Qwen-Image-Edit-2511 vs SenseNova U1.5 Lite: hands-on multi-reference fusion comparison
daniel933912 · reddit · 2026-09-10
A Reddit user ran four sets of multi-reference image fusion tests pitting SenseNova-U1.5-8B-MoT-Preview against Qwen-Image-Edit-2511.
- Texture quality: Qwen wins on lighting, shadows, and overall texture fidelity
- Spatial understanding: SenseNova holds the edge — in the cat-on-toy-scooter test, Qwen hallucinated a weird pillar under the coffee table and misplaced the cat's front paw, while SenseNova avoided both artifacts
- Inputs mixed Krea-2 generated images and real photos, with prompts covering scene reconstruction, cross-image object transfer, and physical occlusion requirements
Both models are open source on Hugging Face (Qwen-Image-Edit-2511 and the SenseNova-U1.5 preview with a GitHub repo).
More from Models
- Local benchmark: Flint matches Qwen3.6 quality at 50/52 while running 3.6x faster on M4 Pro — RichardSocher · 2026-10-27
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27
- One stray 'u' and Claude starts spelling like a Brit — cantrell · 2026-10-27
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27
- ARC Prize redesigns leaderboard: highest reasoning effort by default, CSV export — mhmazur · 2026-10-27
- Cohere: open-weight Qwen3.8-Flash-Next tried uploading probe file to Beijing cloud bucket in sandbox — wstone_bd · 2026-10-27