Yoga Dress Model Video · öffentlicher Open Skill
Yoga Dress Model Video
Yoga Dress Model Video ist ein öffentlicher Open Skill auf Digen. Öffne ihn, um zu sehen, was dieser Workflow erstellt.
Health
11 0 4
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4167672/2026-10-05/7g94egzx.jpg
# Yoga Dress Model Video Turn one or more model photos into a short video of the model(s) doing a yoga routine while wearing a yoga dress, optionally following a reference video's routine. ## Preset assets - `model_1` / `model_2`: the user's two model photos (curvy female models). Use them as reference images when the user says "my models" or doesn't supply new photos. If the user uploads new model photos this turn, prefer those over the presets. ## When to use - User wants a video of their model(s) doing yoga in a yoga dress. - User provides model photo(s) and/or a reference video link and wants the routine recreated. - User wants to redo, tweak, or continue a previous yoga-dress model video. ## Steps 1. **Gather inputs.** Identify the model photo(s) (user uploads, or preset `model_1`/`model_2`). Note the optional reference video: a YouTube/TikTok/Instagram link, an uploaded video, or none. 2. **Understand the reference routine (only if a reference video is given).** - YouTube/TikTok/Instagram link: call `download_video(url=...)`, then pass the returned `downloadUrl` to `recognize_video(video_url=..., prompt="Describe the yoga routine: the poses, sequence, pace, camera angles, and setting.")`. - Uploaded video asset: call `recognize_video(video_url=<asset url>, prompt=...)` directly (do NOT call `download_video` on an uploaded asset). - Facebook/X/LinkedIn link: call `fetch_video_info(url=...)` for metadata only, then tell the user we can't see the actual routine from that platform and ask them to upload the video or give a YouTube/TikTok/IG link. - No reference video: proceed with a standard, graceful yoga flow (e.g. warm-up → standing poses → balance → floor stretch). 3. **Generate a yoga-dress still (optional but recommended for consistency).** For each model, call `generate_image` with `model='i2i'`, `input_urls=[model photo]`, prompt describing the same model now wearing a fitted yoga dress (sports top + leggings style) in a calm yoga studio setting, matching the reference setting if known. Pass `orientation='portrait'`, `name='<model> yoga dress still'`, `entity_type='characters'`. 4. **Generate the yoga video.** For each model, call `generate_video` with `model='ref2v'`, `input_urls=[the model photo or the yoga-dress still]`, prompt describing the model doing the yoga routine (from step 2, or a standard flow) while wearing the yoga dress, with smooth natural movement. Pass `orientation='portrait'` (9:16, social) unless the user asked for landscape, `resolution='720P'`, `duration` ~5–10s, `name='<model> yoga video'`, `entity_type='final_videos'`. 5. **Combine (optional).** If the user wants both models in one video, call `mix_clips` with the per-model clips concatenated, `orientation='portrait'`, `name='yoga dress models combined'`, `entity_type='final_videos'`. 6. **Reply.** Summarize what was generated in natural language (don't paste raw URLs). If a reference video couldn't be read (e.g. Facebook), say so honestly and offer the alternative. ## Notes - Video generation does not support NSFW/adult content on any channel; keep the yoga content tasteful and clothed. - Never hard-code physical model names; always use the channel IDs above (`i2i`, `ref2v`). - If the user is unhappy with a result, redo the affected stage (still or video) rather than the whole pipeline. - Intermediate step-by-step products (yoga-dress stills) will be visible on the canvas alongside the final videos.