30 December 2007

Blown Highlights and Blown Highlights

There is a lot of talk about "blown highlights" in digital photography forums, especially with respect to my camera of choice, the Olympus E-510. I think it needs to be clear what we are talking about. There are two kinds of blown highlights. The first is where highlights are blown in the captured image. These are lost forever to the photographer, even if storing the raw file data from the image capture. The second kind is where highlights are blown in the image developed from the captured image.

The first kind comes from incorrect exposure or dynamic range limitations in the sensor, which the photographer cannot do anything about after the exposure is made. The second kind the photographer has more control over (other than getting the exposure right) through the JPEG engine adjustments. Blown highlights in the developed image can be the result of these adjustments, not a limitation of the camera or exposure.

To clear up any confusion in my mind over the effects of camera adjustments I developed images representing the extremes of the 510's development adjustments. The intention was to show the widest range of contrast possible using the in camera adjustments on a difficult subject.

Noise reduction and sharpness also influence highlights in the result, but their effects are minimal compared to contrast and saturation. I left the sharpness at minimum, as if I intended to bring the image into my image editor before applying sufficient sharpening for printing. Noise reduction, by blurring pixels, can affect the measurement of dynamic range, so it was turned off (the E-510 allows noise reduction to be turned off completely).

Muted

Apartment Building - Min

This scene was shot with the E-510 and developed in Master 2.04 using the following settings:

Picture Mode: Muted
Contrast: -2
Saturation: -2
Sharpness: -2
Noise Filter: OFF

(The Master 2 raw developer honors the E-510 camera settings, giving the same results from developing an image at the computer as in the camera).


Vivid

Apartment Building - Max

This scene was shot with the E-510 and developed in Master 2.04 using the following camera settings:

Picture Mode: Vivid
Contrast: +2
Saturation: +2
Sharpness: -2
Noise Filter: OFF

The difference is best seen by examining the venetian blinds in the window on the corner of the first floor nearest to the viewer. Although it is difficult to see in the photographs resized for the web, the window shows nicely how the contrast and saturation (including Picture Mode) adjustments influence the highlights of the developed image, so I have provided a detail:

E510 Highlight Comparison

In the Vivid image, the highlights on the blinds appear completely blown. In the Muted image more of the venetian blinds are visible. The lower contrast rendering shows more detail than the higher contrast one. What appears as a "blown highlight" shows detail as the contrast and saturation are reduced, proving the detail was there in the original capture. This is not an example of highlight recovery, it shows how the raw image contains detail that gets obliterated by the contrast curve.

(If you would like to read more about how contrast curves influence perception and testing of "dynamic range" read How to Magically Improve You Camera's Dynamic Range).

True Blown Highlights

In fact, this photograph does contain highlights blown at capture, the yellow sidewalk curb is blown along the front length and the far sky toward the horizon is blown a bit (when recovery of highlights is attempted, this part of the sky posterizes). The detail in these areas is completely lost, even in the captured image.

What does not count as a blown highlight is a specular reflection or element in the scene that is naturally so bright it does not contain any detail. Such reflections should be rendered in the captured image and final print as completely white without any texture. The reflection from a wave top, chrome on an automobile, and the like. In the example picture, the yellow curb is painted with a material designed to reflect light to give it greater visibility, which is partly why it exhibits blown highlights.

Local Contrast Enhancement

There is a way to increase the perceived contrast of an image without blowing the details in highlights. One is called "local contrast enhancement" and is a technique for increasing what is called "micro contrast." Micro contrast is the amount of contrast at the edges of objects. Increasing micro contrast makes an image appear sharper, clearer and increases perceived acuity, which is why photographers seek lenses with better "contrast" by which they mean the ability to provide greater contrast between edges and in details. It is the contrast of fine detail, not the overall contrast of the scene.

Basic LCE is simple. Just set apply unsharp mask with a very wide radius and a small amount. You will see the "fog" lift immediately from your picture. Local contrast can be pushed beyond small amounts without blowing highlights by using a "gray mask" (This is a mask inverted to protect highlights from changes) in LAB space.

Using LightZone, one can protect highlights from changes by adding a sharpen tool to the stack and then selecting only the highlights from the tonal range, and inverting the the mask, which automatically creates a mask excluding effects of sharpening from highlights. Another LightZone trick is softening blown highlights to make the less distracting and more acceptable to the eye, by adding a blur tool to the stack and selecting only the highlight tones, so the blurring only applies to the lightest highlights. Some adjustment of the selection range is necessary to smoothly integrate the blurring at the edges and soften highlights into gray tones. This is very similar to the technique of blurring the image slightly in "glamor shots" to give the image an ethereal look.

