UGC T-Shirt Video Ad Maker · Open Skill 公开技能

UGC T-Shirt Video Ad Maker

Creates a vertical UGC ad video: a model wearing a t-shirt with the user's print design walks confidently toward the camera while the camera tracks the movement from different angles. For TikTok / Instagram / Facebook

@myonlineavenue

Shopping

3 0 1

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/582503/2026-10-11/2jytawqx.jpg

# UGC T-Shirt Ad Maker

Creates a vertical UGC ad video: a model wearing a t-shirt with the user's print design walks confidently toward the camera while the camera tracks the movement from different angles.

## Preset asset
- `tshirt_print`: the user's uploaded t-shirt print design image. Use it as the reference for the graphic printed on the t-shirt.

## When to use
- The user wants a UGC ad video for a t-shirt design.
- The user wants to redo or change one step (different model gender/age, different scene, regenerate the video).

## Steps
1. First use `write_todos` to list the steps and update the whole list as you progress.
2. Confirm the uploaded t-shirt print design (preset asset `tshirt_print`) is available. If it has not been uploaded yet, ask the user to upload it first before proceeding.
3. Ask the user for the model's **gender** (male / female) and **age** (e.g. 20s, 30s, 40s). Wait for their answer before proceeding — never guess.
4. Ask the user to choose the **scene**: photo studio, street, restaurant, or park. Wait for their choice.
5. Generate a portrait image of the model wearing a t-shirt with the print design:
   - `generate_image`, `model='i2i'`, `input_urls=[tshirt_print]`, `orientation='portrait'`, `resolution='720P'`
   - Prompt (in English): describe the model (gender, age, confident casual look) wearing a plain t-shirt that displays the uploaded print design on the chest, standing in the chosen scene, natural lighting, full-body shot.
   - `name='model_tshirt_<scene>'`, `entity_type='characters'`
6. Generate the UGC ad video from that image:
   - `generate_video`, `model='i2v'`, `input_urls=[<model image asset_id>]`, `orientation='portrait'`, `resolution='720P'`, `duration=5`
   - Prompt (in English): the model walks confidently toward the camera while the camera tracks the movement from different angles, smooth cinematic motion, the t-shirt print stays clearly visible.
   - `name='ugc_ad_<scene>'`, `entity_type='final_videos'`
7. Deliver the final video to the user in natural language. Call `set_guidance` to suggest next actions (e.g. change the scene, change the model's gender/age, regenerate the video).

## Notes
- Always confirm gender, age, and scene before generating — never guess them.
- If the user wants a different scene/gender/age, regenerate from step 5 (new model image) then step 6.
- Report failures honestly; never fabricate URLs. Politely decline requests for prohibited content.
- Intermediate products (the model image) will be visible on the canvas alongside the final video.