Home
Apple Unveils LiTo Model Transforming Single Images to 3D with 37% Boost in Lighting Accuracy
Apple's AI research team has recently introduced a groundbreaking 3D generation model named LiTo (Surface Light Field Tokenization). This technology tackles a core challenge in 3D reconstruction by generating complete, high-fidelity 3D objects with realistic lighting from just a single 2D image.

The innovation of LiTo centers on a novel latent space and a unified 3D representation framework:
Efficient Encoding: It compresses intricate surface light field data into compact vector sets, mathematically capturing the object's geometry and its interaction with light.
Bidirectional Architecture: Using an encoder-decoder structure, the encoder extracts foundational geometric and appearance features. The decoder then reconstructs this data, accurately rendering advanced visual effects like specular highlights and Fresnel reflections.
Performance: Unmatched Lighting Consistency Across Views
LiTo was trained on a comprehensive 3D dataset containing thousands of objects. Key experimental findings include:
Eliminating Directional Bias: LiTo strictly adheres to the camera coordinate system, correcting the orientation errors common in other models.
State-of-the-Art Metrics: For multi-view lighting consistency, LiTo outperforms the previous leading model, TRELLIS, by approximately 37%.
This advancement significantly lowers the barrier for high-quality 3D content creation. It is poised to deliver superior asset generation for future augmented reality (AR) and spatial computing platforms, including devices like the Vision Pro.
Related article
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber
Related Special Topic Recommendations
Comments (1)
0/500
Apple's AI research team has recently introduced a groundbreaking 3D generation model named LiTo (Surface Light Field Tokenization). This technology tackles a core challenge in 3D reconstruction by generating complete, high-fidelity 3D objects with realistic lighting from just a single 2D image.

The innovation of LiTo centers on a novel latent space and a unified 3D representation framework:
Efficient Encoding: It compresses intricate surface light field data into compact vector sets, mathematically capturing the object's geometry and its interaction with light.
Bidirectional Architecture: Using an encoder-decoder structure, the encoder extracts foundational geometric and appearance features. The decoder then reconstructs this data, accurately rendering advanced visual effects like specular highlights and Fresnel reflections.
Performance: Unmatched Lighting Consistency Across Views
LiTo was trained on a comprehensive 3D dataset containing thousands of objects. Key experimental findings include:
Eliminating Directional Bias: LiTo strictly adheres to the camera coordinate system, correcting the orientation errors common in other models.
State-of-the-Art Metrics: For multi-view lighting consistency, LiTo outperforms the previous leading model, TRELLIS, by approximately 37%.
This advancement significantly lowers the barrier for high-quality 3D content creation. It is poised to deliver superior asset generation for future augmented reality (AR) and spatial computing platforms, including devices like the Vision Pro.
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber











