FLUX:

Flux has better semantic understanding and better image quality, and has more generalization. You can try more gameplay

SDXL:

Improved semantic understanding, able to generate more interesting content,

Resolution sampling, etc. is recommended to be used according to the check XL used,

Sometimes if the content is not ideal, besides changing the prompt words, It's best to try different check models,

If using Comfyui to generate vertical images, it is recommended to add the built-in node "PatchModelAddDownscale". Otherwise, if the vertical screen size is too large, it is easy to produce portrait errors

Recommend using Juggernaut_X,Lora weight recommendation 0.5-0.7,

If there are many people, it is not recommended to use zoom in, otherwise the hands may deform easily,The image quality effect of high-definition amplification may not necessarily be better

You can try to generate more interesting effects with other Loras.

SD_1.5model

Used to generate photos of bar characters, realistic or anime models can be well adapted, It can also cooperate with different LORAs to generate portraits of different nationalities or images of different styles

The training set is 512×768, so it is recommended that this size, the weight 0.6-0.8 is better, and high-definition repair can be opened.

Common reference word: Bar photo, hair, jewelry, multiple girls, necklace, shorts, bra, earrings, nail polish, many people,glasses, jacket, round eyewear,skirt, etc. describe the character, or describe the depth of field, blurry background, blurry foreground, red background,red light,

Face repair with ADetailer.