fire / looking-glass

Some scripts for testing with the Looking Glass 3D display.

Concepts:

• "quilt image": NxM views collected in a single image, usually 2048x2048 or 4096x4096. This is the standard method used by the Looking Glass tooling to store multi-view content. To display a quilt image on a Looking Glass the per-device calibration values are used to generate a native image on the fly, usually in done a fragment shader.
• "native image": An image in the native resolution of a Looking Glass display, 2560x1600 for a small LG model, which can be directly displayed on the device. This image is only viewable correctly on the specific device being targeted, as it uses the per-device calibration values.

Scripts:

• get_calibration_from_eeprom.py: Get the display calibration values (in the form of a JSON string) from a Looking Glass. Save this to a .json file, as some of the other scripts need these values.
• make_quilt.py: Generate a standard quilt from a set of view images
• gen_numbers_quilt.py: Generates a quilt where each tile shows the view number. This can be used to (try to) understand how the individual views (and their pixels) are shown on the Looking Glass
• quilt2native.py: Takes a standard quilt and outputs a native image, based on a device's calibration values. This should match what the official LG tools do.
• linquilt2native.py: Takes a linear quilt (all tile images side-by-side, i.e. num vertical tiles = 1) and outputs a native image.
• frames2native.py: Reads a set of separate tile images and output a native image.

The blender subdirectory contains a patch for Blender 2.79b for rendering quilt images directly (without having to go through rendering 45 separate views and combining them). See the README.md there for more information.