JoyBoy FAQ: local AI, addons, mature packs, and project mode

[Senzo13]

JoyBoy FAQ and Support Thread

Use this thread for recurring questions about JoyBoy setup, local models, addons, packs, VRAM behavior, and project mode. If a question becomes a concrete bug, we can move it to an Issue.

Common questions

I have 8 GB VRAM. Can I still use JoyBoy?

Yes, but use smaller/quantized models and expect image/video workflows to be slower. JoyBoy tries to unload conflicting resources before heavy jobs, especially Ollama before diffusion/video on low VRAM.

Why does JoyBoy unload or keep models depending on the task?

The runtime uses a job/resource model. Chat, diffusion, video, downloads, and imports do not all need the same resources. JoyBoy tries to avoid useless load/unload cycles while still freeing VRAM when a heavy job needs it.

Can I import models from Hugging Face or CivitAI?

Yes. The model importer can resolve provider URLs and store local imports. Some provider downloads may require a key. Imported checkpoints may be stored as source files while JoyBoy applies a runtime loading strategy such as INT8 when possible.

What are addons/local packs?

Addons are local extensions that can add presets, prompts, routing behavior, model sources, UI surfaces, or workflow helpers without polluting the public core.

Are mature packs allowed to be discussed here?

Neutral, project-focused discussion is okay: architecture, install flow, safety constraints, compatibility, and docs. Do not post explicit media, sexual content, sexualized demos, illegal content, non-consensual content, or anything involving minors. Mature packs are optional third-party/local packs, not part of the public core.

What is the difference between Issues and Discussions?

Use Discussions for questions, ideas, workflows, setup help, and design tradeoffs. Use Issues for confirmed bugs, documentation gaps, and scoped implementation tasks.

How can I contribute?

Start with the good first issues label, read the contributor guide, and open a PR from a branch. Direct pushes to main are protected for maintainers.

When asking for help

Please include:

• Your OS and GPU/VRAM.
• The model or addon you are using.
• The exact command or UI action.
• Relevant console logs.
• Whether the issue happens after pressing "Liberer VRAM" and retrying.

Keeping questions structured makes it much easier for contributors to help without guessing.