Jump to content

RTI capture of reflective materials


Recommended Posts

Greetings RTI community!


A colleague of mine is interested in documenting the surface of a 19th c. Japanese magic mirror. In general, such objects are cast in high tin bronze with a design (usually of a Buddha or inscription) on one side and a shiny smooth reflective mirror surface on the other side. When a strong light is shone on them they reflect the cast image from the back side onto a white wall. The overall reflective surface of these mirrors is slightly convex. The predominating theory on how these mirrors work is that there are slight inequalities in curvature, the thicker portions being slightly flatter than the thinner ones, and even sometimes actually concave. So, in theory, RTI should be able to show the image that is reflected because of these slight differences is concavity.


We did two RTI captures the other day of this object, both of which show scratch marks in the mirror’s surface (from use, wear, etc.) but we were unable to see any of the slight inequalities in curvature that we were looking for. So either the surface features we’re looking for don’t exist, or the capture process is problematic. The surface is extremely shiny (it is, after all, a mirror) and we're wondering if anyone on this Forum has had success imaging something with similar surface qualities, and if so, is there anything we might do differently in the capture process to improve our data? We tried processing the data both as LRGB and RGB ptms and also as a RTI file with the HSH fitter--none of these gave us what we were looking for, and we still feel like we have incomplete data.




Anna Serotta

Contract Conservator

Sherman Fairchild Center for Objects Conservation

The Metropolitan Museum of Art

Link to comment
Share on other sites

Hi Anna,


We did some sterling silver spoons to identify the hallmarks last year that were not quite as shiny as yours, but still quite lustrous. The HSH was rather impressive when we reprocessed it. It seems to me that the technique Kathryn Piquette and Lindsay MacDonald used to extract a surface profile from the RTI may be the sort of thing you're looking for to identify the slight curvature. It's about mid-way down on her blog: http://kathrynpiquette.blogspot.ca/


Hopefully Kathryn will see this post and enlighten us on how she and Lindsay did it. It's amazing work! Dynamic relighting in the RTIViewer is great for revealing high-frequency textures like fine scratches, but isn't always the best for very low frequency ones (the slight convexity extending across the entire image). Traditionally a 3D scan would be how this sort of feature would be studied, but Algorithmic Rendering of RTIs appears to be the way to go. I'm sure Carla and Marlin will have a lot to add about that.



Link to comment
Share on other sites

We have shot daguerreotypes successfully, and so have Keats and Mel at Smithsonian Museum Conservation Institute. We were mostly looking for conservation information such as scratches and wear patterns, and the chemicals on the plate.


It would be good to know something about your data set. Do you have a lot of large blow out areas (i.e. all white from specular reflections) in the input data? In that case you might try rebuilding with a subset of the images with less blow out. Also, Mel and Keats used a snoot on the light to reduce blowout on one of their dag datasets, but we've had mixed success with the snoot approach. They talk about this approach in a blog post on our blog a couple of years ago:


Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...