Demo for ML-Intern-lab/Qwen-Image-2.1-viewpoint-orbit-LoRA, a camera-orbit LoRA for Qwen-Image 2.1 trained with ML Intern in HuggingChat on ML-Intern-lab/gso-orbit-rgba.
Upload one transparent PNG, pick a rotation (45 to 180 degrees, relative to the uploaded view) and an elevation, and get the same object from the new viewpoint with transparency preserved. A checkbox runs the base model on the same prompt and seed for comparison; the turntable button chains seven views into a GIF.
Non-commercial research demo. The base model and adapter are under the Qwen Research License. Built with Qwen. Training renders are from Google Scanned Objects (CC-BY 4.0).
Demo for ML-Intern-lab/Qwen-Image-2.1-viewpoint-orbit-LoRA, a camera-orbit LoRA for Qwen-Image 2.1 trained with ML Intern in HuggingChat on ML-Intern-lab/gso-orbit-rgba.
Upload one transparent PNG, pick a rotation (45 to 180 degrees, relative to the uploaded view) and an elevation, and get the same object from the new viewpoint with transparency preserved. A checkbox runs the base model on the same prompt and seed for comparison; the turntable button chains seven views into a GIF.
Non-commercial research demo. The base model and adapter are under the Qwen Research License. Built with Qwen. Training renders are from Google Scanned Objects (CC-BY 4.0).