Skip to navigation – Site map

HomeNuméros9DossierSharp Images and Unsharp Masks

Dossier

Sharp Images and Unsharp Masks

Images nettes, masques flous
Till A. Heilmann
Translation(s):
Images nettes, masques flous [fr]

Abstracts

Today, almost all images, from smartphone photos to magazine ads, are algorithmically sharpened, redefining aesthetic standards and altering our perception of what constitutes a “good” image. Sharp is the visual default of everyday photographic culture. This article explores the genealogy of unsharp masking, a method for enhancing images popularized by Adobe Photoshop. Originally developed as an “analog” technique to prevent loss of information in reproducing high-contrast originals, unsharp masking has shifted from preserving details to increasing acutance—i.e., perceived sharpness. This transition reflects its transformation from a specialized tool in fields like X-ray medicine and astrophotography to a universal optimization process in the digital image pipeline. Normalized and largely automated, unsharp masking profoundly influences visual culture.

Top of page

Full text

  • 1 See Andrey Ignatov et al., “Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, (...)

1Sharpening photographs is a routine operation in digital image processing, whether performed on a smartphone using an app like Instagram (fig. 1) or on a desktop computer with professional software such as Adobe Photoshop. Although the recent rise of deep neural networks for image synthesis has introduced new methods for “improving” pictures by upscaling and adding detail to blurry or low-resolution images, traditional sharpening techniques continue to dominate contemporary visual culture.1

1. Sharpening filter on Instagram, January 31, 2024

1. Sharpening filter on Instagram, January 31, 2024

Screenshot, Apple iPhone SE (2022).

2This essay examines what is arguably the most common digital sharpening method: an algorithmic filtering operation called, somewhat misleadingly, unsharp masking, which was popularized primarily by Adobe Photoshop (fig. 2). Tracing its genealogy back to the 1930s, we can discern a momentous shift in the application of unsharp masking: from a logic of preserved details to a logic of perceived sharpness. Initially devised and used as a technique to prevent the loss of information when reproducing high-quality originals, unsharp masking eventually became the standard method for accentuating contours to “add” sharpness to a picture. It evolved from a specialized tool for specific use cases in fields like medicine, astrophotography, and the prepress industry into a universal optimization procedure applied to all digitized or digital pictures. Parallel to its algorithmic implementation, unsharp masking has undergone a threefold transformation: its reduction to a single purpose (namely, the enhancement of edges); its automation as a standard component of the digital image pipeline; and its normalization across the entire field of photography. As a result, unsharp masking has had a profound impact on today’s visual culture. Almost every photo we see, from our own smartphone shots to advertisements in glossy magazines, has been algorithmically sharpened. The shift in unsharp masking has thus helped redefine the aesthetics of photographic images. With sharpened images having become the norm, our viewing habits, expectations, and ideas of what “good” images look like have changed accordingly.

2. Dialog box for Adobe Photoshop 1.0 “Unsharp Mask” filter, January 31, 2024

2. Dialog box for Adobe Photoshop 1.0 “Unsharp Mask” filter, January 31, 2024

Screenshot, Apple Macintosh Quadra 900 with operating system 7.5.3.

Digital Unsharp Masking

  • 2 See Wolfgang Ullrich, Die Geschichte der Unschärfe (Berlin: Klaus Wagenbach, 2002); Pauline Martin, (...)

3The sharpness of photographic pictures (or lack thereof) has been a hotly contested issue since the beginnings of the technology. It was precisely the impartial distinctiveness and mechanical precision with which photography could render reality that set it apart from other visual arts like painting. In fact, many early critics and practitioners of the medium saw the technology’s inherent “realism” in depicting even the smallest detail as a defect that had to be artistically corrected for photography to achieve the status of a proper art form. Ever since, sharpness—along with its opposite—has served as an ideological touchstone in debates about the specificity and aesthetics of photography.2

4But what does “sharp” mean? And what exactly happens when an image is sharpened, using algorithms or other means? If we disregard artistic considerations and subjective value judgments, then sharpness can be understood as a function of two properties of an image: resolution and acutance.3 Resolution indicates how finely a picture represents details; for digital images, resolution can be easily defined as the number of pixels with a specific color and brightness value. Acutance, on the other hand, refers to the contrast at the edges in an image, usually at the boundaries between objects and their surroundings. The steeper the transition from bright to dark areas, the more clearly individual parts of an image stand out and the higher its acutance. “Soft” images are generally low in contrast, while sharp images have strong edge contrast. And just like resolution, acutance can be measured objectively using a formula that calculates the steepness of the density change at object edges.4

  • 5 Consequently, upscaling images with AI techniques like “super resolution” should not be regarded as (...)
  • 6 Adobe Systems, Adobe Photoshop User Guide (Mountain View, CA: Adobe Systems, 1990), 306, https://ar (...)

