usually cinematic rather than gimmicky: film-look grading, camera-direction prompts ("cinematic medium shot", "slow dolly in"), shallow depth of field, and subjects like sorceresses in candle-lit rooms, city scenery, and character vignettes. Because the model responds strongly to camera language, prompts here read like shot lists, and many posts publish the full workflow so you can swap in your own start frame. Newer community builds may appear under WAN 2.1 / WAN 2.5 tags; if you are picking a model, compare run counts and open a few posts to check motion consistency before committing to a long queue.

Example Prompts

cinematic medium shot of a young sorceress seated at a wooden desk in a candle-lit room
warm fireplace bokeh in the background; Kodak-like film look
shallow depth of field
soft practical light
slight film grain. She wears a black dress with green embroidery and pink ribbons
ruby rings and long nails
a black choker with a green gem; long dark wavy hair
calm focused expression. Books