Full prompt revealed for AI-generated real estate tour videos with timestamped shots and lip sync
FinanceYF5 · x · 2026-10-27
Part 2 of the thread exposes the complete prompt Justine Moore used to generate a real estate tour video with GPT-6 Astra + Kling 4.
Workflow:
- Feed the Zillow listing page to GPT-6 Astra, which picks listing photos, writes timestamped shot prompts, and aligns narration with visuals
- Then pass 8 reference images plus the long prompt to Kling 4 to generate a 30s vertical video in one go
Prompt highlights:
- Visual rules: the agent stands on the real ground with believable proportions; foreground furniture may naturally occlude her lower body for depth; eye-level shots framed mid-thigh up
- Explicit anti-patterns: no tiny figure pasted at the bottom of a background photo, no slideshow or digital zoom of reference photos, no stretched rooms
- Character consistency: one Asian female agent throughout, with locked outfit, voice and lip sync; shot 5 is interior-only with voiceover
- Lines are given verbatim, with no music or improvised dialogue allowed
A reusable prompt template for AI video generation.
Related event: AI Real Estate Video Workflow Runs End-to-End with Kling 4(3 posts)→
More from Multimodal
- Uncensored Wan 1.3B fine-tune hits Hugging Face, runs video gen on 8GB VRAM — Promptmethus · 2026-10-27
- Illustration-to-real tiger headdress recreation wows with near-perfect detail match — gizakdag · 2026-10-27
- Gradium's open-weight TTS runs offline on iOS/macOS, benchmarked vs Kokoro and PocketTTS — amos_gyamfi · 2026-10-27
- SATE-Bench: 316 scenes show MLLMs stop exploring too early with high confidence — Tianchen Guo · 2026-10-27
- Two-year-old patch still fixes ComfyUI's slow checkpoint loading: 225MB/s to 2.5GB/s — Solongtomegrandma · 2026-10-27
- A lava-slide first-person video made with Dreamina for just 29 cents — Certain_Menu_8412 · 2026-10-27