Ostris AI Toolkit adds Qwen-Image-2.1 training with native alpha channel support
ostrisai · x · 2026-09-21
Ostris announced that its open-source AI Toolkit now supports training Qwen-Image-2.1, including native alpha channel support.
The author tested it on a small t-shirt design dataset: no trigger word or dedicated alpha text was needed—just default sample prompts—and results looked good after only 750 training steps. Native alpha support, he argues, unlocks many new workflows for transparent assets, stickers, and print design.
Related event: Ostris AI Toolkit Adds Qwen-Image-2.1 LoRA Training Support(3 posts)→
More from Multimodal
- Indie Britpop AI music video built with Nano Banana Pro, Kling, and free lip-sync via WAN2GP + LTX 2.5 — cocktailpeanut · 2026-11-12
- P2Echo: one text-prompted model segments multi-view echocardiography with high accuracy and low cost — UBC-V · 2026-11-12
- Redditor makes fast-motion short film with H3 and ships Longshot Studio v0.18.5 with face swap — R34vspec · 2026-11-12
- Reimagining Knight Rider's opening as a Netflix drama with Claude Fable 5.2 — justin_hart · 2026-11-12
- Gemini 4 Carbon builds a full 3D aquarium in Blender and Three.js in ~6 hours — sunjiao123sun_ · 2026-11-12
- Awaken Open-Sourced: Turn a Half-Duplex Speech LLM Full-Duplex With Just a Small LoRA — alexcovo_eth · 2026-11-12