guoyww/AnimateDiff — reverse-engineered prompt

Build me a simple local demo for AnimateDiff where I can type a text prompt and generate a short animated GIF or video using Stable Diffusion 1.5 style community models.

I want it to feel easy for a normal user, not like running research scripts. It should install the needed Python requirements, automatically download checkpoints the first time if the project already supports that, and save outputs into a samples folder with clear names. Please include a Gradio web page like the existing demo, with controls for the prompt, negative prompt, model preset, motion option, seed, steps, and output size if those are supported.

Also make sure the command line examples still work for generating animations from the provided config files, including the community model examples, MotionLoRA example, and SparseCtrl RGB or sketch examples. Add a short README section explaining how to run the local app and where the generated animations go. Look up current docs online if you need to.