+- The Voynich Ninja (https://www.voynich.ninja)
+-- Forum: Voynich Research (https://www.voynich.ninja/forum-27.html)
+--- Forum: Imagery (https://www.voynich.ninja/forum-43.html)
+--- Thread: How good is your pareidolia? (/thread-4839.html)
RE: How good is your pareidolia? - Jorge_Stolfi - 10-08-2025
(09-08-2025, 11:55 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.The following image was created during the forensic analyses in 2009
Too bad we don't have the whole VMS scanned at that resolution...
The multi-spectral images have somewhat better resolution than the Beinecke 2014 scans, but unfortunately the pages that they scanned are "boring" in that the original writing was rather well preserved, and hence there does not seem to have been much opportunity for the Retracer to make "interesting" mistakes.
The exception is f1r, which has a lot of damage from many processes, even after the retracing. That oCta example cited above, from the other thread, is from that page, and is an apparent case where the Retracer replaced the orignal glyph.
All the best, --jorge
RE: How good is your pareidolia? - oshfdk - 10-08-2025
(09-08-2025, 11:57 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.When writing a Ch, for example, he would write an e and then an h with the tip of the ligature precisely touching the tip of the e. Ditto when writing s, t, etc. And the original figure outlines are generally ~0.2 mm wide, independently of direction.
It's interesting that you chose ch as the example, because it seems to prove my point. There is a whole solution theory on this forum that works by treating various ways of misaligning ch as different characters, because ch's are often misaligned and in many possible ways (above/below/overlap or gap).
Also, when connecting lines with a quill, as long as the ink is still wet, you will get natural smoothing of the outline at the point where two ink blobs touch one another. I don't think it takes more than normal human ~0.25mm precision to write a proper ch, but the scribe managed to fail even this in a large number of cases.
Looking at the retouch jobs the smoothing doesn't seem to happen when the old ink was already dry.
RE: How good is your pareidolia? - Hider - 10-08-2025
(09-08-2025, 11:55 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.The following image was created during the forensic analyses in 2009, and confirms these dimensions.
Very interesting photos. Not a modest question. Do you have any other similar images. Preferably with a link to the folio. Thank you.
RE: How good is your pareidolia? - oshfdk - 10-08-2025
(10-08-2025, 07:58 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.It's interesting that you chose ch as the example, because it seems to prove my point. There is a whole solution theory on this forum that works by treating various ways of misaligning ch as different characters, because ch's are often misaligned and in many possible ways (above/below/overlap or gap).
Found it: You are not allowed to view links. Register or Login to view.
And also copying from the same thread a collection from ch's from two folios. As you can see, c and h are not perfectly connected with impossible 0.1mm precision, they just smoothed out into a single curve by the action of wet ink. When the offset is really large, there is a visible step, like in the upper right corner. Wet ink won't allow sharp angles and abrupt steps. Retouching over dry ink most probably can leave sharp angles and visible steps or double lines, as seen in other manuscripts and in the example from You are not allowed to view links. Register or Login to view. I posted earlier.
RE: How good is your pareidolia? - Jorge_Stolfi - 11-08-2025
But those ligatures are only ~0.2 mm wide. So it seems that the scribe did in fact manage to connect the first e and the h with ~0.1 mm of accuracy, most of the time.
My retracing hallucination requires the existence of a scribe who was routinely able to follow previous strokes with at least ~0.1 mm accuracy. Whether they were his own strokes or someone else's is a separate question. Unfortunately for me, this pre-condition is obviously hard to prove. For that I would have to find at least one example where a stroke was clearly retraced with that accuracy. But then the previous stroke would be invisible...
There are thousands of cases where a dark half-stroke precisely joins with a lighter half. They could be examples of retracing (and I think that many are); but you would claim that they are just cases where the ink flow suddenly changed halfway through a stroke...
By the way, another "natural" cause of stroke weight variation, that I forgot to mention in the previous post, is that the vellum surface, having been "sized" with chalk or other minerals, is rather abrasive and quickly wears down any sharp tip that the pen may have initially, making the strokes gradually wider For this reason, the pen had to be re-sharpened periodically. There is a nice example of this phenomenon on page You are not allowed to view links. Register or Login to view. of the Starred Parags section. The strokes suddenly become much thinner halfway through line 15, and then gradually return to their 'normal" width over the course of a paragraph or two. (I noticed this also in my own casual experiments with a bamboo pen on plain office paper, even though the two materials should have the same hardness.)
Here is an example of that phenomenon from Biological page f75r:
These are the first five lines of the page, minus the intruding figure and a couple of words at the right margin. Note that the first word has very thin strokes (0.2-0.3 mm), even in the "broadstrokes" where the pen is moving Southeast. Immediately to the right of the figure many broadstrokes already have the "normal" width (~0.5 mm), but they are all dark, suggesting that they were retraced; the original stroke width may be glimpsed in the lighter parts, such as the q glyphs and some plumes and loops. Then the stroke width -- even in the lighter parts -- gradually increases in the next line or two, as the pen tip wears off.
