This LoRA can generate cartoon scenes with empty foregrounds.
Recommended size is 896*1536, with an iteration count of around 20-50, and a LoRA weight of 1, which can be flexibly adjusted as needed.
The platform sampler used is FlowMatchEuler, euler.
Prompt suggestions can be based on example images, and you can add the prefix "dt" and include "nothing on the ground to cover."
Looking forward to seeing your generated images!!!
Description
Recommended size is 896*1536, with an iteration count of around 20-50, and a LoRA weight of 1, which can be flexibly adjusted as needed.
The platform sampler used is FlowMatchEuler, euler.
Prompt suggestions can be based on example images, and you can add the prefix "dt" and include "nothing on the ground to cover."
Looking forward to seeing your generated images!!!
FAQ
Comments (2)
Thank you for your great work！It's a truly valuable lora since most of the AI painting now focus on foreground people but not background scene. And could you maybe share some information about the data you used for training this lora? I'm planning to train my own lora on a different base model.
Sorry, it's been a while, I can't find the data on the cloud platform, the training material can be dozens to hundreds of pictures, I use the traditional tag, or you can use https://github.com/jhc13/taggui?tab=readme-ov-file. This tool is used to label the pictures, and the other parts are dim64/32 or 32/16, 15_xxx, max epochs 8. I hope it will be useful to you.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.