About Bokeh, Hacker News

About Bokeh

What is Bokeh?
Bokeh is a term, originally from Japanese (暈 す– to blur; orぼ け), that has come to mean the subjective aesthetic quality of the out of focus area of ​​an image. Two key terms in this definition are SUBJECTIVE and IMAGE. One cannot measure bokeh, as it is something that may be pleasing to one person and not another. I have my opinion of what makes for good bokeh, and, as I own this site, I get to define good and bad Bokeh for the purposes here! The other key term is “image.” Bokeh is a property of IMAGES, not a property of LENSES. Many people, myself included, even on the 1st version of this web page, get this confused, and this confusion contributes to a lot of misunderstanding in conversations in photography forums regarding this issue. What CAN definitely be said is that certain properties of lenses cause bokeh to be rendered differently by different lenses. When we refer to a “the bokeh of a lens” what we really mean is that particular lenses ability to affect the bokeh of an image for the better or worse. This is an important distinction. Many scenes will produce wonderful creamy bokeh with even the worst of offenders, like mirror lenses. Other scenes with harsh lighting, clothes lines running across the picture, specular highlights are much more difficult to get beautiful creamy blurred background images from. And here is where the qualities of the lens make a difference.

Bokeh of an IMAGE is highly dependent on the lighting of a scene, distance between lens, subject and background, how much Contrast is in a background scene, how much the background contributes to an image (maybe you WANT the background to be partially visible), how busy the background is, as well as lens factors, as well. This website is ONLY meant to address the LENS FACTORS that affect the bokeh of an image rendered by a particular lens, as this is the ONLY part of bokeh that should, at least in theory, be measurable (although many would argue it is not) .

As I said before, there seems to be a lot of misunderstanding about bokeh, even in the photographic website forums. Some people have written that bokeh isn’t affected AT ALL by lenses, others write that the way a lens affects bokeh (What I will call “lens bokeh”) is all about the shape of aperture blades, others that the only thing that matters is separation between the background and subject (ie the size of the circle of confusion), others that bokeh does not matter at all and it is only the snobbery of photographic elites to justify their buying super expensive lenses. Still others argue more about the origin of the word, how it is pronounced, and some just flat out hate the word altogether, suggesting it is just some fad word about something that was known for a long time in other circles. I guess the theory is: “back in the my day we did not need no word for it, that was good enough for us then, and, dammit, that should be good enough for those little whippersnappers now!” And now that a single word for it has been coined, these critics complain because they think the people who invented the word believe they have “invented” the science, understanding, and concept of Bokeh. How dare they invent a word that is short and concise! I would rather not spend time arguing about the words origins, who really invented it, etc, as I think these arguments are completely irrelevant to understanding the issue, and just get in the way and distract from where the conversation should be– learning.

The rest of the misconceptions are right to a certain degree, but there is more to the story of “lens bokeh.” To start with, people who think that caring about Bokeh is just photographic elitism are slightly off base, in my opinion. It is true that people who do only wide sweeping landscape photography hardly need to worry about Bokeh– the important thing for them is a lens that is razor sharp at higher apertures, and they usually don’t want ANYTHING out of focus. But with portrait pictures, flower pictures, macro pictures and when you want to make a subject really pop out of the picture, one will have a significant portion of the picture that is out of focus. Why would one think that this wide swath of film area doesn’t matter? A sharply focussed and busy background often detracts from the image, drawing your attention away from the subject (see the image on the introduction page for an example). An image with bad bokeh does the same, making a busy background out of what should have been smoothly rendered background that would make your subject stand out. Furthermore, the shape of aperture blades DOES make a difference to the bokeh of a picture, but only in that non-circular aperture blades will lead to non-circular highlight out of focus “circles.” This should mostly affect single specular points of light in the background, while the effect on solid shapes and lines should be minimal. There is far more to the quality of blurred areas in images than simply the circularity of the aperture blades. It also has to do with the path of light through the lens elements, how well corrected the spherical aberration is, as well as coma, curvature of field, astigmatism and a few other properties of lenses. At this time, I am focusing on Spherical aberration, since this appears to have probably the most effect on “lens bokeh.” As this site develops I may expand to other lens parameters which affect bokeh, as well.