(For this topic, it does not matter here whether the thicker strokes on the right half of line 1 are "back-tracing" by the original scribe or were made in a later restoration attempt. If it was by the original scribe, perhaps the reason for him going back was in fact that those strokes were thinner and did not look good next to the ones just below them.)
Sudden but short-lived thinning of the traces is often seen at the start of a page (like here) or of a parag. And indeed it seems unlikely that the Scribe would stop in the middle of a parag to re-sharpen the pen. The example of f104v seems exceptional in this regard...
By the way, also on f75r, there is a block of text nested inside the main figure that has anomalously thin strokes and somewhat smaller "font size" (minim height, o-height; only ~1 mm instead of ~1.3 mm):
Yet this text shows the same irregular mix of light and dark strokes as the rest of the text on this page. I would say that it was retraced too, but surely you will say that it is normal ink-flow variation. Would you accept the o in the daro as evidence for my claim?
RE: How good is your pareidolia? - oshfdk - 11-08-2025
(11-08-2025, 12:36 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.But those ligatures are only ~0.2 mm wide. So it seems that the scribe did in fact manage to connect the first e and the h with ~0.1 mm of accuracy, most of the time.
My limited experience with the quill suggests to me that the accuracy you need in this case is the same as the width of the stroke or even a bit less, so about 0.25mm, which is very precise, but manageable. This works because of the tendency of liquids to merge in a smooth roundish shape, this has something to do with the physics, but I won't pretend I fully understand the process (I can look it up, I'm not particularly interested now). After you write e, if the ink is still wet, you can touch with the quill anywhere close enough to the tip of the stroke (you don't even need to hit it precisely, so roughly 0.25x0.25mm area is ok) and move the quill tip and two strokes will merge into a single smooth shape. This will produce a somewhat different angle depending on where exactly you started the next stroke, but if you look at the collection of ch in my previous post, you can see that the connection between e and h has many variations, this is likely because the scribe hit different parts of e when starting h stroke.
(11-08-2025, 12:36 AM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.Yet this text shows the same irregular mix of light and dark strokes as the rest of the text on this page. I would say that it was retraced too, but surely you will say that it is normal ink-flow variation. Would you accept the o in the daro as evidence for my claim?
I would accept this o as the only piece of a block of very thin letters (0.15 mm strokes probably) where there is something that could be interpreted as retracing. Which to me speaks a lot about the whole retracing hypothesis.
It is physically possible that someone retraced the text over hundreds of pages with the meticulousness and precision of a money forger. But why?
RE: How good is your pareidolia? - Jorge_Stolfi - 11-08-2025
(11-08-2025, 10:01 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.My limited experience with the quill suggests to me that the accuracy you need in this case is the same as the width of the stroke or even a bit less, so about 0.25mm, which is very precise, but manageable.
I had quite a bit of experience with pen-and-ink drawing in my teen years. I bet I could do a bit better than 0.1 mm accuracy. But I used a steel pen with very sharp nib, and I was severely shortsighted back then, so I could look at the work from only a couple inches away.
I should have some samples somewhere. Meanwhile, found my old pen and a few bottles that still had usable ink, in spite of their venerable age. (The trick is to keep them upside-down.)
And I so I tried to see whether I could be hired by the Author to write Volume II, if Selene decided to wake him up ahead of schedule.
The size of characters is about the same as in "normal" VMS text (o-height ~1.5 mm) , and the resolution is about the same as in the Beinecke 2014 scans (~15 pixels/mm). The pen strokes are ~0.4-0.6 mm wide.
Granted, the result was, ahem, a bit disappointing. But note that it was the very first time I used that pen in at least 40 years, that the ink has visibly degraded -- it is more watery than it should -- and the paper is not ideal for it. (That black blot near top right was a pen loaded too much ink.) And of course I am probably a bit older than the VMS Scribe was. And, after two cataract operations, I have quite sharp vision but cannot focus any closer than 30-40 cm away.
And I also had to keep the pen upright, otherwise it would catch and dig into the paper on the NW strokes. So I could not make any "mousetail" plumes and tails.
As you can see, I tried to retrace some of the red text with blue ink, but with very limited success.
So this experiment unfortunately does not quite support my claims. But it does not refute them either, right. Maybe someone with better eyes can do a better simulation of the original Scribe and of the Retracer.
By the way: while the original Scribe probably used a quill pen, the hypothetical retracing could have happened any time between 1500 and 1900. So the Retracer could have used a steel pen, that AFAIK became fairly common in the early 1700s.
Quote:It is physically possible that someone retraced the text over hundreds of pages with the meticulousness and precision of a money forger. But why?
Because the original writing had severely faded, to near invisibility. Look at f73r, for example. Look for the faintest strokes; that whole page must have looked like that or worse when the owner decided to salvage it. Think of the retracing as the restoration of an artwork that must have been highly valued to the owner at the time. He naturally wanted the result to be close as possible to the original state. Think of how much money and time people and institutions are nowadays willing to spend to restore a painting, a statue, a mural...