5It is crucial to understand that sharpening, regardless of how it is achieved, only ever affects the acutance of an image, never its resolution.5 Techniques such as unsharp masking are methods for controlling contrast to increase the perceived sharpness of edges in a picture. This effect is demonstrated by figure 3. The upper half of the illustration features a low-contrast transition from light background to dark spiral pattern. The lower half shows the same pattern after unsharp masking was applied. The steepness in change from background to foreground has been increased by darkening the edge on the inside of the spiral and making it brighter on the outside. As a result, the spiral’s border appears sharper, and it stands out more clearly. The same effect can be seen in figure 2: the right-hand side of the image has been processed with Adobe Photoshop’s Unsharp Mask filter. The sky above the trees is brightened, while the trees are darkened. In fact, the filter has been applied so strongly here that the strip of sky above the treetops seems to glow. How is this effect achieved? The user manual for the first version of Photoshop from 1990, the very one shown in figure 2, provides a fairly precise description of algorithmic unsharp masking.6 In short, the program first generates a blurred copy (depending on the radius specified by the user) and then “subtracts” a fraction of this copy (depending on the amount specified), known as the eponymous unsharp mask, from the original image. This process makes edges appear more distinct, resulting in a sharper image.

3. Nevit Dilmen, unsharp mask, April 14, 2008

3. Nevit Dilmen, unsharp mask, April 14, 2008

6Already with its very first release, Photoshop had implemented algorithmic unsharp masking in an almost exemplary manner. The sole change to the filter, made only a few months later with bug fix release 1.0.7 in 1990, was the addition of a third parameter called “Threshold,” allowing users to control the filter more precisely. Since then, the unsharp mask feature has remained unchanged in Photoshop, at least in terms of its basic functionality, up to the current version of 2024.7 It continues to be the preferred tool for sharpening images in PC software for professionals—not limited to Adobe Photoshop alone. Today, all major competitors also incorporate this feature, including the open-source solution GIMP, Affinity Pro, the web application Photopea or Apple’s Core Image framework for macOS and iOS. However, when Adobe first launched Photoshop in 1990, the filter served as a unique selling point for the software. None of its now forgotten alternatives seems to have offered it at the time. The primary competitor, ColorStudio from Letraset, only provided two simple and fully automated filters called “Sharpen” and “Sharpen More”.

“Analog” Unsharp Masking

  • 8 Adobe Systems, Adobe Photoshop User Guide, 306.

7While its availability to PC users may have been unusual at the time, the method of unsharp masking was by no means new or unknown, as Photoshop’s user guide makes clear: “The filter is commonly used in pre-press production to enhance details in the separations by producing exaggerated density at the borders of a color change.”8 Unsharp masking was, in fact, quite common around 1990 and even before—just not in personal computing, but rather in the printing industry. Furthermore, it served a different purpose. Not aimed at optimizing photographs for acutance, it was actually used to address a rather technical difficulty: reproducing correct color borders in CMYK printing. Making images look sharper seems not to have been the primary, and certainly not the only, goal of unsharp masking in the late 1980s. Instead, the problem that it addressed was the reproduction of images (such as color photographs rendered in halftone printing), rather than a lack of sharpness in the images themselves.

8To delve deeper into the aesthetics of sharpness in contemporary visual culture, it is informative to trace the genealogy of unsharp masking back beyond digital implementations for image processing. What constituted unsharp masking in “analog” times? How was it done before images could be computed at the level of individual pixels? To commence our exploration, let us examine some seminal works on photography from the mid-twentieth century. The Focal Encyclopedia of Photography from 1956 features a comprehensive chapter on masking, providing a detailed description of the process:

  • 9 Purves, Focal Encyclopedia of Photography, 705.

