Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
×
EU Graphics

Museum Announces Highest Resolution Ever for an Image of a Painting (thehill.com) 68

Rembrandt's masterpiece The Night Watch "can now be viewed on computers everywhere in ultra high definition," writes The Hill.

But ultra high definition is an understatement, according to long-time Slashdot reader dr_blurb. "Some tech specs: 717 Gigapixels in a 5.6 terabyte image: 8,439 individual images, taken with a 100 megapixel Hasselblad H6D 400 MS camera."

This single image is over four meters in length and three meters in height, reports Digital Photography Review: The museum also points out that the distance between 2 pixels on the image is 5 micrometers (0.0005 centimeters). This means that 1 pixel on the image is smaller than a human red blood cell. According to representatives from the museum, each photo has a depth of field of 125 micrometers (0.0125 centimeters). To ensure each image was properly in focus, the surface of the painting was scanned with lasers. Then the camera's settings were adjusted for optimal image quality. After each image was captured, a neural network scanned it for color accuracy and sharpness.

The level of detail captured, coupled with the size of the file, makes it the largest image of a work of art ever captured. It's 4 times larger than the original digitized version of "The Night Watch" that was published on the museum's website in May 2020, and that file was already 44.8 gigapixels...

It can be viewed on the Rijksmuseum museum's website.

"There were many people who thought it was impossible, and who thought the Operation Night Watch team were crazy to even attempt it," said Robert Erdmann, senior scientist at the Rijksmuseum. "We have surpassed ourselves in what can justifiably be described as a world-class achievement..."
This discussion has been archived. No new comments can be posted.

Museum Announces Highest Resolution Ever for an Image of a Painting

Comments Filter:
  • by backslashdot ( 95548 ) on Sunday January 09, 2022 @04:39PM (#62157789)

    Did we need that much resolution detail versus say capturing the texture in stereo 3D and getting the color (multispectral?) accuracy correct?

    • by cats-paw ( 34890 )

      i'd like to know too.

      i would expect that if they really are getting that resolution there's probably interesting features that can be studied in terms of the painting brushes/implements used, the application and drying of the paint, the condition of the paint, etc...

      you'd think that sufficient information would be available at 10um per pixel or even 20um.

      It's still an amazing achievement, and perhaps the point of it was just to perfect this technique for other applications/fields.

      • by ls671 ( 1122017 )

        For a tecky site, I would have been interested in knowing which file formats support 5 TB images.

        • From TFS:

          8,439 individual images

          .
          It's not a single file.

          • by ls671 ( 1122017 )

            They say they took 8,439 individual images with the camera to get maximum resolution. I assumed that they then assembled all those images into a big complete image file afterward that would be 5.6 TB.

            • I'm guessing that would be unnecessary and would make everything more difficult. It's better to keep the images in a matrix and stream the relevant parts with some on-the-fly stitching wherever they overlap/intersect.

        • by bobby ( 109046 )

          Didn't read the article yet (just following the rules! :) but BigTIFF [wikipedia.org] might be a possibility. I'm not sure who / what supports it, but maybe some of these: http://simplesystems.org/libtiff/bigtiffpr.html

        • For a tecky site, I would have been interested in knowing which file formats support 5 TB images.

          My clients have been sending me 5 TB TIFF files for years - that's not bad going considering they only have a cheap scanner and a single sheet of paper to scan.

    • Re: (Score:2, Insightful)

      by Anonymous Coward

      Wow, beat me to it, and on the first post. I didn't expect an appreciation for this to come up so quickly. I don't know how critical this is for Rembrandt's work, but I was absolutely stunned the first time I saw a Van Gogh in person. It's entirely different from the flat images you see in books. A picture of a Van Gogh is nice. Seeing a Ven Gogh in person leaves you with the impression that he was living in the 1960s psychedelic world before anybody was using those terms. IIRC, it was cyrpus trees.

      • Typical. When someone like John Myatt [wikipedia.org] creates a perfectly detailed high-resolution copy of a Rembrandt or whatever he gets arrested, but when some museum does it it's fine.
        • The Rijksmuseum are the custodians of the artwork (as proxy for the Dutch people, or something similar), though what the copyright status if the art work is not something I know.

          Most museums that I've looked have a notice in the entrance (sometimes in the Ts&Cs on the tickets) that photography on the site is banned, or photography of the exhibits is banned (contra, photography of your group, shoes, whatever else), and so any photographs taken of the exhibits were evidently illegally obtained. So their

    • Looking at the image and zooming in, I would say that actually there could be some benefit in getting even higher resolution than this. 3D texture would be great as well, of course.

