Flowers in 3D

Volume 8, Issue 103; 18 Jul 2005; last modified 08 Oct 2010

I slipped out of the house late on Friday afternoon. I took the camera and explored the flower beds, which are looking pretty good this time of year.

I slipped out of the house late last Friday to escape from a certain sense of ennui. I took the camera and explored the flower beds, which are looking pretty good this time of year.

In the back of my mind was a critical analysis of my photographs, sent in by a gentle reader, which suggested that they were very “two dimensional”. The particular examples cited included works like the pub signs series which are, I freely admit two dimensional, but which in all fairness are photographs of essentially two dimensional objects. But they also included examples from my series of flowers from St Thomas.

So I tried this time to frame images which didn't appear to exhibit the same sense of flatness. Did I succeed?

I should say that I'm not in the least offended by the criticism. It was carefully considered and offered, I think, in a spirit of interest in the images. Hey, I get to write it, you get to read it, the comment link's there for you.

Squared Circles

Very explicitly two-dimensional are a series of images that I took for the “squared circle” group on Flickr. Paul Sumner Downey introduced me to the group when I posted “Fluorescent” some months ago.

These are not all equally obvious examples of squared circles, but I like the fact that they all share a secondary theme as well.


Finally, the abstraction of the Coreopsis seemed to invite the possibility of good desktop wallpaper. I'm not sure the exposure is just right, but maybe that's not too important.

This little inchworm is nestled in the upper-left corner of the wallpaper.

False Color

Here's another shot of the little weed that appears above in a squared circle.

What's remarkable about this weed is the indescribable saturation of the purple in its blossom. These images reflect a pale shadow of the true color; as if I was trying to demonstrate the appearance of bright sunlight on a tropical beach with photos of a sandbox on an overcast day.

I have no idea what part of the chain is most responsible for the failure: the camera's sensor, the compression algorithms, the translation back into RGB values, or something else.


Pretty reasonable sense of depth - the ones that work best are those with sufficiently wide aperture / narrow DoF, that also have something slightly blurry in the background, suitably positioned in the frame as to lead you from one to the other.

As for the purple saturation: overexposure at source would do that, as you've blown mostly the blue and occasionally the red channels; shoot in RAW mode on a less sunny day, underexpose by half a stop or so, or at least do *not* get highlights blinking a warning. Then boost the exposure in PS/CameraRaw or Cinepaint with a gamma curve, taking care never to blow a highlight. Also, the edges have not come out well - over-sharpening and/or colour aberation (also fixable directly in PS/CameraRaw) could solve that, or failing all else, a delicate blurring. Probably the quantization to final jpeg wasn't very good, either: try setting floating point accuracy and about 98% quality, see what happens.


—Posted by Tim on 20 Jul 2005 @ 09:04 UTC #

I'd left this page up to try and give an answer about the purple flower, but I just refreshed and noticed that Tim's given an excellent answer.

The only thing I'd add, is that, at least on my Nikon D70, the histogram, and therefore the highlights markers, are calculated using a luminance (Y) channel that's presumably derived via a normal luminance calculation; one that's meant to approximate the luminance response of the human eye. This dates back at least as far as the 1953 NTSC standard, and while it varies between systems a bit, it's roughly: Y = 0.6G + 0.3R + 0.1B. What I noticed, while shooting some yellowish-orange jonquils, was that I was blowing R and G without being anywhere near clipped in the histogram. This would be even more relevant for your purple flower, where the large blue values would be having little impact on the calculated Y value.

—Posted by Martin Kenny on 22 Jul 2005 @ 01:00 UTC #

P.S. the Flickr image src values in the comment-posting mode are wrong, so all the images are broken while you write a comment — Marty

—Posted by Martin Kenny on 22 Jul 2005 @ 01:03 UTC #