title	emoji	colorFrom	colorTo	sdk	sdk_version	app_file	pinned	license	short_description
HugMaFaceWithYourThighs	🏆	green	gray	gradio	5.49.1	app.py	false	apache-2.0	Text to image generator that allows the use of ant mode

HugMyFaceWithYaThighs

HugMyFaceWithYaThighs is a Gradio application designed for Hugging Face Spaces that pulls together base checkpoints and creative add-ons like LoRA, LyCORIS, embeddings, and more into a single control room. It acts as a lightweight launchpad for experimenting with text-to-image models that run on your private Space GPU.

Features

• 🔍 Base model explorer – Start with the top 20 public text-to-image checkpoints or search the Hugging Face Hub right inside the interface.
• ➕ Additional resources manager – Mix in LoRAs, embeddings, ControlNets, VAEs, and other assets. Adjust weights and see trigger words before committing.
• 🎛️ Full prompt stack – Prompt & negative prompt fields, CFG and step sliders, denoise control, batch sizing, seed randomization, and the classic width/height selectors with quick presets.
• 🖼️ Inline gallery & metadata – Generated images appear alongside the metadata that produced them, so every run is reproducible.
• 🎨 Custom dark theme – A deep blue base with pops of yorange and light blue so long sessions stay easy on the eyes.

Getting started

1. Install dependencies

   pip install -r requirements.txt

2. Run locally

   python app.py

   The Gradio interface will launch in your browser. If you are testing on CPU, stick to smaller resolutions and simpler models.

3. Deploy on Hugging Face Spaces

   • Create a new Gradio Space and upload the repository contents.
   • Ensure the Space has access to the checkpoints and resources you plan to use (gated/private models require approval).
   • Pick a hardware tier that can comfortably host the selected pipeline.

Tips & roadmap

• LoRA handling – LoRA and LyCORIS resources are applied automatically when Diffusers adapters are available. Other resource types are tracked for reference and will appear in metadata for manual workflows.
• Trigger words – When available, trigger tokens from model cards are surfaced so you can drop them into your prompt quickly.
• Extensibility – The code is structured for future additions such as custom schedulers, ControlNet preview routing, img2img/inpainting workflows, or job queues.
• Caching – Model pipelines are cached in-process to avoid re-downloading within a session. Remember to restart the Space after major updates.

If you run into missing features or have new ideas, open an issue or extend the generate_images function to hook into your preferred pipelines. Have fun experimenting!