All the best, --jorge
RE: How good is your pareidolia? - R. Sale - 11-08-2025
On what evidence is the period from 1450 to 1500 eliminated?
Or could it have been retouched by Voynich himself?
RE: How good is your pareidolia? - oshfdk - 11-08-2025
(11-08-2025, 08:07 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.Granted, the result was, ahem, a bit disappointing. But note that it was the very first time I used that pen in at least 40 years, that the ink has visibly degraded -- it is more watery than it should -- and the paper is not ideal for it. (That black blot near top right was a pen loaded too much ink.) And of course I am probably a bit older than the VMS Scribe was. And, after two cataract operations, I have quite sharp vision but cannot focus any closer than 30-40 cm away.
Sorry for stating the obvious, but a magnifying glass, especially the wearable watchmaker kind, can certainly put the visual acuity for most people far above of what was possible for anyone in the 1400s, I don't think artificial lenses would be a problem here either. I used a magnifying glass for my experiment with the pencil above, just to make sure the vision part of the experiment would not become the weak link.
(11-08-2025, 08:07 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.Because the original writing had severely faded, to near invisibility. Look at f73r, for example. Look for the faintest strokes; that whole page must have looked like that or worse when the owner decided to salvage it. Think of the retracing as the restoration of an artwork that must have been highly valued to the owner at the time. He naturally wanted the result to be close as possible to the original state. Think of how much money and time people and institutions are nowadays willing to spend to restore a painting, a statue, a mural...
My question was not why attempt retouching, but why perform it with what I see as nearly superhuman accuracy. After all, a text is a text. Unless the retoucher was somehow sure that a microstroke/microglyph encoding was used, I see no reason to try restoring every minute detail of every stroke. I would expect something more similar to what your attempt shows, than to what we see in the MS.
One possibility that could explain the perfect retouching in the VMS is if due to some properties of the ink and the vellum it would be enough just to add a drop of ink to any part of a dry stroke and it would fill the same shape, with the wet ink sticking to dry ink or oxidized vellum. However, the imperfect retouch job on You are not allowed to view links. Register or Login to view. seems to disprove this, with faint stroke outlines clearly visible next to the dark strokes (at least for Sh).
RE: How good is your pareidolia? - Jorge_Stolfi - 12-08-2025
(11-08-2025, 09:19 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.Sorry for stating the obvious, but a magnifying glass, especially the wearable watchmaker kind, can certainly put the visual acuity for most people far above of what was possible for anyone in the 1400s
Yes, I should try that. I am even thinking of getting glasses made that will restore my old nearsightness...
Being able to focus on things 10 cm away instead of 40 cm is like using a 4x magnifying lens. Surely there were plenty of myopic people in the 1400s too...
Quote:My question was not why attempt retouching, but why perform it with what I see as nearly superhuman accuracy. After all, a text is a text. Unless the retoucher was somehow sure that a microstroke/microglyph encoding was used, I see no reason to try restoring every minute detail of every stroke.
Suppose that the owner of a badly faded, 200-year-old manuscript were to try to restore it -- today. A manuscript that may have been bought by an emperor for 600 gold coins. Would he be satisfied with crudely re-writing the text over the faded original, leaving the later visible? Especially considering that neither he nor the person he hired for the job knew what was the alphabet, hence could not tell whether a tall plume was equivalent to a squat one?
And moreover, if all he wanted was to preserve the text, he would just make a copy on new blank vellum.
Again, the retracing that I see is like the restoration of a valued painting. The owner does not want merely a portrait of a chubby smiling lady; he want the portrait to look like it must have looked right after Leonardo painted it -- with no obvious sign that it was painted over.
On the other hand, I believe that there were at least two restoration episodes, and the second one was more limited but much sloppier than the first one. It did not even try to approximate the color of the original ink.
Quote:One possibility that could explain the perfect retouching in the VMS is if due to some properties of the ink and the vellum it would be enough just to add a drop of ink to any part of a dry stroke and it would fill the same shape, with the wet ink sticking to dry ink or oxidized vellum. However, the imperfect retouch job on You are not allowed to view links. Register or Login to view. seems to disprove this, with faint stroke outlines clearly visible next to the dark strokes (at least for Sh).
That could be possible, and I even think I can see some evidence of it, e.g.in this clip from f8r, line 4, just after the plant:
Looking closely, it seems that, when C of the Ch was retraced, some of the ink flowed into the loop of the k, for ~0,5 mm or so.
However, this phenomenon does not explain the cases where the Retracer added strokes that were not there, like the item B in my original post (where he extended the plume at the wrong end, turning a plain Sh into an "impossible" weirdo) or the honors item (where he completed the body of an s into an o).
And, on that same clip, note how the apparently retraced ligature stroke is unusually broad and has a butt (square) right end, tilted 45 degrees -- just like a normal broad pen stroke.
Also it seems unlikely that ink could flow that way over the whole leg of a t.
All the best, --jorge