how Flux.2 Klein support both T2I and TI2I #12989

Activity

1. yiyixuxu commented on Jan 16, 2026

   

   yiyixuxu

   on Jan 16, 2026

   Collaborator

   More actions

   hey @Bili-Sakura
   flux2 klein support image input https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/flux2/pipeline_flux2_klein.py#L611
2. Bili-Sakura commented on Jan 17, 2026

   

   Bili-Sakura

   on Jan 17, 2026

   Author

   More actions

   hey @Bili-Sakura flux2 klein support image input https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/flux2/pipeline_flux2_klein.py#L611

   Got it. But could you refactor with explicit Text2Image and Image2Image pipeline class inherit from that for better readability.

   As to my understanding, most T2I model support this sort of I2I by simply replace noise to input image latent which is more likely to an inversion style. However, the image editing model behaves differently.
3. Bili-Sakura

   changed the title [-]Add Flux.2 Klein Image2Image Pipeline[/-] [+]how Flux.2 Klein support both T2I and TI2I[/+] on Jan 23, 2026