4 ms·

Congrats. This is very cool. This is probably less a critique of your project and maybe more a celebration of cubism: The clock in the Picasso portrait is not

Congrats. This is very cool. This is probably less a critique of your project and maybe more a celebration of cubism: The clock in the Picasso portrait is not "harmonized" in terms of cubist motifs at play in the composition. The contours appear merely dilapidated. Maybe this passes for someone with little concern for painting or art history but it was like a sort thumb for me. Maybe this is a good study reference on the gap between pattern recognition and human perception. Maybe not. I'd be interested in other thoughts.

teenbear 9y ago
I think the gap in this case is not as wide as you think given that it is only looking at this one image. Had the network been trained for one art style in particular it may be able to pick up on that detail similar to if a person had studied many paintings in that style.
jerf 9y ago
Plenty of them fail to stylistically harmonize even if they visually harmonize. Of course, in some cases, it's debateable whether stylistic harmonization is even possible at the human semantic level. The bicyclist on the bridge in the scream painting was quite bad, and the Eiffel Tower in the cityscape was noticeably degraded by the fact the painting didn't have the colors that the artist would probably have actually used, so it's rather bluish. What's more impressive is that several of them work quite well, and barring anachronism, at the sizes given in the samples you might not know what was added if they didn't point it out. The park bench I found particularly impressive, for instance. I suspect if we increased the resolution it would break down, but at this size it works pretty well.
- macawfish 9y ago
  I was very impressed by the one with the large book added next to the person in the grass. The curvature of the pages changed to be more gentle.
- extralego 9y ago
  Yeah good point about the resolution. So I wonder what effect lowering the resolution would have? In general, it’s definitely impressive how well they work.
  - jerf 9y ago
    I've measured the "realism" of graphics in terms of resolution for many years now, because nowadays we generally have realistic graphics, as long as the resolution is small enough. (Compare with, say, an NES game, where no matter what you do to that image nobody will ever find it realistic.) There's a lot of techniques/artists/etc. I've seen that are perfectly realistic looking at 320x200 (say), but couldn't sustain the illusion of realism at 1920x1080. This way I get a reasonable metric of whether or not graphics are "photorealistic", or in this case, match the style of the painting. It would be interesting to see them do a higher resolution version, or if they have them. It's possible it would indeed fit right in even so; now that I think about it the microscale of these paintings are probably very stereotypical and the algorithm might be able to reproduce them well. It would be especially interesting to see if it would correctly reproduce brush strokes, which have a lot of context to them. Given that "deep learning" can do things like reproduce the structure of a TV script accurately, it doesn't seem out of the question.

teenbear 9y ago

I think the gap in this case is not as wide as you think given that it is only looking at this one image. Had the network been trained for one art style in particular it may be able to pick up on that detail similar to if a person had studied many paintings in that style.

jerf 9y ago

Plenty of them fail to stylistically harmonize even if they visually harmonize. Of course, in some cases, it's debateable whether stylistic harmonization is even possible at the human semantic level. The bicyclist on the bridge in the scream painting was quite bad, and the Eiffel Tower in the cityscape was noticeably degraded by the fact the painting didn't have the colors that the artist would probably have actually used, so it's rather bluish. What's more impressive is that several of them work quite well, and barring anachronism, at the sizes given in the samples you might not know what was added if they didn't point it out. The park bench I found particularly impressive, for instance. I suspect if we increased the resolution it would break down, but at this size it works pretty well.

macawfish 9y ago
I was very impressed by the one with the large book added next to the person in the grass. The curvature of the pages changed to be more gentle.
extralego 9y ago
Yeah good point about the resolution. So I wonder what effect lowering the resolution would have? In general, it’s definitely impressive how well they work.
- jerf 9y ago
  I've measured the "realism" of graphics in terms of resolution for many years now, because nowadays we generally have realistic graphics, as long as the resolution is small enough. (Compare with, say, an NES game, where no matter what you do to that image nobody will ever find it realistic.) There's a lot of techniques/artists/etc. I've seen that are perfectly realistic looking at 320x200 (say), but couldn't sustain the illusion of realism at 1920x1080. This way I get a reasonable metric of whether or not graphics are "photorealistic", or in this case, match the style of the painting. It would be interesting to see them do a higher resolution version, or if they have them. It's possible it would indeed fit right in even so; now that I think about it the microscale of these paintings are probably very stereotypical and the algorithm might be able to reproduce them well. It would be especially interesting to see if it would correctly reproduce brush strokes, which have a lot of context to them. Given that "deep learning" can do things like reproduce the structure of a TV script accurately, it doesn't seem out of the question.