Thanks for the comment Glenn NK. You are, of course, correct that bokeh is related to the quality of a blur and DOF to the quantity of oof, but in truth they are very closely related. Depth of field is determined by how narrow or wide a cone of light coming from an object is after it passes through the aperture (among other things). The image will be considered sharp in all areas in which the circle of confusion is less than, say 0.02-0.035 mm, depending on how tight your tolerances are. Outside of this and we get a blur, and the circle of confusion diameter at a particular distance shows how much blurring there will be at this distance. If my premise is correct, that how a lens affects the bokeh (quality of blur) of an image relates to qualities of the out of focus circle from a point of light, then it has to be understood that this is very much affected by depth of field. In other words, the circle of confusion gets bigger with a shallower depth of field, all else being held constant. The character of the oof circle of confusion may also vary at different distances from perfect focus, and this also affects the image.
As you pointed out on your post, I am trying to keep as honest as I can and say it is opinion when it is. As bokeh is a subjective property of images, it cannot be measured, and even people might disagree on what makes a lens better at rendering bokeh than another lens. This is why I focused so much on the science behind the way a lens affects bokeh (note, I am being very careful NOT to call it bokeh itself, only lens parameters which afect the out of focus area), which should be a lot less subjective. Some people may prefer mirror lenses with their little circles. I am only speaking for myself, that I prefer nice smooth, non-edged lines and more blur rather than less blur, no sharp line interactions between specular highlights I am willing to bet, however, that many, if not most, other people would also prefer this kind of bokeh to more harshly rendered bokeh.
My central premise with the article, however, doesn't have to do with images, but rather how a LENS will affect the bokeh of an IMAGE. And I think that is a LOT less complex than factoring in the background elements etc. Afterall, the light path through the lens is the same no matter how bright an object is, a lens doesn't care how much contrast is in a scene, a lens doesn't change what it does based on the scene. A lens does what it does unaffected by background- same light path, lenses don't bend when they see something really bright or a low contrast scene. Therefore, we can measure how a lens will affect an out of focus point of light and infer from this what it will do to an image. As an example, look at the last 3 large pctures in the "about bokeh" section:
http://bokehtests.com/Site/About_Bokeh.html
They show the same image made up from different shaped OOF circles of confusion. This IS how a lens makes an image-- every single point of light of an object the camera sees is actually imaged as a little blur circle. When in focus the circle is really small, but not when out of focus. In the example, I made the blur circle equal to 20% of the lens diameter for 3 different blur circles. You can see that all three blurred images are distinctly different, and not only where specular highlights, and the only difference is the shape of the circles they were built from. Anyone that doubts this-- please feel free to get a copy of photoshop and do the same experiment yourself. I think the excercise makes it pretty clear that the shape of the circle of confusion makes a huge effect on the bokeh of the image. And what's more, this is a property of the LENS, that will be INDEPENDENT of the image used. Some lenses will tend to blur backgrounds smoothly, others not. And if you read the beginning part of the link, you would see the contribution of spherical aberration. Indeed, if you have ever seen or read about the Nikkon DC lenses, all they do is manipulate spherical aberration, and images with them (look online) at different settings exactly reproduce my little diagrams of uncorrected or overcorrected spherical aberration, both for front and back bokeh. And of course, they are noted for having excellent bokeh by most people when used correctly.
So, while it is subjective what looks nice in a background for a particular image, I would contend that the way a lens affects this background is at least measurable, and MOST people would agree that the smoother rendering will usually be more pleasing (unless you are going for some kind of special effect). Subjectivity is an interesting concept. If we think of the taste of two brown substances, well, chocolate may not be objectively measurably better tasting than poop, but I would be willing to be you won't find too many takers of the poop if offered the choice of both. At least not on the planet earth.
Klaus
--
http://www.bokehtest.com