SEEDANCE 2.5 prompt generates photorealistic late-night gym vlog that beats the 'AI slop' look
SimplyAnnisa · x · 2026-09-03
The author shared the full prompt used with SEEDANCE 2.5 (via TapNow) to generate a 30-second, 16:9 late-night gym vlog featuring a Korean woman in her mid-20s.
The key trick: deliberately instructing the model to shoot like a friend holding an old consumer DV camcorder — raw, slightly shaky, imperfect and spontaneous, explicitly "not cinematic or commercial" — to avoid the polished AI-generated look. The author says the result feels way too real. A useful prompt-design reference for video generation creators.
Related event: SEEDANCE 2.5 generates hyper-realistic gym vlog from one prompt(3 posts)→
More from Multimodal
- Stanford's LILIE loss cuts text-to-image gFD by 14 points over pixel-only training — StanfordUniversity · 2026-10-21
- Ex-Google Brain/DeepMind team launches Vyb, an AI design model with editable layered output — testingcatalog · 2026-10-21
- MOSS-Image's text encoder is multimodal and accepts image inputs — RandumbRedditor1000 · 2026-10-21
- Making a classroom case-explainer video with Opus 5.5 — technollama · 2026-10-21
- FastVideo releases FastH3 Omni weight collection with FP8 and NVFP4 ComfyUI builds — multimodalart · 2026-10-21
- Opus 5.5 distills a 30-min pi-durable lecture into a 7-min narrated explainer video — solyarisoftware · 2026-10-21