One selfie in, video out: Seedance 2.5's character reference trick with prompt
techhalla · x · 2026-09-12
A user demonstrates that Seedance 2.5 can generate a video featuring a real person using just a single selfie as the character reference, calling the capability surprising. The full prompt is shared in the original post, so others can reproduce the workflow directly.
More from Multimodal
- Amazon FAR studies image tokenizers as visual languages in unified multimodal models — Amazon-FAR · 2026-09-12
- Paul Walker's brother says AI could revive the late actor in Fast & Furious using unused footage — Polymarket · 2026-09-12
- AI-Enhanced Helmet-Mounted GoPro Footage from D-Day — dgnr8ai · 2026-09-12
- AI short made with Magnific and ElevenLabs features unprompted synthetic Wahlberg — charis_ai · 2026-09-12
- OpenAI releases GPT-Image-2.5 in Flare and Sunburst variants, up to 50% faster — dl_weekly · 2026-09-12
- Non-AI footage blended with AI-generated video blurs the authenticity line — alex-224 · 2026-09-12