oxoria 2026.1.3

Stand on the shoulders of giants

Oxoria 🎨

A modern, open-source reference management tool for artists and 3D creators.

What is Oxoria?

Oxoria is a feature-rich reference management tool designed for digital artists and 3D creators. Whether you're painting, illustrating, or modelling in 3D, Oxoria helps you collect, organize, and retrieve your visual references — all in one place.

Try it now: pip install oxoria → launch with oxoria

---

✨ Key Features

🔍 Semantic Search

Find references by meaning, not just filename. Type a concept like "religion" and Oxoria will surface relevant results such as "shrine" — powered by natural language understanding applied to your own image memos.

📸 Instant Screenshot Capture

No need to open the app first. With Oxoria Screen Capture Util running in the background, press Cmd+Shift+O anytime to capture your screen and import it directly into your reference library.

🔁 Duplicate Detection

Oxoria stores a perceptual hash for every image. Re-importing the same image won't create duplicates — your library stays clean automatically.

📦 Fully Portable

Your entire reference library lives in a single central repository folder. Copy it to a new machine and pick up exactly where you left off.

🐍 Python API

Every UI action and internal operation maps to a Python function. Combine them freely to automate workflows, build custom pipelines, or extend Oxoria however you like.

---

🛠️ Tech Stack

Category	Details
Language	Python 3.13
GUI	PySide6
ML / Search	transformers, torch, faiss-cpu, optimum[onnxruntime]
Image Processing	Pillow, ImageHash
Utilities	numpy, psutil, setproctitle

---

🗺️ Roadmap

• 3D Object Import — Import .obj and other 3D formats as interactive references via OpenGL
• Canvas Version Control — Git-backed versioning of canvas layouts using PyGit
• Window Overlay Mode — Keep references on top with an adjustable transparent window while working in other apps
• Partial C++ Migration — Performance-critical modules (e.g. 3D rendering) migrated to C++ as needed

---

⚠️ Known Limitations

• Startup time — transformers and torch currently add ~8 seconds to startup. A planned migration to a Rust-based tokenizer and a lighter ONNX runtime should significantly reduce this.
• Duplicate sensitivity — The current 16-bit dHash may flag near-identical images (e.g. expression variants) as duplicates. Upgrading to 32-bit dHash is planned.
• API coherence — The Python API is functional but spread across modules. A cleaner, more unified interface is in progress.
• Test coverage — Automated unit and regression tests are not yet in place; this is a known priority.

---

📄 License

Licensed under the Apache 2.0 License — free to use, modify, and redistribute for both personal and commercial purposes.

---

🔗 Links

• 📁 GitHub Repository
• 📦 PyPI Package

---

Oxoria is still under active development. Feedback and contributions are very welcome! A stable build is planned for release on Steam once the core feature set is complete.