How to Build 3D Models Using Pictures from Different Angles

This patent describes a system that creates a three-dimensional digital object by combining multiple flat images taken from different viewpoints, then lets a user fine-tune how those images wrap around the object.

What it covers

The system first takes several "textures" (pictures) of an object, each showing a different side or view. It then automatically creates a rough 3D shape of the object and tries to stick these pictures onto the right parts of that shape, which is called an "initial alignment" (Abstract). A user can then give "input" to adjust how these pictures are placed and stretched on the 3D model (Abstract). Finally, all these separate pictures are blended into one single, complete picture that covers the entire 3D model, defining how it looks from all angles. This finished 3D model, along with its combined picture, is then saved. For example, if you wanted to create a 3D model of a toy car, you would take pictures of its front, back, sides, and top. The system would then build a car shape and let you adjust how those photos fit onto the digital car.

What it doesn't cover

• —Generating 3D models solely from a single photograph or texture.
• —Creating 3D models without needing multiple textures from different views of the object.
• —Systems that generate 3D models without any user input to refine texture placement or alignment.
• —Methods that create 3D models using only depth data, LiDAR scans, or other non-texture-based inputs.
• —Generating 3D models where the multiple textures are not combined into a single, unified texture.

The clever bit

The core idea is taking multiple flat images, each from a different view, and intelligently combining them into a single, seamless texture that accurately covers a 3D model, allowing for user-guided refinement to ensure visual quality.

Why it matters

Creating realistic 3D models is crucial for many digital experiences, especially in extended reality (XR) like virtual and augmented reality. This method helps streamline the process of making digital objects look accurate and detailed by efficiently using multiple visual inputs. This is important for industries building virtual worlds, training simulations, or digital product prototypes.

Real-world examples

• 1.Creating assets for video games
• 2.Building virtual objects for augmented reality apps
• 3.Generating digital twins of real-world items
• 4.Producing visual effects for movies
• 5.3D scanning services for e-commerce