Thanks to dpreview.com forum members (Nik121 and gollywop) for their comments.

My attempt at defining some unfamiliar terms.

Blown Highlight. A highlight without detail or tonality, which equals or exceeds the white level of the capture. People find highlights with detail or gradient more attractive and less distracting. A "hot" highlight is one that appears blown or is distracting, but not actually blown.

Specular Reflection. A point source of light or bright reflection, such as the crest of a wave, which does not contain any detail. These highlights will always be "blown" as they should be, represented by white in the image. A point source may not be a reflection, but for all practical purposes it falls under this category in photography.

Highlight Recovery. This is where extra information the camera captured in the exposure is used to reconstruct highlights. Ordinarily, this can only be taken so far, and anywhere detail was not captured in the original raw image, the highlight will be replaced by gray.

Captured Image. This is the raw image data, which cannot be viewed as an image without development.

Developing. The raw image data is converted into a viewable form through development, which either happens at the computer or in the camera through a JPEG engine.

JPEG Engine. The hardware and software the camera uses to develop the captured image in the form of raw data into a usable image in the JPEG format.

Open Flash Charts

I recently discovered a wonderful new open source project for creating Flash charts. It is open source, non-proprietary and best of all for a non-profit on a tight budge, it is free. In the last week I deployed Open Flash Charts after integrating the package into our Folkstreams content management system. For users of our system (through their personalized area My Folkstreams), this will be a great improvement in the quality of charts. We make the statistics on visitors and video views available to filmmakers, and the Flash charts are simply beautiful compared to our old ones based on phplot.

You can download the code for OFC (Open Flash Charts) from their homepage. It is the work of John Glazebrook and he must be a designer, because the default charts in the tutorial are beautiful and take advantage of the interactive features of Flash. I discovered a few kinks that need working out, but overall, this is an excellent addition to the open source code making up the Folkstreams platform.

Get off my duff? Forget it.

The strangest thing about the web is that I no longer have to look for things. I can just Google them up quicker than I can locate them myself. When I wanted to know the guide number for my ancient Vivitar 215 flash, instead of spending an hour looking through my boxes of equipment stored about the house to find the original manual, I was able to find a scan of the manual online using Google. When I had a question about my camera, the manual was sitting five feet from me on the shelf, but I decided to Google for it. In less time and effort than getting off my duff and tugging it out of my bookshelf, I had the PDF file from the camera manufacturer's site open in my browser.

I even had the PDF file on my local hard disk, but it was lost among the ten thousands PDF file's I've downloaded over the years. So it was quicker to stay in my browser and use the power of Google search to find the manual than to use the obsolete and clunky interface to my information the Windows Explorer offers. This is a new age dawning. There is something significant in this. I just don't know where its going.

It is more efficient, easier and simpler for me to find a manual by finding its Doppelganger halfway around the world through thousands of miles of telephone and network connections, routers and gateways than it is to look in a box ten feet away from me in the closet for the original dead trees version.

Stranger, is that I don't have to worry that my ancient manual won't be there. I can rely on the assumption that if I am interested in it, someone, somewhere out there has taken the time to scan it and float it onto the network. And it is there. Just Google for a Vivitar 215 flash. What does it mean when I can rely on a "smart mob" acting just like me to ensure whatever I want or need will be at my fingertips because I know our needs are similar?

14 December 2007

What bokeh can tell us about art.

The study of bokeh tells us something vital and fascinating about art. In an article on bokeh, the author draws attention to the "cat's eye effect" where out of focus highlights appear as ovals with sharply converging corners (the shape of a cat's eye). The concept is illustrated with a photograph by Edo Engel of a street sign surrounded by what appears to be a whorl of cat's eye highlights. The author states "When there are many OOFH's [out of focus highlights] scattered across the frame, the cat's eye effect yields the impression of a rotational background motion." (http://www.vanwalree.com/optics/bokeh.html 2007)

What is fascinating about this, besides an interesting effect?

We learn from this phenomena the elongated shape of the disks implies movement. Any elongated shape implies motion, which automobile designers take great advantage of. This sense arises from how the human visual perceptual system interprets elongated forms. These forms, for some evolutionary psychological/perceptual reason, suggest movement.

The phenomena is used by the artist to create a meaningful statement, by understanding the effect the disks have on the human perceptual system, by known that human beings seeing the elongated disks will experience a sense of movement, that the sense of movement can be introduced into an otherwise static and flat scene. The sense of movement, further, can be used to evoke metaphor, as the photographer does by placing the sign for "Wildey" street in the center of motion, the name suggestive of "wildness."

