With ACE++, we can maintain good consistency of characters and items when using Flux Fill, without any additional aids. Simply by loading the ACE++ LoRA model, it seamlessly integrates into the Flux Fill workflow, allowing anyone to get started quickly without any barriers.

When you want to maintain consistency in your character, use the "Portrait" group;

Upload your character image to node ”Edit Image“, Choose ACE++Portrait LoRA model, Enter prompt words, then click the Queue button!

If you want to swap faces, use the "Portrait With Reference Image" group;

Upload the character you want to keep to the node ”Reference Image“ and the background image to the node ”Edit Image“, Open 'Edit Image' in Maskeditor and draw the mask(where you choose to swap the face.), Choose ACE++Portrait LoRA model, Enter prompt words, then click the Queue button!

When you want to realize object migration, use the "Subject"Group;

Upload your object image to node ”Edit Image“, Choose ACE++Subject LoRA model, Enter prompt words, then click the Queue button!

Upload your object to node ”Reference Image“ and your background image to node ”Edit Image“, Open 'Edit Image' in Maskeditor and draw the mask(where you choose to place your object.), Choose ACE++Subject LoRA model, Enter prompt words, then click the Queue button!

It is recommended to use the default parameters within the workflow, but you can try random seeds.

Welcome to use the ACE++ on hugginface and modelscope:

https://modelscope.cn/studios/iic/ACE-Plus

https://huggingface.co/spaces/scepter-studio/ACE-Plus

The address of the ACE project homepage: https://ali-vilab.github.io/ACE_plus_page/

LoRA Download：

portrait：https://huggingface.co/ali-vilab/ACE_Plus/tree/main/portrait

subject：https://huggingface.co/ali-vilab/ACE_Plus/tree/main/subject

The address of the ACE project homepage: https://ali-vilab.github.io/ACE_plus_page/