Qwen Image Edit Fusion

Blend products into new scenes with perspective & lighting correction

Model Overview

⚡️An image‑editing tool that fuses a product or object into a different background, automatically fixing perspective, lighting and seamless blending using a curated Fusion LoRA.

Best At

• Product placement: Merge a product into marketing mock‑ups or store shelving while ensuring realistic lighting and perspective.
• Image compositing: Replace or augment backgrounds in commercial photography without manual retouching.
• Creative scene building: Quickly prototype how a product will look in varied settings.

Limitations / Not Good At

• Works best when the source image has clear foreground/background separation; heavy occlusions may confuse the model.
• Extremely small or extremely large objects may not scale perfectly across all aspect ratios.
• Complex photo‑to‑anime or style‑transfer tasks are outside this preset’s scope.

Ideal Use Cases

• E‑commerce product mock‑ups (e.g., a phone on a modern desk).
• Social‑media posts that showcase items in new environments.
• Design comps for advertising agencies preparing product shots.

Input & Output Format

• Input: image (jpeg/png/gif/webp) required, prompt (optional text), aspect_ratio (string), go_fast (bool), num_inference_steps (int 1‑40), lora_weights (URL or slug), lora_scale (float), true_guidance_scale (float), seed (int), output_format (webp/jpg/png), output_quality (0‑100), disable_safety_checker (bool).
• Output: An array of URI strings pointing to the edited image(s).

Performance Notes

• Lightning preset (go_fast=true) delivers results in ~8 diffusion steps (fast), while the detailed path runs 40 steps.
• Outputs are served as webp by default (high quality 95).
• The model is GPU‑bound; batch size beyond single image may be slower.