This is a realistic/photographic/artistic style LoRA trained on Qwen-image. It performs pretty well across various style categories, and some optimizations have been made to address the issue of repetitive "Qwen faces."

You can check out the sample images for generation reference, as well as the images posted in the community feedback section.

Currently, generation is mainly done via WebUI, so you can also use WebUI.

Prompt Reference:

outdoor photography, beautiful woman portrait, summer white outfit, long ponytail, natural, gentle breeze, messy hair aesthetic, dynamic turn, street roadside, relaxed vibe, close-up shot, eye-level angle, close-up on eyes, wide-angle lens, overexposed, sense of depth, soft side light, light and shadow contrast, soft focus, film grain, Japanese fresh style, cool tone, lens aesthetics, cinematic texture, high-end feel, masterpiece,

Use Cases: Portrait photography, atmospheric posters, artistic photography

Recommended Model Parameters:

Sampler: Euler

Recommended Dimensions: 3:4 or 9:16 both work fine

Sampling Steps: 40~60

CFG scale: 4

Model Weight: 0.95

Reprinted from: Qwen - Ultra-Realistic Portrait - Xiaohongshu Fashion Photography - LoRA