A negative may have a good tone range, but too high contrast for convenient printing on the required material. The way to solve the problem is to make a very weak and soft, but full-toned, positive contact print from the hard negative on another plate or film. When dry, this is bound into contact in accurate register with it. The mask will then tend to cancel out the negative and it will have the effect of considerably reducing the contrast of the negative, which may be very useful.9

9The similarity to the operation of the digital filter is evident. The photomechanical process also begins with the creation of a soft or “blurred” copy of the original image; this copy is then aligned, or set in register, with the original, and subsequently a new image is produced by exposing the combined unsharp mask and original. Depending on its density, the mask selectively reduces the amount of projected light, thereby acting as a control in the generation of the new image.

10As the passage quoted above illustrates, the objective of unsharp masking was not solely to sharpen images but rather to reduce their tonal range to ensure faithful reproduction. The range of a film negative is often so extensive that fine details in dark or light areas of the image can be lost during printing. In other words, unsharp masking was not employed exclusively or even primarily to enhance local contrast—i.e., to make edges appear sharper. Its main purpose was to reduce the global dynamic range of images. The issue to be addressed was not a lack of sharpness in the images themselves but rather their reproduction for other media, their transition from film to print, from developed negative to ink on paper.

  • 10 Purves, 706.
  • 11 C. N. Nelson, “The Theory of Tone Reproduction,” in The Theory of the Photographic Process, ed. Cha (...)

11Nevertheless, the sharpening effect achieved by unsharp masking was already well known back in the 1950s. As explained in the Focal Encyclopedia: “It is claimed that unsharp masks improve the rendition of detail, which may seem a contradiction. . . . In practice the visual effect is a general one of apparently greater sharpness.”10 What is intriguing here is the qualifying phrase “It is claimed,” suggesting that the visual effect was sometimes misunderstood or incorrectly explained. Hence the remark that sharpness is, in fact, only “apparently” increased. A cautionary assessment of the process is also provided by another reference work on photography published a decade later. The Theory of the Photographic Process from 1966 warns that increasing edge contrast is not necessarily beneficial to the picture’s quality: “A disadvantage that arises from the use of the mask is that the large ‘uniform’ areas in the print are not actually uniform [in the negative], and the abrupt edges have a dark band on one side and a light band on the other side. Some observers decide that these features detract from the quality of the print.”11

  • 12 Andreas Feininger, The Complete Photographer (Englewood Cliffs, NJ: Prentice-Hall, 1965), 158; Anse (...)

12Reviewing the literature, it becomes evident that unsharp masking was not a technique commonly used by photographers prior to its algorithmic implementation for PCs in the 1990s. In fact, prominent textbooks from before this era make no reference to the process whatsoever. For instance, neither Andreas Feininger’s widely read and repeatedly reissued The Complete Photographer nor Ansel Adams equally popular The Print discuss unsharp masking. Adequate sharpness of the negative and accurate reproduction of the tonal range were simply taken for granted.12

  • 13 John A. C. Yule, “Unsharp Masks and a New Method of Increasing Definition in Prints,” The Photograp (...)
  • 14 Yule, “Unsharp Masks,” 321–24.
  • 15 Yule, 321.
  • 16 Yule, 322.

13If we follow the references to unsharp masking in the professional literature, we can trace the technique back to an article published in 1944 in The Photographic Journal by the Royal Photographic Society. This text, authored by British scientist John A. C. Yule, stands as the earliest known treatise on the process in the English language. In his article titled “Unsharp Masks and a New Method of Increasing Definition in Prints,” Yule gives an extensive account of the process.13 In addition to explaining the usefulness of unsharp masking for controlling the tonal range of images and for color correction, Yule demonstrates the enhanced quality in the reproduction of line drawings (as in topographical maps) and even presents some stylistic experiments, such as the transformation of photographs into pencil-like drawings (fig. 4).14 He also remarks that unsharp masking can render pictures with greater clarity: “Surprising though it may seem, the sharpness of detail in a reproduction can be improved by the use of masks in which the images are not sharp.”15 But, again, this effect is not discussed as being the technique’s main purpose by Yule: “When the correction of colors or values is not required, it is not usually justifiable to use masks solely for the sake of improvement in sharpness achieved by making masks unsharp.”16 The increased acutance seems like an unexpected, albeit not unwelcome, by-product of the process. Yule’s primary intention, however, is to facilitate the reproduction of photographs and other images for various media.

