--Flux2 I used it with res2m+beta57 and guidance 4, and the results were great. The image quality is similar to my Lenovo Qwen, but I noticed that Flux2 is more flexible and handles prompt adherence better. Recommended weight: 0.85-0.9

-- Z-image Turbo Recommended weight: 0.80-0.9 I used the euler + sgm_uniform algorithm and set CFG to 4. P.S.: Currently, ComfyUI requires this custom node pack to properly load the LoRA: https://github.com/PGCRT/CRT-Nodes

--Chroma For fast generation, dpmpp2m + beta and CFG 3 work well. A more experimental setup is res_2s_rkmk2e + beta57 and CFG 3, which produces more precise models with better limbs and fingers, but it can introduce some strange graininess in tiny details, and it takes longer.

When using under Wan or Flux, add the trigger word: l3n0v0 Recommended LoRA weight: 0.8~1

Official website: https://civitai.com/models/1662740 While everyone is chasing ultimate image detail, surreal clarity, and idealized lighting in AI generation, another group of people is doing the exact opposite—working to make AI-generated images look closer to real-life photography.

They deliberately retain the optical flaws unique to smartphone cameras, strive for the natural and candid composition of street photography, and even intentionally preserve minor motion blur and imperfect lighting conditions. The "imperfections" in these generated images—slight out-of-focus elements implying a candid snapshot, uneven exposure recording real ambient lighting—endow them with a convincing sense of realism, creating the illusion that this is a "real moment captured by a lens."

Petite body, casual mirror selfie with soft grin, head slightly tilted, crop top and high-waist shorts, bedroom closet background with hanging clothes, long light brown straight hair parted center, oval face with gentle cheekbones, shot with a silver iPhone (three cameras) held by the subject, mixed natural and artificial light, everyday candid feel, amateur cellphone quality visible sensor noise, artificial over-sharpening, heavy HDR glow, amateur photo, blown-out highlights, crushed shadows.