Also, one has to be aware of the fact the “front bokeh” and “back bokeh” are not the same thing. That is, a lens that has excellent bokeh for backgrounds, with a lovely blur to the out of focus (OOF) areas, may have TERRIBLE looking blurred foreground objects. In fact, based on my limited knowledge of optics, it seems that this is not really a “may have” but a “must have” situation. The only lens I know that defeats this is the Minolta (now Sony) 326 mm f / 2.8 STF (smooth transition focus) lens, which has a special apodization filter built in that specifically dims the OOF areas towards the edges. Here bokeh is affected more by this filter than by the properties of the lens itself. Nikon makes some “DC” lenses, which stands for defocus control, which lets you manipulate spherical aberration to produce better front or back bokeh. The advantages are, you can autofocus the camera and you do not lose any light (the minolta lens has an aperture value of 2.8, but only lets enough light through for the transmissive brightness to be equivalent to an aperture of 4.5, and the apodization filter disables autofocus), but the disadvantage is that when you dial in good front bokeh, you are also dialing in BAD back bokeh, and vice versa.

How an image is formed
In order to really wrap your hands around Bokeh, one needs to understand a little bit about lenses in general. As an introduction to basic lens optics, I will give a short primer on how single (thin) lenses work. Camera lenses are NOT thin lenses, but their optics are far beyond my limited knowledge to understand. And for our purposes, the principles that I am going to highlight for a thin lens do, in fact, also apply to “thick” lenses. First of all, the focal length of a lens (you know that mm lens you own) is the distance at which an object that is an infinite distance away will focus to on the other side of the lens. Take a picture of the moon and you can be sure that very close to 90 mm behind your lens, inside your camera, there is a sharp virtual image of the moon. Luckily for you, this happens to be EXACTLY where the film plane is, and you will get a nice sharp image on your film. So, what about the leaves on the trees that are between you and the moon, that invade your shot? Well, they are a lot closer than the moon. Despite what we think of as “not focussed” when we look through the viewfinder, these ALSO are in extremely sharp focus. The only thing is, they project a sharp virtual image at perhaps mm behind the lens instead of mm. So, if the film plane was placed at 100 mm back, THEY would be in focus. But then the moon wouldn’t because it’s virtual image will now be mm behind the lens, in other words, 8mm in front of the film plane. The focusing ring we use does not “defocus” or “focus” any areas, all it does is effectively move the lens element (remember, think of it as a single lens) forward or backward, changing where all the virtual images are in relation to the film plane. The moon will always image (mm behind a) ***************************************************** mm lens, but since this lens moves forward and backwards, the sensor / film plane will not always be exactly mm behind the lens. Thus, we move the lens so that it is now mm in front of the sensor plane, and now the leaves are in focus. Of course, now the moon has focussed 8 mm in front of the sensor plane, and is now not in focus. Below is a diagram showing this effect. All objects that are in the foreground will end up with their focal points BEHIND the film plane, and all objects that are in the background will focus their image BEFORE the light hits the film plane. It turns out that this is HUGELY important to bokeh, as we will see in a bit.

Depth of field
Bokeh is very affected by the relative distance of lens to subject and lens to background. If the background is 1 foot behind a person at 263 yards, don’t expect to get much blurring! If on the other hand, your subject is 2 feet away and the background is the moon, at several hundred thousand kilometers away, you would be hard pressed to find a lens that could get it all in focus, and you would expect the diameter of the out of focus area to be very large, indeed. Thinking of the principle of different focal locations for different distances, one should imagine light as a cone with a focal point near the film plane, and the cone extending out to the outer reaches of the last lens element. If the focus is exactly at the film plane, you will get only the tip of the cone, and will image a point. If the cone focusses behind the film plane, you will get a little circle where the cone of light intersects the film. If it focuses in front of the film, the light will continue from the point of the cone coming off the lens to form ANOTHER cone going the opposite direction, and you will still end up with a circle of light. The distance between the film plane and the focal plane for any point affects the size of that little circle of light. Large distances between subject and background or foreground will make large circles of light, and small distances will make very small circles of light.

Bad Bokeh
Bad bokeh is, unfortunately, something that i am very familiar with, and frankly, did not really notice until I took a critical look at my pictures. Essentially, instead of making the background creamy and smooth, with objects blending into each other, a lens with bad “lens bokeh” creates harsh, sharp lines around objects, with small specular highlights adding on top of each other to make a very busy series of bright spots with sharp borders. It is highly distracting and draws your attention away from the subject. Any branches or thin lines in the OOF area are transformed from a single OOF line to TWO extremely SHARP lines with some blurring between them. While my lenses didn’t have terrible bokeh, they were far from great, too. The worst offenders are mirror lenses which are famous for creating perfect rings around any specular highlight.

