ComfyUI Node
FrameThrower Reference Node
FT / FrameThrower reference frames. Search the film-still library and output the frame, its scene description, its credit line, and optional depth / DW pose / lineart.
FrameThrower Reference Node
- image
- depth
- pose
- lineart
- prompt (text)
- credit (text)
◄query►
◄modehybrid►
◄index0►
◄pinned►
◄filters►
◄smarttrue►
◄auto►
◄lineart_strength3.0►
◄query_in—►
CategoryFrameThrower
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| query | STRING | — | |
| mode | COMBO | hybrid | 2 options: hybrid, description |
| index | INT | 00–499 | — |
| pinned | STRING | — | |
| filters | STRING | — | |
| smart | BOOLEAN | true | — |
| auto | STRING | — | |
| lineart_strength | STRING | 3.0 | — |
| query_inopt | STRING | Words to search for, from another node. Automatic — the grid follows the wire as you type upstream, no queue needed. |
Outputs (6)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | The frame itself. Into Preview Image, Save Image, or a VAE Encode / ControlNet / IPAdapter that takes an IMAGE. |
| depth | IMAGE | Depth map. Into a depth ControlNet. Runs only while this socket is wired. |
| pose | IMAGE | Pose skeleton, OpenPose layout. Into a pose ControlNet. Runs only while this socket is wired, and is empty when nobody is in the frame. |
| lineart | IMAGE | Lineart. Into a lineart ControlNet. Runs only while this socket is wired. |
| prompt (text) | STRING | The frame's scene description, as plain text — not conditioning. Into CLIP Text Encode, which turns it into conditioning. To read it on the canvas, add the node called 'Preview as Text'. Not a Text node: that one makes text rather than showing it. |
| credit (text) | STRING | Title, year and director, as text. Into 'Preview as Text' to read it, or 'Save Text' to write it beside the picture, so the attribution travels with whatever you make. |