Qwen-Edit-2509-Upscale-LoRA

Still work in progress, checkpoints are being uploaded on the go

Current best: LoRA A: qwen-edit-enhance_64-v3_000001000, LoRA B: qwen-edit-enhance_000004250

V4 Candidates LoRA A: qwen-edit-enhance-v4-16_a_000003000/1000, LoRA B: qwen-edit-enhance-v4-64_a_000003750/4750

LoRA A and LoRA B are meant to be wired one after another

Comfy workflow is in the files

Demo: 🤗 Click here

Prompt: Enhance image quality, followed by textual description of the scene. The more descriptive it is, the better the upscale effect will be

All images below are from 8 step Lighting LoRA.

• ModelSamplingAuraFlow is a must, shift must be kept below 0.3. With higher resolutions, such as image 3, you can set it as low as 0.02
• Samplers - LCM (best), Euler_Ancestral, then Euler
• Schedulers all work and give varying results in terms of smoothness

Trained on a filtered subset of Unsplash-Lite and UltraHR-100K at 2.4MP

• Style: photography
• Subjects include: landscapes, architecture, interiors, portraits, plants, vehicles, abstract photos, man-made objects, food
• Trained to recover from:

• Low resolution up to 16x
• Oversharpened images
• Noise up to 50%
• Gaussian blur radius up to 3px
• JPEG artifacts with quality as low as 5%
• Motion blur up to 64px
• Pixelation up to 16x
• Color bands up to 3 bits
• Images after upscale models - up to 16x

Not trained on: 2d, illustrations, artworks, text, graphics, CGI