 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
| PhotoKB HomeDiscussion Groups Digital Photography Digital PhotoDSLR CamerasZLR CamerasPoint & Shoot CamerasFilm Photography 35 mmLarge FormatMedium formatDarkroomFilm and LabsOther EquipmentPhoto Technique Nature PhotographyPeople PhotographyTechnique GeneralGeneral Photo Topics General TopicsAustralian PhotographyUK PhotographyDirectoryPhoto Clubs |
Photo Forum / Film Photography / Medium format / December 2003Why do pictures appear sharper than they should?
There is a never ending discussion of resolution vs print size and capture media. The mathematics and empirical testing usually show that the usual expected resolution is in the range of 40-60 lines per mm. A good print should have 6 to 8 lpm. So simple logic means that the maximum enlargement should be 5 to 8x. Thus the best size print one could expect from 35mm would be on the order of 8x12 inches with correspondingly smaller sizes from digital cameras. In spite of this people still get prints that are "sharp" with much larger magnifications. Personally, I've been scanning 35mm color negative film with the new Minolta 5400 lately and can print out inkjets that look "sharp" all the way up to the 18x maximum scanning resolution. I'm not one of those who doesn't know what a "sharp" print looks like either, since I use formats all the way up to 4x5. So what's going on? My conjecture (a theory in progress): For people pictures shot at normal distances we are used to seeing detail only in limited areas of the face such as the eyes (lashes and reflections in the pupil) and perhaps loose strands of hair. For landscapes and the like, we can't see all that much detail in distant leaves and grass, but we do see bare branches, telephone wires and the like as sharp. For buildings and other man made structures the detail is seen in the building edges and things like window frames. In all cases the "sharp" things are not those with a lot of fine detail, but rather those with good edge contrast. In other words acutance. Most digital processing involves a certain degree of sharpening. This doesn't do much for real detail, but does increase acutance. This makes those features that we search for in real life appear "sharper" so we read the image as being sharp. We don't expect to see much fine detail so we are not surprised when it is lacking as long as those sharpness indicators have good edge definition. There are categories of images where the detail is important such as scanning electron microscope images and we always comment on how much detail we see in them when viewed. This shows that we don't normally expect to see the fine structures in an image. So I'm guessing that since the images conform to our expectations from viewing such scenes in real life we accept them as sharp even though the resolution figures would indicate that they are not really that detailed. As I said, a theory in progress, comments welcome..  SignatureRobert D Feinman Landscapes, Cityscapes and Panoramic Photographs http://robertdfeinman.com mail: robertdfeinman@netscape.net > [...] > As I said, a theory in progress, comments welcome.. Theories are good, entertaining and fulfilling to pursue. A great deal of research has been published on the subjects of the physiology and psychology of human vision. The US government funded an enormous amount of it when I last cared to look in the seventies. Your local university probaby subscribes to the expensive online databases and archives of these work. > In all cases the "sharp" things are not those with a lot of fine detail, > but rather those with good edge contrast. In other words acutance. > Most digital processing involves a certain degree of sharpening. This > doesn't do much for real detail, but does increase acutance. At the expence of smoothness of tonality and increased noise from my experience. This is the same thing rodinal developer does with B&W film, increased edge contrast at the expemce of grain. Some people are more interested in "sharpness", other prefer smooth tones and lack of noise. To get both requires a larger negative or more clean info in a digital file to print from. Increased contrast (or lack of) is something many people don't consider when choosing optics. IMHO contrast is probably at least as important as actual resolution as a higher contrast lens will appear sharper than a lower contrast lens even if they actually are resolving the same lpmm. I've found this is what makes most prime lenses appear better than zooms more so than actual resolution. -- Stacey > Increased contrast (or lack of) is something many people don't consider when > choosing optics. IMHO contrast is probably at least as important as actual > resolution as a higher contrast lens will appear sharper than a lower > contrast lens even if they actually are resolving the same lpmm. Exactly. In other words, image quality is better measured by MTF at mid-frequencies than by counting angels on heads of pins (looking at frequency components that have magnitudes too small to contribute to image quality or be used for practical imaging). > I've found > this is what makes most prime lenses appear better than zooms more so than > actual resolution. Sounds quite likely. David J. Littleboy Tokyo, Japan Much is made of contrast and acutance (Mackie line) in explaining the apparent sharpness of images. Most of it is right. But consider the appearance of a single line with a width less than the resolution capability of the system. You still see the line. It may have less contrast and definition than it ideally would, but there is a visible line. The mind further processes this data to let you know that it is a line and to associate it with the image context. The situation with two closely spaced lines is not much different. Below the resolution of the system, they become blurred together, but you can still distinguish two lines even a quite a bit beyond the system resolution. The mind will process the blurred image and convey the message, even though the details are not clear. > There is a never ending discussion of resolution vs > print size and capture media. [quoted text clipped - 44 lines] > > As I said, a theory in progress, comments welcome.. quoting Robert F's interesting posting: In all cases the "sharp" things are not those with a lot of fine detail, endquote: I don't agree that _all_ sharp things must lack fine detail. A sidelighted shot of peeling paint with cracks on a barn might have a lot of high contrast (lighted/shadow) in the wood grain and peeling paint edges, but