18236583. Photo Relighting Using Deep Neural Networks and Confidence Learning simplified abstract (Google LLC)
Photo Relighting Using Deep Neural Networks and Confidence Learning
Organization Name
Inventor(s)
Tiancheng Sun of Mountain View CA (US)
Yun-Ta Tsai of Los Gatos CA (US)
Jonathan Barron of Alameda CA (US)
Photo Relighting Using Deep Neural Networks and Confidence Learning - A simplified explanation of the abstract
This abstract first appeared for US patent application 18236583 titled 'Photo Relighting Using Deep Neural Networks and Confidence Learning
Simplified Explanation
The patent application describes an apparatus and methods for applying lighting models to images of objects using a trained neural network. Here are the key points:
- A neural network is trained to apply a lighting model to an input image of an object.
- The training process utilizes confidence learning based on light predictions and prediction confidence values associated with the lighting of the input image.
- A computing device receives an input image of an object and data about a specific lighting model to be applied.
- The computing device uses the trained neural network to determine an output image of the object by applying the particular lighting model to the input image.
Potential Applications
This technology has various potential applications, including:
- Computer graphics and animation: The ability to accurately apply lighting models to images can enhance the realism and visual quality of computer-generated graphics and animations.
- Product visualization: This technology can be used to simulate different lighting conditions on product images, allowing customers to see how a product would look under various lighting scenarios.
- Virtual reality and augmented reality: By applying lighting models to virtual objects in real-time, this technology can improve the immersion and realism of virtual and augmented reality experiences.
Problems Solved
This technology addresses the following problems:
- Accurate lighting simulation: By training a neural network to apply lighting models, this technology can provide more accurate and realistic lighting effects in images of objects.
- Time-consuming manual adjustments: Previously, adjusting lighting in images required manual adjustments, which can be time-consuming and labor-intensive. This technology automates the process, saving time and effort.
- Inconsistent lighting conditions: Different lighting conditions can affect the appearance of objects in images. This technology allows for consistent lighting application, regardless of the original lighting conditions.
Benefits
The benefits of this technology include:
- Improved visual quality: By accurately applying lighting models, the visual quality of images can be significantly enhanced, resulting in more realistic and appealing visuals.
- Time and cost savings: Automating the lighting adjustment process reduces the need for manual adjustments, saving time and labor costs.
- Consistency: This technology ensures consistent lighting application across different images, eliminating variations caused by different lighting conditions.
Original Abstract Submitted
Apparatus and methods related to applying lighting models to images of objects are provided. A neural network can be trained to apply a lighting model to an input image. The training of the neural network can utilize confidence learning that is based on light predictions and prediction confidence values associated with lighting of the input image. A computing device can receive an input image of an object and data about a particular lighting model to be applied to the input image. The computing device can determine an output image of the object by using the trained neural network to apply the particular lighting model to the input image of the object.
