Half a croissant, on a plate, with a sign in front of it saying '50c'
h a l f b a k e r y
Where life imitates science.

idea: add, search, annotate, link, view, overview, recent, by name, random

meta: news, help, about, links, report a problem

account: browse anonymously, or get an account and write.

user:
pass:
register,


                                                     

Reconstruct old Masters by reverse-engineering them

Use established techniques to recreate 3-dimensional scenes from the past
  (+2)
(+2)
  [vote for,
against]

3d design usually involves someone painstakingly constructig a 3- dimensional model consisting of objects, light- sources and all manner of texture or material properties to define how light behaves when bouncing around the scene.

Once built, a viewpoint is defined, and the whole lot sent to the rendering machine where all the calculations are done and the scene is rendered in all its imaginary glory.

What this idea is all about is reversing that process and taking, as a starting point, a representation from say, Caravaggio's "The Calling of St Matthew", and with all manner of technical wizardry, reverse engineer the three-dimensional scene, along with light-sources, apertures and so-on.

The result would be a (computer) model describing the set, along with lighting specifications, which, if realised in the studio, could be populated with members of your family or social circle before taking a photograph.

That resulting photograph *ought* to be a fairly faithful recreation of the original masterpiece, subject to surrealist or non-Euclidean tendencies on the part of the original artist. I imagine sets constructed from works of the more Baroque masters might require a small fortune in expensive lighting, but should yield the most satisfying photographic results.

zen_tom, Jan 15 2019

Arnolfini Portrait https://en.wikipedi.../Arnolfini_Portrait
" ... considered one of the most original and complex paintings in Western art ..." [8th of 7, Jan 15 2019]

Janelle Shane trained an AI to mix drinks http://aiweirdness....ural-net-mix-drinks
[Ian Tindale, Jan 15 2019]

Animated Art https://www.youtube...watch?v=l-7IV2qryiQ
[bigsleep, Jan 19 2019]

VR Van Gogh https://www.youtube...watch?v=jBOL5yakREA
[theircompetitor, Jan 19 2019]

[link]






       Surely this is only true if the old masters were striving for a photorealist look? I suspect that much of the lighting in these pictures is impossible in real life.
hippo, Jan 15 2019
  

       So I don't think it would work for cubist or abstract images (though it might be fun to see what got squirted out in these cases) but if as a human, we respond to graduations of shading and interpret those inputs as a projection onto a 2d frame of a particular shape, viewed under particular lighting conditions, then that ought to be extrapolate-able - yes, it might require specialist lighting techniques, like barn-door baffles, gel-frames or whatever they used to use in 30's Hollywood films, to cast light only on the subject's eyes - some kind of cutout mask - but I'm sure it's got a name - but with enough tools available to the artificial virtual model builder, it should be possible to recreate - at least the general mood and dynamics of an image.
zen_tom, Jan 15 2019
  

       I really would like to see this attempted with some of the Vermeer interiors though.
hippo, Jan 15 2019
  

       Also, you'd have to fill in a huge amount of missing information. For instance, you'd have to know that Whistler's mother actually had a tattoo on her right cheek.
MaxwellBuchanan, Jan 15 2019
  

       Turpentine and a cloth? Get one of them computer controlled robot arms and take one layer off at a time.
not_morrison_rm, Jan 15 2019
  

       Johannes Gechler, a Dutch artist in the early 1900's, only produced about 50 paintings in his lifetime. But X-rays showed why: he actually painted the hidden parts of things, and the backs of things, and the insides of things, in consecutive layers before the final layer containing the viewer-facing parts of things. You can actually take scans of his paintings and expand them front-to-back, and you get a reasonable approximation of the complete scene.
MaxwellBuchanan, Jan 15 2019
  

       Sounds a bit like Richard Dadd.
8th of 7, Jan 15 2019
  

       "Johannes Gechler" sounds like "Richard Dadd"?? What accent are you using?
MaxwellBuchanan, Jan 15 2019
  

       Glad you didn't pick Judith Beheading Holofernes, though that idea and Max's anno make me think of the invisible? How are we dealing with the texture of the wound, the handle of the sword? Is the risk here that we are in "Computer... Enhance" territory here?
calum, Jan 15 2019
  

       // make me think of the invisible? //   

       Well done. Now, don't think of a pink elephant.   

       // we are in "Computer... Enhance" territory //   

       There's nothing in the help file that says you shouldn't propose such things.   

       <Deckard>   

       "Enhance 57 to 19. Track 45 left. Stop. Enhance 15 to 23. Give me a hard copy right there."   

       </Deckard>   

         

       // What accent are you using? //   

       Norfolk. One protolinguistic metasyntactic grunt sounds pretty much like another.
8th of 7, Jan 15 2019
  

       Basically a special case of photogrammetry then.
pocmloc, Jan 15 2019
  

       //There's nothing in the help file that says you shouldn't propose such things.//
True. I was only highlighting it as I think that's the most interesting part - what do we need to teach the computer to extrapolate wounds, or what the painter's painting looks like in Las Meninas?
calum, Jan 15 2019
  

       If we take (not literally) an old master painting such as a beautiful Dutch interior scene we can analyse it and attempt to reconstruct a 3D version of the scene depicted. This is a problem to which there are many solutions. So, there is ambiguity in the painting, in the objects depicted, their orientation in 3D space, and how they are lit, and there may be several viable 3D models and lighting set-ups which render to produce a more or less good facsimile of the painting. However, there will always be one 3D model which produces the best solution - both in terms of the accuracy of where things are in the scene and in terms of how accurately they are lit - and this solution will be that the 3D model contains a copy of the painting you are looking at. I.e. Vermeer's painting "Girl with a pearl earring" isn't a painting of a girl with a pearl earring, but is a painting of the painting "Girl with a pearl earring". There's a risk that if you build a system to search for the best solution, it will always find this solution - if you don't want this, you might have to explicitly prohibit it.
hippo, Jan 15 2019
  

       What about The Arnolfini Wedding ? <link> Would the reflection in the mirror have to show the graphics team hunched behind their monitors ?   

       // what the painter's painting looks like in Las Meninas? //   

       Ah, but that's "context-sensetive", shirley ?
8th of 7, Jan 15 2019
  

       //the reflection in the mirror//

That's not a mirror - it's a lens, with an identical but mirror-imaged room behind it with two people standing in it with their backs to the viewer.
hippo, Jan 15 2019
  

       Amazing. Who knew, huh ?
8th of 7, Jan 15 2019
  

       I'm still waiting for the the Arnolfini Bitterly Disputed Divorce painting to turn up.
not_morrison_rm, Jan 15 2019
  

       It's on the back of the Arnolfini Wedding painting, didn't you know ? At least, half of it is ... she got half in the settlement, he kept the other half, neither of them wanted the wedding picture so for a long time it was hung with the "front" facing the wall ...
8th of 7, Jan 15 2019
  

       Elon Arnolfini should have had a better divorce lawyer.
not_morrison_rm, Jan 15 2019
  

       //That's not a mirror - it's a lens//   

       There's quite a weight behind the theory that many Dutch and Italian painters from the 13th century onwards used lenses as a method to sketch in a canvas using a camera obscura.   

       With the Arnolfini Portrait, 3D models have been made of the chandelier and when rendered it matches the painting exactly.
bigsleep, Jan 19 2019
  

       Nice, a picture is a thousand models.
wjt, Jan 19 2019
  

       this is pretty baked, see link
theircompetitor, Jan 19 2019
  
      
[annotate]
  


 

back: main index

business  computer  culture  fashion  food  halfbakery  home  other  product  public  science  sport  vehicle