Image Mixer is a fine tuned version of Stable Diffusion Image Variations that has been trained to accept multiple CLIP embedding concatenated along the sequence dimension (as opposed to 1 in the original model). At inference time, Image Mixer combines the image embeddings from multiple images (or text) to mix their concepts. The model was trained on a subset of LAION Improved Aesthetics at a resolution of 640x640.
How to Use Image Mixer
Image Mixer lets you fuse the concepts of two or more images. To get started, upload images you want to mix and select the mixing strengths of both images.