    • We're doing 0.7 terapixels!!!
  • by JaredOfEuropa ( 526365 ) on Sunday January 09, 2022 @04:43PM (#62157799) Journal
    Next: the NFT for this thing goes on auction.
    • Next: the NFT for this thing goes on auction.

      But it won't fetch nearly as much as the Photoshopped version which swaps out the people for bored apes.

    • Next: the NFT for this thing goes on auction.

      And the purchaser only has 5 minutes to download it after the payment clears, then it's deleted.

  • have to ask why? is this just to see if they can do it? can't see much use in capturing at a resolution way way way beyond what even the artist could have perceived while creating the art work.
    • Re:why? (Score:5, Interesting)

      by AleRunner ( 4556245 ) on Sunday January 09, 2022 @05:18PM (#62157889)

      have to ask why? is this just to see if they can do it? can't see much use in capturing at a resolution way way way beyond what even the artist could have perceived while creating the art work.

      One answer is because then you can see the detail of how it was constructed. Zoom in on the child's right eye and you can see the direction of the individual strokes and how one of the later ones crosses over an earlier one giving an impression of puffiness when you zoom out. It's interesting and I guess pretty useful if you want to copy the style. No doubt if any of us had any level of art education they would be able to tell you about how this shows something specific about the artist and how they learned from some other earlier artists. If you are looking at another painting claimed to be by the same artist and you saw the same effect done in a completely different way that might be a good hint that the painting is done by a different artist or even a forgery.

      • That's an argument for capturing imagery down to a resolution several times better than the tools the artist used. You could argue for months whether the tool in question was a paintbrush (say a half-mm across, controlled deliberately by the artist), or a bristle in a paintbrush (say, 0.05 mm across, controlled implicitly but not explicitly by the artist).

        They've gone a factor of 10 beyond that argument. Which .. isn't stupid. The argument can be had, and at the end of it, the bloodied survivor from the Ar

    • have to ask why?

      Clearly you've never played Animal Crossing. This is so they can make really damn sure that sly kitsune didn't pull a fast one on them.

    • I suspect part of it is insurance against loss of work.

      Art theft, fires, floods, dipshit work experience intern trips over and destroys the artwork with a fork. all these things still happen in 2021. Not to mention the general physics of it, paints discolor , get oxidised and so on.

      If the artwork goes away, at least we've got an extremely high res capture of the work for future art historians , and general public, to study and look at.

      But yeah the suggestion earlier in these comments about trying to get a 3

      • Art theft, fires, floods, dipshit work experience intern trips over and destroys the artwork with a fork. all these things still happen in 2021.

        Most of these you can manage away, or engineer away. Crappy little museums with no budget might let an enthusiastic amateur at a work with a brush, but I doubt that applies to a Nation State grade museum who can call on the Army to post outward pointing machine gun nests at each corner with shoot-to-kill orders for fork-bearing interns.

        Many of these major artworks

  • The level of detail captured, coupled with the size of the file, makes it the largest image of a work of art ever captured.

    Those are two independent measurements.

    You can argue that a photograph taken through a microscope which has finer details is on some sense "larger" at least with respect to "how close did we zoom in."

    On the other hand, I wouldn't be surprised if there are some other cases of "nearly 10,000 images, or more" being stitched together to create a composite image that is "larger" in pixel count or size-in-TB even if the pixels weren't as high-resolution.

    Now, if they are claiming to be the most detailed image in b

    • On the other hand, I wouldn't be surprised if there are some other cases of "nearly 10,000 images, or more" being stitched together

      Artists have been doing that sort of thing and calling it "collage" since ... the 1930s?

      Digital artists have been doing that since the 1980s, limited only by the technology of the day.

      • A collage typically is an artwork in and of itself.

        We are talking about using technology to make many photos look as if they were a single photograph.

        If they created a composite photo with the smaller photos arranged in any order other than the "natural" order, that would be an artistic collage.

  • by DDumitru ( 692803 ) <`moc.ocysae' `ta' `guod'> on Sunday January 09, 2022 @05:04PM (#62157845) Homepage
    Their web viewer is impressive. A simple zoom in with the mouse wheel all the way to the cracks in the paint. No lag. A very good job in terms of keeping the UI simple and responsive. Bravo.
    • Also impressive, and hasn't been mentioned in the whole thread: The picture itself!
      • Also impressive, and hasn't been mentioned in the whole thread: The picture itself!

        An enthusiastic Yes! to both!

        The page scrolls and zooms amazingly well. It's beautiful bit of coding.

        The painting itself is one of my favorites. I can't get enough of the Dutch masters. I zoomed into one of the lace collars to see how it was painted. When you get close enough to see the brushstrokes, you can't tell it's lace any more. I'm completely in awe a human made this.

