Bursting Out - Images

Bursting Out - Image Model

This image model was based on a similar data set that drove early versions of the Wan 2.1 Bursting Out LoRA. All faces in the dataset were cropped out, so the model should not have any face bias if you want to apply it to your favorite character.

Important: This model is 100% fictitious and is not based on images of any real person. All source images were generated from still image checkpoints using Stable Diffusion. No images (neither actual nor AI-generated) of any real person was ever involved in the making of this model.

The first release of this model is for the Z-Image-Turbo renderer. The speed and realism of Z-Image are very impressive, so it's quickly becoming a favorite for people whether they're just making images or if they're making starting frames for image-to-video clips. Either way, you can now dress your female characters in clothes that are far too small for them.

I recommend using a strength of about 0.75, but feel free to play with it to find your own best settings. I also prefer to use the dpmpp_2s_ancestral sampler paired with the EulerDiscreteScheduler from FlowMatch which you can install in ComfyUI using this custom node.

Limitations:

• This image model does not do plain t-shirts well, as it keeps trying to add chest windows and other odd fabric constructions to show cleavage.

• This model also does not do shots from behind very well. It's probably the fault of my image dataset, which didn't include many images from that angle.

I hope you enjoy using this model, and please don't forget to tag this model page when you post images so I can see what people are making with it. Have fun!