lazy Hunyuan Img2Prompt2Vid Workflow

Sick of keeping track of Lora triggers?

Sick of coming up with good captions when AI slop is good enough?

Image to prompt, add a photo and watch the magic of a poorly trained AI trying to figure out what that wolf is doing to that tiger and why they are in a Starbucks.

Florence takes forever to load, you can turn off unload model but you may run OOM.

Lora auto fills Triggers into the prompt from civitai

disable Character Lora if you are using a Lora with a trigger that auto populates and use one of the other ones.

I could have created a another Lora stack but again, I am lazy.

interpolation sucks. don't use it or just replace it with another one.

Refine is and upscale is a better combo.

Workflows for image gen and prompt were CTRL+C and CTRL+V from multiple others.

If you want credit you need to let me know and Ill give credit.

Lora logic is the only thing I added.

Vid generation is super fast, and super blurry.

Should be simple enough to replace the vid gen with another since there's only two connections.