4. Artistic effect of converting a photograph into an image that resembles a drawing using an unsharp mask

4. Artistic effect of converting a photograph into an image that resembles a drawing using an unsharp mask

Illustrations taken from John A. C. Yule, Unsharp Masks and a New Method of Increasing Definition in Prints, “The Photographic Journal” 84 (1944): 324.

© Michael Pritchard / The Royal Photographic Society, Bristol, UK

  • 17 Yule, 327.

14From a historical perspective, the most intriguing aspect of Yule’s article is not the text itself but an appendix published alongside it. This appendix summarizes the discussion of the article at a meeting of the Royal Society’s Scientific and Technical Group in the spring of 1944. In it, we find a statement by a certain Dr. Spiegler, who claimed to have conducted his own experiments and gained experience with unsharp masking even earlier.17 Who was this Dr. Spiegler?

15Gottfried Spiegler was a Jewish medical physicist from Berlin who, in the 1930s, had worked as the director of the X-Ray Research Institute in Vienna.18 After the annexation of Austria, Spiegler fled to England in 1939, where, after facing many challenges including internment as an “enemy alien,” he eventually secured a position at the Royal Marsden Hospital in London. At his institute in Vienna, Spiegler had worked with X-rays. A recurring problem with such images was their high density. When reproducing X-ray images in print, diagnostically important details would often be lost in the very dark or very light areas. To address this issue, Spiegler, together with his Galician laboratory technician Kalman Juris (who was later murdered in Auschwitz), developed a method of controlling the density range in copies of X-ray images. This method was essentially the photomechanical process of unsharp masking, although Spiegler and Juris never used the term.

  • 19 Gottfried Spiegler and Kalman Juris, “Ein neues Kopierverfahren zur Herstellung ideal harmonischer (...)

16At the beginning of the 1930s, Spiegler and Juris published several articles on their innovation in scientific journals.19 The first text from 1930, in particular—bearing the somewhat laborious title “A New Copying Process for the Production of Ideally Harmonic Copies from High-Contrast Negatives”—is a remarkable document. It not only contains the oldest known description of unsharp masking but also features extensive reflections on imaging technology in general, on the relationship between image and object, and on the question of image manipulation. However, with respect to the genealogy of the method, the main point is that Spiegler and Juris are very clear about their intentions and the purpose of their invention:

  • 20 Spiegler and Juris, “Ein neues Kopierverfahren,” 518 [translated].

Considering the great effort, time, and worry that doctors and staff in many X-ray institutes have to put into producing good copies from high-contrast negatives, the method described here should represent a step forward. What was previously left to individual skill has now been put on a technical basis; since the method permits repeatability [Reproduzierbarkeit], it can be regarded as objective.20

17Spiegler and Juris aimed to improve the reproduction of images, not their sharpness. In its earliest days, unsharp masking was focused on ensuring the visibility of fine details and structures in the very light and very dark areas of X-ray images for diagnostic purposes rather than on achieving higher acutance or sharper edges (even if these occurred as a by-product of the process).

Algorithmizing Unsharp Masking

18Having reached the chronological and technological “zero point” of unsharp masking (as far as our current genealogical knowledge goes), let us reverse our course and trace the evolution of the process back to the present in fast forward. How did the algorithmic implementation of unsharp masking come about, and what changes have occurred in its application?

  • 21 Mary Johnson, “Use of the Herschel Effect in Improving Aerial Photographs,” Journal of the Optical (...)

19The literature provides ample evidence that unsharp masking quickly became well known at the end of World War II. It was discussed and experimented with across all scientific, technological, and industrial fields concerned with imaging and the reproduction of images, including aerial photography and cartography.21 Spiegler and Juris had registered patents for their photomechanical process in Germany, Austria, and France as early as the 1930s. Yule did the same as an assignor for Kodak in the USA in the 1940s.

  • 22 Johnson, “Use of the Herschel Effect”; E. Zieler and K. Westerkowsky, “The Ampliscope, an Experimen (...)
  • 23 Dwin R. Craig, “The LogEtron: Fully Automatic, Servo-Controlled Scanning Light Source for Printing, (...)

