ComfyUI Node
TS Angle Select
Point the camera at the subject and get the prompt that asks a model for exactly that view. Drag the camera around the stage, pick a height and a framing; the phrasing comes from the chosen preset, so the same three settings can drive different models.
TS Angle Select
- prompt
◄presetQwen Multi-Angle►
◄azimuth0►
◄heighteye-level►
◄framingmedium►
CategoryTS/Text
Inputs (4)
| Name | Type | Default | Description |
|---|---|---|---|
| preset | COMBO | Qwen Multi-Angle | Which model's vocabulary to speak. • Qwen Multi-Angle — Qwen-Image-Edit with the Multiple-Angles LoRA. The wording is the LoRA's trigger phrase, not a description: it was trained on these exact words, and the '<sks>' token has to be there. Presets are plain JSON in nodes/text/angle_presets — a new model is a new file, not a code change. |
| azimuth | INT | 00–315 | Where the camera stands, in degrees around the subject. 0 is in front, 90 is the subject's right, 180 is behind. Eight positions, 45 degrees apart — the same eight the model was trained on. Anything in between is snapped to the nearest one, because a phrase for it does not exist. |
| height | COMBO | eye-level | How high the camera sits. • low — below the subject, looking up. • eye-level (default) — level with the subject. • elevated — a little above, looking slightly down. • high — well above the subject, looking down. |
| framing | COMBO | medium | How much of the subject is in frame. • wide — the whole subject and its surroundings. • medium (default) — roughly waist up. • close-up — head and shoulders. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| prompt | STRING | The camera-angle prompt. Feed it to the text encoder — on its own for a pure re-angle, or joined with your own description. |