2D to 3D image conversion - GreycLab/gmic-community GitHub Wiki

This article aims to provide a rough guide to using Tom Keil's 3D Conversion filter for the G'MIC Gimp plugin, in the Stereoscopic 3D branch. Plain language has been used as much as possible and may not be technically accurate.

  1. Brief Settings Description
  • Stereoscopic settings
    • Scene selector: Choose the image scene type, automatic to let the filter decide
    • Far point deviation: Difference between each eye, affects how close objects look in 3D
    • Stereoscopic window position: Adjust viewpoint of the observer from left to right
    • Depth field control: How much distance between near and far objects
    • Feature analyzer threshold: Fine tune the edge detection
    • Feature analyzer smoothness: How much to smooth out edges
    • Local detail enhancer: Increase apparent depth of smaller bumps
    • DOF analyzer: How much to estimate depth based on how in-focus objects are
    • Frequency analyzer: How much to estimate depth based on daylight absorption
    • Flip left/right: Swap the left and right eye images
    • Use individual depth map: Use a custom depth map from the layer below instead of generating one
  • Image settings
    • Black & White: Convert the input image to black and white
    • Output format: The final 3D image type e.g. for red/cyan glasses
    • Gamma compensation: Fine tune the output image brightness
    • Anti-ghosting: Reduce "ghost" images of one eye's image being seen in the other
    • Color boost: Adjust the color saturation of the final image
    • Anaglyph glasses adjustment: Tweak colors to suit individual 3D glasses
    • Autocrop: Crop the sides of the image to avoid edge glitches
    • Bidirectional rendering: Move the viewpoint of both eyes not just one
  • Frame settings
    • Render multiple frames: Output multiple images from different sideways viewpoints, as layers
    • Render routine for wiggle animations: Center the images for creating a "wiggle" animation
    • Frames: Number of viewpoint images to create
    • Frame size: Maximum height/width of the output images, height to width ratio is preserved