SDXL:
Improved semantic understanding, able to generate more interesting content, everyone can try more.
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
I often use Dreamshaper XL/Jumpernaut XL/LEOSA and M HelloWorld XL for large models.
I haven't tested any other check, everyone can try more
Lora weight recommendation 0.5-0.7
However, there is no need to give up on 1.5. Currently, 1.5 has more ecology, and some images still require 1.5.
1.5 model:
Direct input Simple prompts for store categories can generate street shop facades, such as: coffee shop, flower shop, bookstore, etc
If not generated, you can add words like storefront, sign, entrance, etc., or increase LORA weight
For example: coffee shop entrance, coffee shop sign, coffee shop front. If it appears indoors and you want to be outdoors, you can add a hint word like "street" to control it
The weight is recommended 0.6-0.9, the higher the weight, the more obvious the store, the lower the weight, the better the generalization of the prompt word
checkpoint recommends the realistic class
Can have interesting effects with many architectural LORAs
Recommended size: 768×512, because I trained set uniform this size can also try others