also have lots of fine detail in the pattern of the cracks and surface of the paint. A macro shot of a bug could have lots of fine detail in the body hairs of the spider, say, but also lots of sharp black/white transitions due to lighting of the bug hairs etc. A model's hair can be both sharp and have lots of fine detail, ditto cracked skin on hands etc. Further, most photo lenses are compromises which try to maximize both resolution and contrast, not one or the other. Just look at the modern photo tests which did both resolution and contrast tests; lots of lenses were delivering over 50% contrast levels at high resolution values too. And I quoted MTF% charts for three B&W films which had 50% to 65% MTF contrast response at over 100 lpmm. I _DO_ think that Robert F. is on to part of the answer why digital does so well against film, namely image processing. The frequent use of sharpening filters and masks serves to boost the apparent image sharpness at edge transitions. This is one of the reasons that grain clusters get enhanced during scanning over what I see on the raw films, because of the image processing going on in the scanner (and also digicams) and software. This edge enhancing (sharpening) effect offsets the effects of lower inherent resolution (many digicams at 50 lpmm Nyquist limits etc.) and smoothing by software interpolation. The digital images also get rid of lots of color data, but our eyes are tolerant of those defects (i.e., they are a well chosen tradeoff ;-). So digital images look better than they should thanks to image processing, and esp. sharpening effects, as Robert F. noted. Conversely, film images which are scanned look worse, because sharpening effects highlight grain defects which are missing in the direct capture digicams (and of lessor impact in chemical prints made without image processing). regards bobm Signature*********************************************************************** * Robert Monaghan POB 752182 Southern Methodist Univ. Dallas Tx 75275 * ********************Standard Disclaimers Apply************************* >I _DO_ think that Robert F. is on to part of the answer why digital does >so well against film, namely image processing. The frequent use of >sharpening filters and masks serves to boost the apparent image sharpness >at edge transitions. This is one of the reasons that grain clusters get >enhanced during scanning over what I see on the raw films, because of the >image processing going on in the scanner (and also digicams) and software. Image processing is what makes the human vision system so marvelous. As sensors go, our eyes aren't all that keen. But the processing beats all <G>. I think you overstate the presence and significance of sharpening in digital capture, at least on the professional level. When USM is overdone, or done poorly, it always shows. Unsharp masking is often the subject of one or more chapters in any good Photoshop book, and there are dozens of variations on the the theme. Those who've invested in a dSLR and truly care about their images will use USM knowingly and sparingly, and will generally avoid in-camera sharpening if possible. rafe b. http://www.terrapinphoto.com > >I _DO_ think that Robert F. is on to part of the answer why digital does > >so well against film, namely image processing. The frequent use of [quoted text clipped - 25 lines] > rafe b. > http://www.terrapinphoto.com I don't know anything about how digital cameras process the captured data, but I think there is something being done to all images, even those delivered "raw". After all, the data needs to be averaged (or something) to eliminate the color filter pattern at a minimum. As for film and scanning, I've found that even with the Minolta 5400 at maximum resolution I can improve image "sharpness" by applying a modest unsharp mask. Another trick I've taken to using is to select areas of uniform color with little detail (say a white painted church wall, for example) and apply a 2 pixel gaussian blur. I do this as well to large areas of sky. By selecting using the color range I avoid picking up the edges and the slight blur mostly eliminates the grain from those areas in which it is most apparent. The ultimate aim is to produce an image in which the photographic artifacts don't interfere with the aesthetic objectives. We should all be grateful that digital processing has opened up this new array of tools for us. Several of my tips highlight this, for example, selectively adjusting the color of shadows. A process that would be extremely difficult using conventional enlarging. One small point, I did say this in the original about fine detail: "There are categories of images where the detail is important such as scanning electron microscope images and we always comment on how much detail we see in them when viewed. This shows that we don't normally expect to see the fine structures in an image." I think macros of insects and the like can be included in that category and thus the viewer's expectations for sharpness may be higher than for normal pictorial subjects. SignatureRobert D Feinman Landscapes, Cityscapes and Panoramic Photographs http://robertdfeinman.com mail: robertdfeinman@netscape.net > [...] > "There are categories of images where the detail is important such [quoted text clipped - 5 lines] > and thus the viewer's expectations for sharpness may be higher than for > normal pictorial subjects. Those electron microscopes actually produce rather low resolution images. SEM images do not (!) appear sharper than they should be. These microscopes magnify a lot, yes. And they have enormous depth of field. And that's where the "how much detail we see" comes from: it's finer detail than you would be able to see any other way. They show "fine structures", i.e. structures of (sub)microscopic scale. But don't let that fool you into thinking that the images they produce themselves are high resolution, have lots of detail. They are not (!) sharp, not very detailed at all. "Qualitywise" SEM pictures still are at the level of webcams, or mediocre amateur video. They use a rather modest number of scan lines, and on any one of these the number of sampled points is rather limited too. And that coarse scanning structure is what these images consist of. Not "fine structures". So be warned: "fine structures in an image" can mean different things, and it can lead to all sorts of misunderstandings when you start confusing these different meanings. |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2008 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.