This shows how an astute and observant artist can discover and use perceptual effects. To do so requires an approach to art that does not just involve knowledge of rational qualities and effects, but also irrational effects, which can give rise to apparently rational perceptions and ideas. It tells us that one cannot just "think" one's way to artistic success but must take into account perceptual and psychological effects, which are entirely irrational. It does not make any sense that out of focus ovals impart in the viewer a sense of motion, but nevertheless, they do.

In a second example, the author discovers color harmonies between the background out of focus highlights and foreground out of focus flowers. No viewer would ever think about this, but they do experience it perceptually, which sneaks into and influences our thoughts and feelings about the scene unconsciously.

It is the astute artist who looks for and discovers perceptual effects, which are frequently seen as defects, and employs them to produce meaningful art works.

There is a long history in music of categorizing sounds according to their "consonance" or "dissonance" with the former being pleasing and the latter unpleasant. However, the perception and aesthetic usefulness of consonant or dissonant sounds depends on their context. The same is true of bokeh. There is no good or bad bokeh, and I would hazard that "pleasant" and "unpleasant" is too limiting a range. The use of bokeh depends on the subject. I have seen extraordinarily beautiful photographs exhibiting the double line "nisen-bokeh" effect where the out of focus double bands echo the straight repetitive forms of conifer leaves.

12 December 2007

Cicada song

Cicada song
in the leaves—between the
roots generations.
-sek, Dec 2007

06 December 2007

reFrame : Yet Another Photo Sharing Idea

Here is an idea I had recently for a new photo sharing application, which would make it easier for anyone to use photographs in their own context. A site is created where users can sign up. They submit the name of their Flickr photostream. The site pulls in any photos from their stream that have rights set to Creative Commons remix license. Any user of the site can select any image pulled from the users Flickr photostreams collectively. There could be a single photostream "lightbox" used to select images from, I'm not going into details here.

The idea is to let any user "reframe" any image contributed to a pool of images by other users. Reframe means to give the image another context. For example, an expert on historic photographic processes might frame an image with a text explaining the history and chemistry of the process that made the photograph and how to identify an example of this type of photographic image. A family historian might frame the same image with the biography and family history of the subjects in the photograph. A single image has a potentially unlimited number of contexts or "frames." The system would allow anyone, in the style of a wiki, to "reframe" any image.

Users of the site would have to agree that others can place their images in any possible context, possibly unintended or unflattering, which is why there is a requirement for the non-commercial remix license. Of course, you can do this already, but I do not think there exists and application that makes this easy and puts it all in one location.

This might be combined with my idea for a photo wiki system that encouraged the "quick-slow" process enabled by the so-called bliki, where the same contextual system could allow a quick caption when the image is posted and later more sophisticated commentary and use of the image would follow by creating "pages" associated with the image.

One might object, saying that anyone is free to combine images and text if there were a word processor style system that allowed images to be freely dropped into text anywhere. But the web has shown that it is better to provide a system that structures content and interaction as it being created (wiki allows this process to be continuous). This is where the quick and easy part comes in...it is not so easy to arrange photos and text with a word processor. You do not see many people using a word processor instead of a blog or photo sharing site, although they could create richer documents and post them to their own website using today's word processing applications.

I wish archives and institutions would catch on to the power of reframing images in their collections using contexts contributed freely by users. The academics, visitors, people on the web, anyone should be able to frame images of artifacts or media artifacts themselves, historic photos, old films, video, etc. to create the richest possible understanding of the holdings. And make both the artifacts and knowledge about them more accessible.

I'm thinking of grabbing phpflickr, Dojo and Codeigniter and putting this together, but with the work on Foody and Folkstreams, I really have limited time. Steal this idea, please.

05 December 2007

From the Brother's Grimm on DVD

My friend and partner in developing Farm Foody and project director of the Folkstreams project, Tom Davenport, has opened a store for his From the Brother's Grimm series of films for sale direct to individuals (for institutional use, see his Davenport Films site). Tom is a farmer and filmmaker in Delaplane, Virginia.

The films were frequently featured on PBS in the local D. C. area, so they should be familiar to a generation of children who are now adults. They are live action retellings of classic folk tales in an American setting. Some tales are from Appalachia while others are interpretations of European folk tales with strong overtones of Appalachian culture and setting.

Willa, a favorite, draws upon traditional medicine show culture, documented in films like Free Show Tonight available for anyone to watch on the folkstreams.net website. Mutzmag
is a powerful film in an Appalachian setting, which contains a fair amount of traditional fairy tale violence, but the lessons are appropriate given the dangers children face today. Perhaps they could learn a few survival lessons from Mutzmag's clever outwitting of the ogres and other less than savory inhabitants of the forest, who have designs for her.