Flux.1&Wan2.x img2img workflow with Teacache & Florence caption

<Wan2.x_ver.1.1>

We found that still images can be generated in Wan2.2 by replacing the checkpoint file with the “T2V_LowNoise model.”

We are also modifying the “Your own prompt” node.

<Wan2.1_ver.1.0>

Compatible with t2i using Wan2.1.

<ver.2.3>

Teacache and Florence2 nodes have been updated with ComfyUI v0.3.39.

<ver.2.2>

Fixed automatic/manual switch for prompts.

<ver.2.1>

Multiple LORAs can be set up in one node.

<ver.2.0>

1. node changed due to Teacache update.

2. added DetailDaemon node. Image information at upscaling is increased.

If you use GGUF for checkpoint files, please replace the loader node (it is FP file loader by default).

<ver.1.0>

The features of this workflow are as follows.

1. images created by SDXL, Pony, etc. are regenerated by Flux. a switch is provided to switch between the FP and GGUF model loaders, so you can use them according to your environment (“true” will switch to the GGFU loader).

2. The description of the image is automatically generated using Florence, but if you want to use your own text, change the switch to “true”.

3. If you want to change the image size, change the Latent scale magnification.

4. A teacache node is inserted to improve the generation speed.

Drop the json file or attached image into ComfyUI for use.

Have a good image generation life!