One sentence to a moving night sky: single-GPU self-hosted pipeline with Z-Image, BiRefNet, Qwen-Image-Edit and Wan 2.2
llama-of-death · reddit · 2026-10-29
A Reddit user demos Guaardvark (MIT-licensed, Linux + one NVIDIA card), a self-hosted pipeline taking one sentence to two video clips in a 2:32 run:
- Z-Image Turbo text-to-image: paints a crystal stag over waterfalls; same seed reproduces it exactly. One prompt per line yields a sapphire wolf, rose-quartz whale, jade koi dragon.
- BiRefNet cutout: "Cut the stag out of this picture" returns a clean background with every antler tine intact.
- Qwen-Image-Edit 2509: outpaints to 16:9 (400 px per side), then one sentence turns day into a moonlit aurora night while keeping everything else in place.
- Wan 2.2 14B films the day scene, then the day-to-night transition.
All local, one GPU—a directly replicable local media-generation stack.
More from Multimodal
- Adobe's EditAgent routes image edits across editors with evidence-based listwise judging — adobe · 2026-10-30
- Short film with an all-AI cast is indistinguishable from real actors, says viewer — TotGaH · 2026-10-30
- ElevenLabs SFX v3 handles multi-sound prompts that broke earlier AI SFX tools — HeyAmit_ · 2026-10-30
- IntReal: 100K Real-World Interleaved Multimodal Samples to Benchmark Unified Models — nanjinguniv · 2026-10-30
- Tencent's WeVisCut trains a 9B agent to run the whole editing room, using 33x fewer tokens than Opus — tencent · 2026-10-30
- Hand-drawn-style AI short celebrates Korean Seollal with a good-news magpie — narrator_mochi · 2026-10-30