Good Bokeh:OK, here is where I get to define what I personally like about Bokeh. This is all my PERSONAL opinion, but I suspect many out there in the photography community might agree with me. Most of my portrait pictures, which is the usual time I shoot wide open to accentuate the subject, have the person as the closest object and a background, but little or no foreground. That is just the way I shoot. So for me, what I care about is BACK Bokeh. Front Bokeh be damned. Most of the time, I want an über creamy background with no details, nothing sharp at all, nothing distracting, so your eyes are immediately led to the subject, which I want in perfect sharp focus (if you want softening, I would argue you can simulate a soft focus lens in photoshop). Yes, there will be times when I want the opposite, but they will be rare for me, and it is not worth shelling out extra cash to get a special lens for front bokeh. Any points of light need to blur out to an imperceptible edge, not unlike a gaussian blur in Photoshop. Out of focus lines need to be similarly blurred. Intersecting points of light need to blur into each other without creating edges that are sharp, so the whole effect is a creamyness, a “bland” background that nobody will notice (unless they have studied bokeh), but people will notice the picture and say “There is just something about that picture that really appeals to me.” THAT is good bokeh, in my opinion. You may have your own idea of ​​what good bokeh is, but this site will be evaluating lenses based on MY opinions. And lets face it, there are actually times when BAD bokeh really works for a shot, but that is usually for people who have a whole lot more artistic talent than I do.

Spherical Aberration:
It took me some time to understand what spherical aberration is. Essentially, lenses are usually made with spherical elements. These are very much easier to make than aspherical elements, and are therefore much cheaper to make. Even the best, most expensive lenses use more spherical lens elements than aspherical elements. Unfortunately, this comes at a cost. A spherical lens is NOT a good optical design, because the focal point is not the same for light hitting different parts of the lens. Specifically, at the edges of the lens, curvature of light is too high, and the focus comes in front of the main focal point of the lens. This is really only important for the edge of the lens, as the central part of a lens focuses very close to the “focal point.” The result is a significant difference in the bokeh in front of the focal point and the back of the focal point.

Lenses are designed to correct for spherical aberration with multiple aspheric elements and different lens groups with different refractive indexes. They have to do this because where spherical aberration exists, there is no such thing as a single focal point (see diagram above if you don’t understand what I mean), and it is my understanding that you cannot get a truly sharp subject. But, remember, this is a Bokeh website … we do not care in this discussion about the in-focus subject. We are concerned about the out of focus area. And the correction of spherical aberration is the key. In a “perfect lens” there is no spherical aberration, and therefore light is evenly distributed throughout the blurred spot (see diagram above). This leads to a perfect circle of out of focus highlights, with equal intensity from center to edge then abruptly stopping at the edge. Some lenses come very close to being “perfect.” Mind you, this describes a “perfect” lens from the point of view of a lens designer, for whom any uncorrected “flaws” in a lens are sign of shame. Unfortunately, “perfect” lenses generally do NOT make great Bokeh. Not terrible, but not great. We like the flaws in lenses! I tried to find a diagram of the light path that would explain this, but was unsuccessful. Therefore, I did some photoshop work to simulate the light path by drawing a lens, and lines converging to a focal point, with the rays that were further from the axial center focussing closer than the focal point. As we know where the focal point will fall for an out of focus area (background objects focus before hitting the film plane, and foreground objects focus behind the film plane), we can use our knowledge of spherical aberration to see how the bokeh of a lens will be for both foreground and background. If you examine the diagrams below, you will see that for a lens that is not well corrected for spherical aberration, foreground elements will give you a ring artifact, and background elements will give a soft edged circle. And if it is overcorrected, the opposite arrangement will be had. Thus, for a lens that is poorly corrected for spherical aberration:

And for a lens with overcorrected spherical aberration we see the diagram below:

One thing you will see on examining these diagrams, is that not only will you change the QUANTITY of bokeh as the focal plane gets further and further from the film plane, but the actual shape and distribution of light changes as well. Bokeh is not the same for different distances into the background. Of note, if my diagrams are accurate, and I am not 200% sure they are, then unless you are correcting spherical aberration perfectly, then you should lose some contrast and some sharpness in your images. Note that in image B on both the uncorrected and overcorrected spherical aberration diagrams (the in focus area), surrounding the spot in the middle is a ring of light, faint, but visible. Every single point of light in the image will also have a ring of defocused light around it, and this will be detrimental to the local contrast of the image.