        The great thing is, if you see a Rembrandt in a museum, you can't get close enough to see this level of detail. I can admire the brushw

    • Shift-click zooms out again

    • by lorinc ( 2470890 )

      Rob has been developing this sort of tech for years. He made several other web based viewers that are also super nice when looking for details in such images. I'll chime in if I can recall the links.

      • Isn't that just sparse virtual texturing? Not to downplay it, and surely done well without much visible popping
    • I don't know.... I feel the constant urge to take a clone and a blur tool to it and repair all the cracks... I know I really shouldn't, but here we are... Maybe a little bit? Just the really large ones, I promise.

  • Atoms say otherwise.
  • by SuperKendall ( 25149 ) on Sunday January 09, 2022 @05:22PM (#62157897)

    I've been to Amsterdam and seen the original, it's a really huge painting (roughly 12 x 14 feet) so it certainly benefits from such a large capture image.

    I'm surprised they were able to do the whole thing so well with that shallow a depth of field; I couldn't find out if they measured the distance via laser per image, but it seems like you would have to... and with that kind of depth of field you would have to have the camera itself absolutely parallel to the image. Even variation is paint height seems like it would make a difference. The web viewer doesn't seem to let you zoom in var enough to tell if there are any depth issues like that.

    • For soem reason the original link I followed seemed to lead me to a medium resolution version, I finality found the actual high res version [rijksmuseum.nl] and was able to zoom quite far in...

      On closer inspection there are some minor depth of field issues where some areas in the image appear sharper than others, pretty easy to tell when zoomed in the full amount you can see cracks which should all be uniformly sharp.

    • I've been to Amsterdam and seen the original, it's a really huge painting (roughly 12 x 14 feet) so it certainly benefits from such a large capture image.

      It's also just been restored over the last 2 years so it probably looks even more spectacular than you remember.

      I couldn't find out if they measured the distance via laser per image, but it seems like you would have to... and with that kind of depth of field you would have to have the camera itself absolutely parallel to the image.

      No you actually don't. Even with macro lenses beyond 1:1 the ability to capture a large depth of field is highly dependent on lighting and exposure. Hard to do for a moving subject or with a hand as anyone into macro photography will tell you, but quite trivial on a fixed tripod, or in this case an XY gantry. High f-stops and long exposures will easily get you more than enough depth of field to co

      • As I recall, chopped off and thrown away a few centuries ago because it would not fit though a door.

      • It's also just been restored over the last 2 years so it probably looks even more spectacular than you remember.

        Thanks, I love visiting Amsterdam and so will make a point the next time I go to se the painting again, I really loved seeing it the first time! It was several years ago now so for sure I've not seen the refreshed version.

        High f-stops and long exposures will easily get you more than enough depth of field to cover the thickness of a paint stroke.

        I do a lot of photography including some panoramic

        • Also I did wonder if the "AI" they used to stitch was just Auto Pano Pro or some custom stitching software.

          Yeah they do love throwing that word AI around a bit. There's very few programs out there that successfully handle a project of this magnitude, Autopano giga (the big brother of Auto Pano Pro) would be an example.

          Also yeah somehow that summary text didn't register in my brain, that goes to show you shouldn't post on Slashdot at 2am XD. With that being said given the DoF mentioned I now wonder if they just used the motorised inspection microscope to do the pano.
          When paintings are restored (which this has bee

  • Imagine painting the Night Watch and leaving out Fred Colon. Pretty sure that's Nobby Nobbs at the back eating rat onna stick though.
  • How could it possibly be impossible? We have microscopes that allow you to take pictures. Microscope + tracking system to scan over image + software to stitch together the resulting image (e.g. remove overlap) should get you pretty much any resolution you want down to the diffraction limit of light.

    It might be a huge pain in the ass or kinda expensive but how could it be impossible? Even limits in the system for scanning over the image don't seem like much of a problem since you can just do a whole bunch

    • How could it possibly be impossible? We have microscopes that allow you to take pictures.

      Of course it can be done. I've build several digital photography rigs for clients wanting lots of photomicrographs over the years (sometimes with as much as several days between the order and shipping to site), and field curvature means you have to resort to stitching arrays together when the objective turret doesn't give you the appropriate choice of magnifications.

      But I can't remember even attempting to stitch more t

      • by ebvwfbw ( 864834 )

        At that resolution all you'd have to do is blow on it and the digital pix is no longer correct. I'm sure dirt and dust would change it.

        Woops, have to start over.

        • It's not "resolution" that the issue, but "depth of field". (Though the two are related for a particular lens.) And the "start over" is restricted to re-taking the current photo (and possibly making an entry into the observing log.

