Found 1 document(s)
Page 1 of 1
-
Recent generic reference-based image generation models natively support multiple conditions by concatenating all reference image and text tokens into a joint sequence processed by a single Diffusion Transformer (DiT). Ho...