Car Edit 4K · compétence Open Skill
Car Edit 4K
"Transforme fotos de carros em edições automotivas cinematográficas em 4K. A Car Edit 4K preserva o modelo, a carroceria, as cores e os detalhes originais do veículo, aprimorando iluminação, reflexos, nitidez e enquadramento para criar resultados realistas, sofisticados e impactantes."
Auto
1 0 0
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/3150498/2026-10-10/w66g45qc.png
# Car Edit 4K Create high-quality 4K car edits: cinematic car stills and short car-edit videos (the popular social-media "car edit" look), generated from a text description or from a reference car photo the user uploads. Optionally add a generated soundtrack and stitch multiple clips into one edit. ## When to use - User wants a 4K car image or a cinematic car-edit video (drift, night/neon, urban, showroom, rain, sunset, etc.). - User uploads a photo of a car and wants an edit that keeps that same car. - User wants to redo, restyle, or continue a previous car edit. ## Steps 1. **Clarify briefly** (only if not already clear): format (video / image / both), the car to feature (make/model/color or a description), the desired style/mood (e.g. night neon, urban, drift, showroom, rain, sunset), and orientation (default `portrait` 9:16 for social edits; `landscape` 16:9 if requested). If the user gives a one-liner, pick a sensible cinematic default and confirm in one line. 2. **Generate the car still(s)** with `generate_image`: - No reference photo → `model='t2i'`, `orientation` per request, `resolution='4K'`. Write a detailed English prompt: the car (make/model/color), the scene, lighting, camera angle, and the cinematic "car edit" mood (e.g. "cinematic low-angle shot of a matte black sports car drifting through a neon-lit city street at night, rain on asphalt, motion blur, dramatic rim lighting, ultra-detailed, 4K"). - Reference photo provided → `model='i2i'`, pass the uploaded image in `input_urls`, keep the same car and restyle the scene/lighting. - Pass `name='car_still'` and `entity_type='car_stills'`. 3. **For a video edit**, animate the still with `generate_video`: - Use `model='i2v'` with the generated still as the start frame (`input_urls=[<car_still asset_id>]`), `orientation` matching the still, `resolution='1080P'` (or `2K`; 4K is auto-clamped on video). Prompt the motion: e.g. "the car drifts past the camera, camera slowly tracking, cinematic, smooth motion". - Pass `name='car_edit_clip'` and `entity_type='car_edit_clips'`. 4. **Optional soundtrack**: if the user wants music, call `generate_music` with `model='music'` describing the vibe (e.g. "dark cinematic electronic beat, deep bass, energetic, for a car edit montage"). Pass `name='car_edit_music'` and `entity_type='car_edit_music'`. 5. **Optional mix**: if there are multiple clips (or a clip + music) to combine into one edit, call `mix_clips` with the clips in order, adding the music as an audio track. Pass `name='car_edit_final'` and `entity_type='final_videos'`. 6. **Reply** in natural language summarizing what was made (car, style, format, resolution). Don't paste raw URLs. ## Notes - Prompts for image/video generation must be written in English; talk to the user in their language. - Keep the same car consistent across stills/clips: when continuing, reuse the previous still as the `i2v`/`i2i` input. - Report failures honestly; never fabricate URLs. Video generation does not support NSFW/adult content — politely decline such requests. - Intermediate products (stills, clips, music) appear on the canvas alongside the final edit; that's expected. - For multi-step jobs, use `write_todos` to list steps and update as you go; after finishing, call `set_guidance` to suggest next actions (e.g. restyle, another angle, add music).