The Voynich Ninja

In You are not allowed to view links. Register or Login to view. by Stephen Bax he discusses a set of multispectral scans performed by Lazarus Project Imaging.

Are these scans available?

As far as I know, the full images are not available outside the project in which they were generated.


The few Jpegs that have been handed out are extracts of the multispectral scans, highlighting some particular frequency bands or combinations thereof.

I asked Stephen Bax about progress on the multispectral images back in October, this was our exchange:
Me:
October 19, 2015 – 7:54 pm
So, it’s been a few months since this post and I was wondering if there’s anything new going on with this? I followed the link to the Lazarus Project site but they have nothing up about the VMS… what’s going on?

Stephen Bax:
October 20, 2015 – 12:06 am
Yes, it is true that they didn’t post anything about their work on the Voynich as it was really a side issue for them! As for something new, we are taking small, small steps every time. Give us another ten years!

Although I am graeful that he took the time to reply to my question, I'll admit that was completely bewildered by his answer.
10 years to analyze multispectral images??? Really?

As an aside, the lossless tiffs posted by the Beinecke (not the jpgs! They are lossy images) can be processed into different colourspaces which allows faint or obscured details to become more prominent.

I suspect that the 10 years do not specifically refer to the digital image processing.

However, the things that could be done with them are far-ranging. I don't remember where I wrote this, but they could apply a (digital) filter that highlights the presence of iron, to better bring out the faded alphabet in the right margin of f1r, but I was told it would be a big effort, that should be done letter by letter.
I don't know enough about this topic to judge it any better.

It would be interesting to track down any similar multispectral scans for other manuscripts to get an idea of how it might eventually be used on the VMS.

I imagine the data format is analogous to a typical uncompressed image with a larger number of channels. Any idea how many?

I would still be interested in looking at any "multispectral JPEGs" that might be available.

Put simply, multispectral images are various different photos taken in different wavelengths that are then combined into image files.
This allows different aspects to be contrasted. For example, in space photos, it allows perception to be judged.
The technique has been used on ancient and delicate document (manuscripts, papyrus, even damaged manuscripts IIRC) to produce a contrast between letters and the substance.
As Rene says, if you take a shot in wavelengths that detects iron, you could then contrast ink types.

An important factor to consider is what wavelengths are being used to develop the original multispectral images. The British Library has a system they use You are not allowed to view links. Register or Login to view. where they explain how they use these types of multispectral images to process manuscripts.

Also worth pointing out that jpgs are useless for this sort of thing, a particular bugbear of mine. Why? Because JPG is a "lossy" format that discards original information from the file to save space. The JPG algorithm then guesses what that information was when it recreates the image. So if pixels 1-9 are all a particular type of red, it deletes pixels 2-8 and later tries to guess what sort of red they were.

That's why the Beinecke provide two types of images to download - the smaller JPG and the much larger "lossless" TIFFS, which keep all the pixel information.

(28-04-2016, 12:26 PM)davidjackson Wrote: You are not allowed to view links. Register or Login to view.Put simply, multispectral images are various different photos taken in different wavelengths that are then combined into image files.
This allows different aspects to be contrasted. For example, in space photos, it allows perception to be judged.
The technique has been used on ancient and delicate document (manuscripts, papyrus, even damaged manuscripts IIRC) to produce a contrast between letters and the substance.
As Rene says, if you take a shot in wavelengths that detects iron, you could then contrast ink types.

An important factor to consider is what wavelengths are being used to develop the original multispectral images. The British Library has a system they use You are not allowed to view links. Register or Login to view. where they explain how they use these types of multispectral images to process manuscripts.

Also worth pointing out that jpgs are useless for this sort of thing, a particular bugbear of mine. Why? Because JPG is a "lossy" format that discards original information from the file to save space. The JPG algorithm then guesses what that information was when it recreates the image. So if pixels 1-9 are all a particular type of red, it deletes pixels 2-8 and later tries to guess what sort of red they were.

That's why the Beinecke provide two types of images to download - the smaller JPG and the much larger "lossless" TIFFS, which keep all the pixel information.

TIFF is a good format, widely used in desktop publishing and the printing industry—I prefer it—but JPG isn't always lossy. It depends how it is stored. There are settings for JPG that determine the amount of compression (and loss) and it can be saved as lossless.

Having said that, not every implementation of JPEG has a lossless option. For example, some of the Adobe implementations (which unfortunately affects a lot of software) are lossy and since Adobe is always changing things, it's hard to keep track of which ones they are. I'm not sure why they decided to do this, maybe because they offer so many file formats and assumed users would choose something else (like TIFF) when they wanted lossless.

Quote:TIFF is a good format, widely used in desktop publishing and the printing industry—I prefer it—but JPG isn't always lossy. It depends how it is stored. There are settings for JPG that determine the amount of compression (and loss) and it can be saved as lossless.

True, true. But when was the last time you saw one of those in the wild, as it were?
BTW, the Beinecke jpgs are lossy, IIRC.

I'd really like to see some more of these, especially from the zodiac images - the goats, bulls and scorpio.

The goat folio Bax has up on his site shows some fluorescent green paint - can anything be deduced from that? It's exactly in those areas where I expect later overpainting...