20With the rise of electronics, particularly through the new medium of television, and with the emergence of digital computers, first attempts were soon made to base the method on the firm mathematical foundations of signal processing and information theory. Although photomechanical unsharp masking produced good results in the darkroom, it was often described as “time consuming and expensive” and “rather complicated.”22 Electronics, and later computers, promised to simplify and improve the process. A first patent for “Unsharp Mask in Electronic Color Correction,” applied for in 1948, was granted to TIME Inc. in the USA in 1952. At the beginning of the 1950s, engineer Dwin R. Craig developed the LogEtron, an early device for the automatic control of contrast using electro-optical unsharp masking. The LogEtron apparently found its market and even spurred on some imitators, particularly in the field of medical imaging23—the stated goal being, once again, to preserve diagnostically important details in copies of X-ray images.

  • 24 Leslie S. G. Kovásznay and H. M. Joseph, “Image Processing,” Proceedings of the IRE 43, no. 5 (1955 (...)

21In fact, unsharp masking played a central role in the newly emerging field of image processing during the 1950s and 1960s. One of the seminal texts, simply titled “Image Processing,” used the well-understood photomechanical technique as a primary test case to develop a “method for processing pictures by electronic techniques.”24 Over the course of several pages, the article translates unsharp masking (discussed as “contour enhancement”) into algebraic formulas. Craftsmanship and experience in the photo lab are reformulated as mathematical descriptions of image parameters and operators (fig. 5). Practice is turned into algorithms.

5. Mathematical and electronic execution of an unsharp mask

5. Mathematical and electronic execution of an unsharp mask

Illustration taken from Leslie S. G. Kovásznay and H. M. Joseph, Image Processing, “Proceedings of the IRE” 43, no. 5 (1955): 565.

© 1955, IEEE

22Remarkably, the article barely addresses the visual effect of the method it discusses and algorithmizes. The acutance of images, the judgments of human viewers, and the subjective perception of sharpness play only a subordinate role. The focus is on questions of repeatability and reproduction—i.e., on technical factors such as low bandwidths and low signal frequencies. Improved visibility is mentioned only in passing and, in an astonishing (if not explicit) continuation of Spiegler and Juris’s contribution, it is discussed again in the context of X-ray images:

  • 25 Kovásznay and Joseph, “Image Processing,” 567.

The potential uses of the contour enhancement process may include transmission of an image with less impairment through a lesser bandwidth channel or to pre-emphasize a picture that is to be sent through a low definition channel, in a manner analogous to pre-emphasis in an audio amplifier for the purpose of improving the response at higher frequencies. This apparent improvement may be made on inherently low definition patterns such as an X-ray picture of soft tissues to make the detail more apparent, that is, more quickly visible.25

  • 26 Russell A. Kirsch et al., “Experiments in Processing Pictorial Information with a Digital Computer, (...)
  • 27 David F. Malin, “Unsharp Masking,” AAS Photo-Bulletin 16 (1977): 10–13.

23Although the mathematical foundations of algorithmic unsharp masking and the first experiments in digital image processing date back to the 1950s,26 the process could only be realized on digital computers in the 1970s. The necessary computing and storage capacities were simply too high before integrated circuits were produced at scale. Whenever and wherever the highest possible visual quality was required, photomechanical unsharp masking was still preferred and used until the late 1970s, particularly in astrophotography.27

  • 28 Alan Martin Gilkes, “Photograph Enhancement by Adaptive Digital Unsharp Masking” (master’s thesis, (...)
  • 29 Gilkes, “Photograph Enhancement,” 8.
  • 30 Gilkes, 8–9.