Spherical aberration can also be reduced by stopping down the lens, thus not having as many rays hit the edge of the lens. But if you are reading this article, then chances are you cringed when you read the words “stop down.” This is a BOKEH thread. Open those lenses up, baby!

Lines from dots. What can you tell from points of light?
It turns out that knowing the shape and character of the out of focus specular highlight areas tells you a lot about the character of the bokeh of a lens. Many people hold particular ire for “Nisen Bokeh.” This is basically the appearance of double lines around out of focus straight lines in the background, like telephone wires, branches, etc. It turns out that one can think of a line as just an infinite number or infinitely small points strung up in a line. The beauty of this is, we can just add up a bunch of these out of focus points in a row and see how well the lens will do with nisen bokeh! Furthermore, a solid is just an infinitely large series of infinitely thin lines stacked up against each other. Again, if we know the shape of the out of focus point, we can determine the shape of a line it makes, and therefore can determine the shape of a solid it makes. See the diagram below for more clarity.

Based on this and knowing the kind of dot we have for a lens, we can determine the shape of lines in the background, if they render pleasingly or not. Perfect lenses show no nisen bokeh, but they have sharp edges, which are relatively harsh and do not blend as well as soft edges. Mirror lenses, and lenses that are over-corrected for spherical aberration show nisen bokeh, also have harsh edges, and do not lead to a pleasingly creamy background. Only lenses with poor spherical aberration correction show nice, pleasing creamy backgrounds, with no sharp lines to speak of in out of focus areas, good blending of overlying objects, etc. These are the lenses we want.

We can see what images will look like if we stack multiple specular highlights up against each other. You can decide for yourself which is the less distracting for a background.

Simulated Bokeh- how the shape of specular highlights affects a background
I have made the following scene to demonstrate this effect. Superimposed on this I have made what would be the blurred image (where the circle of confusion=(********************************************************% of the sensor width) for a lens with poor bokeh, one with good bokeh, and one with neutral bokeh. Surprisingly, while there are differences, the bad bokeh lens is not as bad as you might expect. I suspect that if I expanded this to make the circle of confusion>(*****************************************************%, the two would look even more similar. Here is the scene:

The white lines on the image are meant to indicate areas of specular highlight. Here is the image of the scene as taken through a Mirror lens, or at least one with poor Bokeh.

Here is one to emulate good bokeh (simulated to be the Minolta / Sony STF (:)

Finally, here is what an “ideal” lens with perfectly corrected bokeh would look like for the same scene.

You can see that the poor Bokeh lens has harsh lines in the blurred areas, as edges of objects do not have a smooth transition from the edge of he object to the background (in this case black). Furthermore, the highlight regions show the characteristic bright circle pattern that many people (including myself) feel is so objectionable. The good Bokeh Lens shows lovely smoothing of all lines, everything blurs together nicely, and is much more smooth and creamy. The “ideal” lens (perfectly corrected spherical aberration) is half way between, with lines, but less clear than with the poor bokeh lens. Note that the most profound differences are in the stems, where there are thin elements that have varied definition base don the bokeh of the lens.

The only difference between the three above images was this: they were built up with different shaped blurred circles. The three circles of confusion that were used can be compared here:

Bad Circle of Confusion Good Circle of Confusion Neutral Circle of Confusion

******************************************************** This example uses a circle. of confusion of 58% of the lens horizontal diameter, a fairly large blur that would represent a large distance between the background and the subject. This is the best case scenario for bokeh- even a lens with poor bokeh will produce a fairly smooth appearing background given enough separation between the subject and background (single specular highlights aside). Had I known this at the time I made these diagrams, I would have chosen a 50% blur perhaps, as I suspect it would have shown more substantial differences (and if I ever have a whole bunch of extra hours I may go about doing this again- not a fun thought for me).

Clearly, from the example above, the shape of the out of focus area of ​​a specular highlight ALONE appears to drastically impact the bokeh of a lens. This is the principle by which I choose to test lenses. SO MUCH can be said about the way a lens affects bokeh based on the circle of confusion, that I feel pictures of circles of confusion at various distances and various f-stops should give at least a pretty good idea of ​​the “lens bokeh” of a lens. The great thing about this is that, assuming it is completely true, we can create a very reproducible method of testing any lens, that is not so subjective as looking at pictures of light blobs through trees, with different white balance, different amounts of contrast , etc. Bokeh will always be a subjective quality of images, affected by many things, especially the way a picture is taken and the background that is chosen by the photographer. But it would be folly to ignore the contribution of lens design to the rendering of images.