Creator shares GPT Image + Seedance 2.5 prompt combo for realistic selfie-style AI videos
techhalla · x · 2026-09-28
Creator techhalla demos a repeatable workflow for ultra-realistic selfie-style videos: generate a photorealistic initial frame with GPT Image 2.5, then feed it into Seedance 2.5 with a specially crafted prompt. He claims it "works every time" and has published 7 examples so far, with more coming, all collected in his Space for others to try.
More from Multimodal
- RTX 5090 Owner Generates 180 Clips Locally to Build a Transformation Video in ComfyUI — Clasyc · 2026-09-28
- CYCLE: An AI-Generated Horror Short Film Worth Watching — Direct_Shirt8320 · 2026-09-28
- RPG v7 for Krea 2 Lands in Beta on CivitAI, Tuned for Comfy — anashel · 2026-09-28
- A One-Click Tool for Consistent Character Prompts in Midjourney/SD — Guilty_Service1262 · 2026-09-28
- Google Flow Spotted Referencing Nano Banana 2.1, New Image Model Release Imminent — koltregaskes · 2026-09-28
- BFS face & body swap LoRAs for Qwen Image 2.1 released with ComfyUI workflows — fruesome · 2026-09-28