f116v is a decryption attempt?

[Jorge_Stolfi]

I was reading Nick Pelling's blog (You are not allowed to view links. Register or Login to view.) and came across his 2022 suggestion of raking IR on f116v, and I wondered whether the AI could get anywhere with the 2014 multispectral set (Dr. Davis wrote up here: You are not allowed to view links. Register or Login to view., and which is freely downloadable) to see if pen furrows would come out.

So the method (get the AI to take multiple images of the same surface and subtract stuff to produce the topographic image, like a shading gradient) works... it doesnt simulate raking (which I thought it might, anyway I thought this was interesting to see how far AI stuff has come with image processing) but there's just no furrow there to find at this lighting.

Great images, thanks!

For each dominant wavelength, the Lazarus multispectral imaging project team took three images of the page, using two light sources located at some angle (very roughly 45 degrees) from the camera axis: one tilted towards North (top edge of the page), one towards South (bottom edge).  The third image (usually first in their numbering) was taken with both sources. 

This third image should be redundant, since the physics of light scattering and the camera transfer function are supposed to be linear.  That is, the image with both lights should be just the sum of the other two.    It would have been more useful to use three sources at three independent directions (say, N, SE, SW).  Then one could easily determine the local surface inclination and the vellum color at every pixel.  This is the so-called "photometric stereo" method, first devised in the 1980s (but used informally much earlier to get relief maps of the Moon surface).  

One can do photometric stereo even with only two light directions (as shown in JoeyB images C and D) but the theory says that the results will not be very reliable.

Shallower rake on You are not allowed to view links. Register or Login to view. could show if there were any furrows at all.

In theory, the relief map produced by photometric stereo does not depend on the inclination of the light sources. If there is a bump or dimple 1 mm tall, it will show as 1 mm tall in the resulting relief map, no matter where the sources are located.

Raking light would let one see the bump without any processing, but will not reveal bumps that could not be detected  with other lighting. In fact, for photometric stereo it is important that every point of the surface be illuminated by enough sources, since the math does not work on parts of the surface that are in shadow.

The angles between the lights would affect mainly the the "noise" in the map. The ~90 degree separation used in the Lazarus project images would in fact be very good in that regard.  If only they had used three sources instead of just two...

which Nick indicated in his blog could help figure out if the marginalia is one hand or two or three from different periods and if a later hand bit differently into the surface as it stiffened over time

That might work for ballpoint writing on paper.  But velum or parchment is stiff tough leather, like a drum membrane.  I don't think that the sharpened and split tine of a quill pen would be able to scratch or dent it at all.  And even the darker ink traces on the VMS are very thin layers, as one can see in the McCrone macrograph below (172 pixels/mm, ~4300 pixels/inch).  Too thin to be detectable by photometric stereo (or raking light).
   
(Beware that the right ~2/3 of this image is out of focus.) 

Instead of raking light, it would be much more useful to have images with at least that much resolution of well-chosen spots on You are not allowed to view links. Register or Login to view. and on other chosen pages.  Those images could be captured even with a toy hand-held microscope, without any special lighting or mounting, without even touching the book.  But I suppose that the Beinecke managment is so fed up with Voynich weirdos that they would not allow even that.

A pity, because such images would definitely prove or refute the various "retracing" and "magic ink" theories, including the "Big Wipe-off" theory for f116v,  and reveal practically everything one could ever know about the original traces.

All the best, --stolfi