4 ms·

Surprisingly good quality, considering his method. You get the best sense of his method when it goes a bit wrong, in photo 27: http://www.boston.com/bigpicture/

• daeken 14y ago
  What's most interesting to me, personally, is the water in these photos. Because the water is constantly moving, capturing the filtered images repeatedly won't work -- you end up capturing different images for RGB. That's what gives you that really neat rainbow effect; it looks like oil in a rainpuddle. Edit: You can see some of this in photo 18 where the stems of the flowers were presumably moving very slightly: http://www.boston.com/bigpicture/2010/08/russia_in_color_a_century_ago.html#photo18 http://www.boston.com/bigpicture/2010/08/russia_in_color_a_c... Edit #2: I just realized that I've used the same exact "effect" in demo code. I had a function in a shader that would generate a pixel color given a time value; I ran it three times for each pixel with a slightly different time value and then grabbed an R/G/B from each of them. You ended up with the same exact rainbow effect that was subtle when motion was limited and extreme when high speed motion was occurring. Not that that adds a ton to this thread, but I got a kick out of it.
  • vhf 14y ago
    >Because the water is constantly moving, capturing the filtered images repeatedly won't work -- you end up capturing different images for RGB. You're right, of course. But don't forget that the shutter speed was very low on these images. It would have harmed if the shutter speed was high : short exposure time of quickly moving water would have given very different water images for RGB. When exposing longer, motion blur makes the water seem to flow, becoming soft and blurry.