Mixing Midjourney and ChatGPT to Create Cinematic Miniature Horror Scenes
umesh_ai · x · 2026-08-09
The author shares a stunning AI image generation workflow that combines a specific Midjourney style mixing parameter (--sref 3802363335) with ChatGPT to create a cinematic miniature horror scene.
The quoted original tweet details the exact prompt used: it requires an uploaded reference image as an exact facial identity anchor. While preserving zero facial drift, it generates a tiny version of the person standing on a weathered wooden table, holding a glowing candle and cautiously looking over her shoulder, creating a strong forced perspective and horror vibe.
More from Multimodal
- Developer Reports Extremely Slow VAE Decoding with Minimax Video Model — Glittering-Cold-2981 · 2026-08-09
- Seedance 2.5 vs. Minimax H3: A 30-Second Single-Take Video Generation Test — beatlepol · 2026-08-09
- Testing MiniMax H3: Complex Camera Movements and Beat Syncing via Prompts — Muted-Celebration-47 · 2026-08-09
- Kroma v0.2 Released: Full Fine-Tuned Model Based on Krea2 — mikemend · 2026-08-09
- HyperFrames MCP demo: Generate videos with sound effects directly in Claude — toolstelegraph · 2026-08-09
- Grok Imagine Prompt Engineering: Structuring Cinematic Realism — techhalla · 2026-08-09