          You missed the bit in TFS about using some sort of auto-focussing rig at every photo station on an X-Y frame for positioning the camera. Surprisingly, the people who did this had actually done it before, encountered this problem (well, I'd encountered it in my much more limited e

  • But why would you want to bother with this?

    The artist designed their artwork to be viewed by the human eye. Sure, there might be details about their brushstrokes that you can see in super high-resolution but none of that is part of Rembrandt's artistic brilliance. The only reasons for paying any more attention to the Night Watch than a four year old's painting on the researcher's refrigerator are the artistic choices he made about how the painting looks to the human eye.

    How the paint happens to look at th

    • I understand that when it comes to restoring the painting you might want to look at hints about what the paint in that section originally looked like that aren't big enough to see with the naked eye. But if that's your goal why would you be scanning it in just the visible wavelengths? For that goal you first want to do some research about the wavelengths that are most likely to cut through the layers of gunk and distinguish the paints used in the original from any later restorative efforts.

      This feels like

    • In the article they mention it's to aid scientists in determining how the painting ages.

    • Nothing lasts forever. Many paintings have been lost to wars. Detailed digital scans are an effective way of providing recovery from a disaster.
  • by test321 ( 8891681 ) on Sunday January 09, 2022 @06:08PM (#62157995)

    In case anyone is wondering, 5 micrometers pixel size would be 5080 dpi, meaning it's only slightly more than the 4800 dpi of a consumer-grade flatbed scanner. Artists already scan at this resolution https://www.parkablogs.com/pic... [parkablogs.com]

    The feat is to do that on such a large canvas and tile the photographs flawlessly. But by itself the resolution is not unheard of.

  • by war4peace ( 1628283 ) on Sunday January 09, 2022 @06:23PM (#62158029)

    1. You can read the banner/emblem text pretty well.
    2. Zooming to lower/right (at the edge of the image) you will see some stitching inconsistencies.
    3. I want to know what "Macbook 001" means.
    4. Zoom to the bottom and you will see very thin individual canvas thread strings.

  • With a 3d printed impasto.
    Better than the real thing. :-)

  • Merely making high resolution images of fine art paintings is missing tons of information. In fact, the other parts of the painting that really matter in real life is understanding that paintings are three dimensional. Oil on canvas has peaks and ridges, the types of pigments and the angles of the strokes reflect light in different ways.

    Itâ(TM)s nice to have a high resolution flat image, but without the three dimensional and texture info (specular reflectivity etc) then itâ(TM)s never going to be

  • From the tech specs... they used 100 different cameras and I assume over 8000 angles.

    I'm no expert, but I'm running some mental DSP algorithms, and while I strongly believe there is great value in the source data, A single 100 megapixel image is almost guaranteed to be substantially more accurate than the heavily post processed version.

    A few images combined from several well chosen perspectives to produce a bump map and texture map would also be of interest.

    It sounds to me as though a 4 axis CNC CCD motion
    • there is great value in the source data,

      Do your first stage of processing by choosing good glass out the front of the sensor. Help it with good lighting.

      That's in "Photography 1.0.1."

      Having "mental DSP AI chops" in your phone means only that you'll be able to produce much better photos if you couple that back end to a good set of glass out the front.

      When digital photography sensors got above about 5 mega pixel, they were more or less matching what the unaided human eye can perceive. And at that point peo

  • Please do the Bayeux Tapestry next. The sistine chapel would be great too.
  • When covid isolation started, I've opened 5000 puzzle box with The Night Watch by Ravensburger, I have bought in 1989.
    Became discouraged after a few days - resolution was awful and I've lost interest in completing something that blurry and with wrong colors.

  • 640 x 480 should be enough for everyone.

  • Now you can see rembrandt down to the molecular level. this will be useful for google street view when you zoom in.

  • by Babel-17 ( 1087541 ) on Monday January 10, 2022 @02:07PM (#62161503)

    If that was happening in a Pink Panther movie, with Peter Sellers as Inspector Clouseau near the controls, the outcome would be disastrous, and excruciatingly funny.

  • In one of the later books in the Dune series, he depicted, IIRC, one of the Bene Gesserit holding her finger to a plate beneath a painting, and she could could then experience the laying of brush strokes of paint that it took to create it. That presupposed an incredible detail of analysis, one where science blended into magic, as it was so advanced.

  • Though that was way back in the 70s or 80s. I think they stated that their desire was to make available to the public high quality reproductions of art, because people couldn't be expected to get to galleries to see them, nor could the availability of them be counted on. I forget exactly what was the minor controversy that caused, but I think it had to do with maintaining standards, and perhaps with art getting churned out of a factory.

Technology is dominated by those who manage what they do not understand.

Working...