Generative 3D: The Next Frontier in Game Production and Mirage's Pioneering Role
Mirage 3D: Mirage is at the forefront of generative 3D, building upon the works of Nasir, NVIDIA, Google, and OpenAI. They have optimized various open-source repositories to produce usable game assets, especially in the GLTF format. They have now open-sourced Mirage3D, a platform optimized for generating GLB files. The goal is to create a central hub for open-source generative 3D models that can produce usable 3D game assets.
Challenges in Generative 3D: Generative 3D development lags behind other domains like audio, video, and images because of the scarcity of 3D data online. While there are some open-source 3D models, they still fall behind the quality and efficiency of closed-source models.
Notable Developments:
- Open-Source: OpenAI’s Shap-E and the One-2–3–45 technique are among the few standout open-source repositories. Shap-E, for example, can generate meshes from text or images in just about a minute.
- Closed-Source: Companies like CSM.ai and Luma AI have made significant strides. CSM.ai’s image-to-3D service, CSM Cube, has shown remarkable results, while Luma’s Imagine3D is touted as the best text-to-3D model currently available.
The Future: Rumors suggest NVIDIA is collaborating with Turbosquid on a text-to-3D model, and other companies like Midjourney are set to release similar models. Additionally, gaming giants like Epic Games might be delving into 3D model creation because of their vast asset libraries.
Impact on Gaming: Generative 3D holds the promise to revolutionize various industries, particularly the gaming industry. With the ability to rapidly produce high-quality 3D assets, game production can be quicker, more diverse, and innovative.
Mirage Space: https://app.mirageml.com/
Mirage GitHub Repo: https://github.com/MirageML/Mirage3D
-
This discussion was modified 2 years, 8 months ago by
Joker69.
- 0 Replies
Sorry, there were no replies found.
Log in to reply.