An AI dream: 100 seamless scenes generated overnight with no human input
wine_dark · reddit · 2026-08-19
The author let LTX 2.5 dream unattended overnight via a custom PyTorch script: each 13.3s scene uses the last frame of the previous scene as its first frame, yielding 100 scenes stitched into one video after 7 hours on an RTX 5090. Qwen 3.8 auto-wrote the continuation prompts from the last scene, producing a coherent AI dream of moons, stars, threads and foxes.
More from Multimodal
- Music video agent NoSpoon heads to production; dev debates one-shot vs granular control — Kyrannio · 2026-08-19
- Local MiniMax H3 img2video demo turns sketch into photorealistic scene — Important-Radish-722 · 2026-08-19
- Requesting lightweight ComfyUI faceswap workflow — Bossdon01 · 2026-08-19
- Reshoot-Anything: Self-supervised model for in-the-wild video reshooting — rsasaki0109 · 2026-08-19
- Opinion: Suno generates listenable sound bites but falls short on full tracks — eschadiol · 2026-08-19
- ComfyUI Workflow: Automated MiniMax Video Segmentation Upscaling — Francky_B · 2026-08-19