google/stereo-magnification
Neural network model for novel view synthesis from stereo image pairs using Multiplane Images with predicted alpha layers and color blending.
Not currently ranked — collecting fresh signals.
star history
This research project trains a deep learning model on stereo image pairs to synthesize novel viewpoints by predicting color images, alpha layers, and blending weights across multiple depth planes. The network supports different prediction modes for foreground and background color images and can be trained with either pixel-based or VGG perceptual loss functions. The trained model interpolates between views by blending source images at predicted depth planes.
Frequently asked
- What is google/stereo-magnification?
- Neural network model for novel view synthesis from stereo image pairs using Multiplane Images with predicted alpha layers and color blending.
- Is stereo-magnification open source?
- Yes — google/stereo-magnification is open source, released under the Apache-2.0 license.
- What language is stereo-magnification written in?
- google/stereo-magnification is primarily written in Python.
- How popular is stereo-magnification?
- google/stereo-magnification has 416 stars on GitHub.
- Where can I find stereo-magnification?
- google/stereo-magnification is on GitHub at https://github.com/google/stereo-magnification.