Video Generation

Live catalog

Supported video models

Choose a model before you generate. Availability and modes are loaded from the current AnyCap catalog.

Loading live models…

Modes

Mode	Use it for
`text-to-video`	Create a video from a prompt
`image-to-video`	Animate one or more supported image references
`edit-video`	Refine a supplied video using natural-language instructions
`first-last-frame-to-video`	Generate a transition between named first and last frames
`multi-shot-video`	Generate a sequence from shot-level prompts and durations
`multi-modal-reference`	Guide a video with supported image, video, or audio references

A selected model supports only a subset of these modes. Query the live catalog instead of assuming a mode from this reference:

anycap video models \
  | jq -r '.models[] | .model as $model | .operations[]?.modes[]?.mode | [$model, .] | @tsv'

Generate from text

anycap video models
anycap video models <model-id> schema --mode text-to-video
anycap video generate \
  --prompt "slow camera move through a quiet greenhouse" \
  --model <model-id> \
  --param aspect_ratio=16:9 \
  -o greenhouse.mp4

Animate a local image

anycap video models <model-id> schema --mode image-to-video
anycap video generate \
  --prompt "gentle wind moves the leaves; slow camera push-in" \
  --model <model-id> \
  --mode image-to-video \
  --param images=./greenhouse.jpg \
  -o greenhouse-motion.mp4

The CLI validates and uploads the local reference, waits for generation, and downloads the result.

Schema-driven controls

Duration, resolution, aspect ratio, reference limits, and other controls differ by model. Read the exact selected mode before passing parameters:

anycap video models <model-id> schema --mode image-to-video \
  | jq '.schemas[0].schema.model_params'

Output

{"status":"success","local_path":"/absolute/path/to/greenhouse-motion.mp4","model":"<model-id>","credits_used":5,"request_id":"req_abc123"}

For a complete workflow, see Generate video from a local image.