Gameplay Footage Generator · skill pública de Open Skill
Gameplay Footage Generator
Gameplay Footage Generator es una Open Skill pública en Digen. Ábrela para ver qué crea este flujo.
Other
9 2 4
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4005040/2026-09-20/jjveyaxq.png
# Gameplay Footage Generator Generates realistic, high-fidelity gameplay footage as vertical **9:16** video clips from any game concept or scenario. Prioritizes smooth motion, consistent characters and environments, and authentic game visuals (camera movement, environments, lighting, HUD elements, physics). Optionally adds voiceover narration and background music and combines everything into a single final video. ## When to use - The user wants a gameplay video clip from a text description (FPS combat, open-world exploration, racing, action, story moment, etc.). - The user provides a reference image (character, environment, or scene) and wants a gameplay clip that keeps that subject consistent. - The user wants to iterate on a gameplay clip (redo, change the scenario, adjust camera/lighting). - The user wants voiceover narration and/or background music added to the gameplay clip(s), or wants multiple clips combined into one final video. ## Steps 1. Clarify the gameplay scenario if vague: game genre, setting/environment, camera style (first-person, third-person, cinematic), mood/lighting, and any HUD elements. Confirm it's a single clip or a sequence, and whether the user wants voiceover and/or music. 2. Decide the generation mode based on what the user provides: - **Text only** (no image): call `generate_video` with `model='t2v'` and an empty `input_urls`. - **Reference image(s) for consistency** (character/environment/scene): call `generate_video` with `model='ref2v'` and pass the reference image asset_id(s) in `input_urls` (at least 1 required). - **Single start image → animated clip**: call `generate_video` with `model='i2v'` and pass the start image in `input_urls`. 3. On every `generate_video` call pass: - `orientation='portrait'` (9:16 vertical). - `resolution='720P'` by default (raise to `1080P` when the user wants higher fidelity). - `duration` ~5–8s per clip. - `name`: a short shot title (e.g. "FPS corridor firefight"). - `entity_type='gameplay_videos'`. - `prompt` in English, describing the gameplay in detail: genre, camera movement (e.g. "smooth first-person tracking shot"), environment, lighting, HUD elements, physics/action, and the desired mood. Emphasize smooth motion, consistent character/environment, and high visual fidelity. 4. If the user wants a reference image created first (no image provided but wants a specific character/environment), call `generate_image` with `model='t2i'`, `orientation='portrait'`, `entity_type='characters'` (or `locations`), then use that image as the reference for the video step. 5. **Voiceover (optional):** if the user wants narration, call `generate_tts` with `model='tts.design'`, `text` = the narration script (in the user's language), and `voice_instruction` in English describing the desired voice (e.g. "deep, energetic male narrator, action-movie trailer tone"). Pass `name` (e.g. "gameplay narration") and `entity_type='voiceovers'`. 6. **Background music (optional):** if the user wants music, call `generate_music` with a `prompt` describing the style and mood (e.g. "high-energy action game soundtrack, driving electronic beat, cinematic tension"). Pass `name` (e.g. "gameplay soundtrack") and `entity_type='music'`. 7. **Combine (optional):** if the user wants a single final video, call `mix_clips` with `clips` listing each gameplay video (asset_id in `url`) and the voiceover/music as audio tracks (asset_id in `audio`/`audios`, with `volume` and optional `delay`). Pass `orientation='portrait'`, `name` (e.g. "GTA gameplay montage"), and `entity_type='final_videos'`. 8. Report the result naturally (don't paste raw URLs). If the user wants changes, re-run the relevant step with adjusted prompt/parameters. ## Notes - Video generation does not support NSFW/adult content on any channel — politely decline such requests. - Never hard-code physical model names; always use the channel IDs above (`t2v` / `ref2v` / `i2v` / `t2i` / `tts.design` / `music`). - Repor