RTI – alternative visualization method

Problems with still image visualization

Franco-German Figurines Project FGFP Terracota Jordan Universität Würzburg Thomas Graichen Astrid Nunn Regine Hunziker-Rodewald Universite Strasbourg RTI Reflectance Transformation Imaging

The default rendering mode in the RTIViewer seems to be a combination of surface normal visualization and an interpolation between input images. This is most evident on objects with pronounced topography and lots of self-shadowing across multiple images
(see left, screenshot from RTIViewer, raking light from right).

The loss of topography information of shadowed areas is an obvious disadvantage when using screenshots of the default rendering mode. A way around this could be to use screenshots with multiple lighting angles, which would bring our image count up to at least 4, and only for a single view of an object
(see below, screenshots from RTIViewer, f.l.t.r. raking light from right, left, top, bottom).

Franco-German Figurines Project FGFP Terracota Jordan Universität Würzburg Thomas Graichen Astrid Nunn Regine Hunziker-Rodewald Universite Strasbourg RTI Reflectance Transformation Imaging

The solution: composite imagery

This can be remedied by using the maximum lighting value for each pixel across all images in combination with the normal map visualization from CHI's RTIViewer. With some photoshop magic one can produce a composite image that incorporate both surface topography and color information (see below, f.l.t.r. maximum pixel values, normal map, final composit image).

Franco-German Figurines Project FGFP Terracota Jordan Universität Würzburg Thomas Graichen Astrid Nunn Regine Hunziker-Rodewald Universite Strasbourg RTI Reflectance Transformation Imaging

Here some further examples (not to scale) of my shading method (left) compared to the resp. web viewer visualization (right, use bulb icon and mouse to change lighting directions):

Franco-German Figurines Project FGFP Terracota Jordan Universität Würzburg Thomas Graichen Astrid Nunn Regine Hunziker-Rodewald Universite Strasbourg RTI Reflectance Transformation Imaging

Amman, The Jordan Museum, JMA 1675 (J 13022).

Franco-German Figurines Project FGFP Terracota Jordan Universität Würzburg Thomas Graichen Astrid Nunn Regine Hunziker-Rodewald Universite Strasbourg RTI Reflectance Transformation Imaging

Madaba, Archaeological Museum, M 4949.

Franco-German Figurines Project FGFP Terracota Jordan Universität Würzburg Thomas Graichen Astrid Nunn Regine Hunziker-Rodewald Universite Strasbourg RTI Reflectance Transformation Imaging

Madaba, Archaeological Museum, M 9928.

Franco-German Figurines Project FGFP Terracota Jordan Universität Würzburg Thomas Graichen Astrid Nunn Regine Hunziker-Rodewald Universite Strasbourg RTI Reflectance Transformation Imaging

Madaba, Archaeological Museum, M 9989.

To achieve a composite image with GIMP just follow these steps:

  • generate an RTI-file as usual
  • open the RTI-file with the RTIViewer
  • save a full view screenshot of the normal map filter as "normalmap.jpg"
  • import the cropped assembly pictures as layers in GIMP
  • set the mode of all layers to "lighten"
  • save as "albedo.jpg", close project
  • open albedo.jpg
  • import "normalmap.jpg" as layer and strech to bounds of albedo.jpg
  • set mode of albedo layer to "Multiply"
  • select normalmap layer
  • colors -> desaturate (luminousity)
  • colors -> curves -> set min. to 0|0 and max. to 205|255
  • select "albedo"-layer
  • colors -> brightness/contrast to 40/30
  • save result

Python script comming soon!