Image Generation and Editing

Supported workflows

Mode	Use it for
text-to-image	Create a new image from a prompt
image-to-image	Edit, restyle, or transform reference images

Support varies by model. Discover the current catalog before choosing one.

Generate from text

anycap image models
anycap image models <model-id> schema --mode text-to-image
anycap image generate \
  --prompt "editorial product still life, warm daylight" \
  --model <model-id> \
  --param aspect_ratio=16:9 \
  -o editorial-hero.png

The aspect_ratio example is valid only when the selected schema advertises it.

Edit a local image

anycap image models <model-id> schema --mode image-to-image
anycap image generate \
  --prompt "remove the background and preserve the subject" \
  --model <model-id> \
  --mode image-to-image \
  --param images=./photo.png \
  -o subject-transparent.png

Local files are uploaded automatically. You can also pass an HTTP(S) URL.

Multiple references

When the schema allows multiple images, pass a JSON array. Repeating the same --param images= flag overwrites the previous value.

--param images='["./structure.png","./palette.png"]'

Output

{"status":"success","local_path":"/absolute/path/to/editorial-hero.png","model":"<model-id>","credits_used":1,"request_id":"req_abc123"}

Use jq -r '.local_path' to extract the downloaded file.

Precise human-guided edits

Use Annotate to collect numbered spatial feedback, then pass the annotated image back to image generate in image-to-image mode.