24However, the earliest digital implementations of unsharp masking marked the beginning of a significant shift in the purpose of the method. For example, a mid-1970s MIT thesis on “Photograph Enhancement by Adaptive Digital Unsharp Masking” states that the aim of the study is to clarify “the potential for producing subjectively better output images.”28 The algorithmic optimization of images is now aimed at and based on the viewer’s judgment: “‘Subjectively better’ is intended to embrace both a human observer’s evaluation of the aesthetic qualities of an output image and his evaluation of how much the visibility of useful information is improved by filtering.”29 The study starts from the basic premise that every reproduction of an image, such as an X-ray, implies a potential loss of information in the long chain of transmissions from one medium to another. Digital unsharp masking is supposed to compensate for this loss. And it is in this context that the increase in acutance, the perceived improved sharpness of images, takes on its own aesthetic value beyond the quality even of the original picture: “Edge sharpening is also useful for enhancing images beyond the requirements of mere restoration of original image quality. A human observer will often judge a sharpened descendant image to be more informative and more aesthetically satisfying than the source image from which it came.”30

25Fifteen years later, when Thomas Knoll implemented unsharp masking for his new image-editing program called Photoshop, the tension between optimizing images for reproduction in a different medium and improving them for perception in the viewer’s eye was still minimal. After all, reproducing images digitally was a significant technical challenge at the time. For an image to be edited and processed with Photoshop or another program, it had to be in digital form, which was a problem because digital cameras of adequate quality were not readily available on the consumer market around 1990. Most of the images processed on a computer originated as a print on paper, a film negative, or a transparency. The first link in the chain of operations that transmitted, transformed, and reproduced images was usually an electronic scanner, needed to convert “analog” pictures into streams of data. And owing to their basic working principle, scanners had to interpolate the brightness and color value of each individual pixel from multiple adjacent photo cells. Consequently, edges in scanned images appeared slightly blurred—a result of reproducing pictures in the desired digital format.

  • 31 Adobe Systems, Adobe Photoshop: Classroom in a Book (Carmel, IN: Hayden Books, 1993), 134.
  • 32 Adobe Systems, Advanced Adobe Photoshop: Classroom in a Book (Indianapolis, IN: Hayden Books, 1994) (...)

26Photoshop’s Unsharp Mask filter was specifically implemented to counteract the blurring caused by the initial scanning process through algorithmic sharpening. In other words, the filter’s original purpose was to dis-simulate the digitization of images. Early documentation of the program confirms this. The official Adobe Photoshop textbook, first published in 1993, plainly states: “The scanning process can cause an image to appear slightly out of focus or ‘soft.’ You can sharpen an image using the Adobe Photoshop sharpening filters.”31 The second edition from 1994 summarizes the function of the filter as follows: “The Unsharp Mask filter adjusts the contrast of edge detail, creating the illusion of more image sharpness. This filter can be useful for refocusing an image that has become blurry from interpolation or scanning.”32 Remarkably, Adobe is quite candid about the filter’s visual effect being, in fact, only “illusory”—i.e., increasing perceived sharpness.

  • 33 Kim Baker and Sunny Baker, Color Publishing on the Macintosh: From Desktop to Print Shop (New York: (...)
  • 34 See, for example, the advertisement for Agfa’s StudioScan in Popular Photography (October 1994): 12 (...)

27Other sources from the time describe the situation similarly. For example, a 1992 handbook on the use of image-processing programs for prepress work explains: “Images lose focus during scanning, and this makes them look flat. Scanned images may also lack definition and subtle detail. Because each step in the film reproduction, plating, and the printing process will exacerbate these problems, a filter is used to sharpen parts of the image to make it crisper. . . . In most cases use Unsharp masking (not Sharpen) to increase the focus if necessary.”33 Soon after the initial release of Adobe Photoshop, all quality scanners were sold with additional software to facilitate the post-processing of digitized images with unsharp masking and other filters.34

The Visual Default of Sharpness

28From Spiegler and Juris’s first description of the process around 1930 to the early successes of Photoshop in the mid-1990s, unsharp masking—whether realized photomechanically, electronically, or digitally—primarily addressed difficulties in reproducing images, rather than a lack of sharpness in originals. However, with the adoption of the method in image-editing programs for the consumer market, the filter’s bias, already noticeable at the very beginning of its digital implementation in the 1970s, grew stronger and stronger: algorithmic sharpening of images increasingly aimed to please the human eye. Aspects of reproduction and the technical difficulties of digitization became less and less prominent.

29On the discursive level, the change is most apparent in how unsharp masking was described to the users of image-editing programs. At the end of the 1990s, Dan Margulis, an influential public expert on Photoshop at the time, declared the filter to be a universal tool for sharpening images in his popular Makeready column: “Unsharp masking is an artificial method of making images appear more in focus. It is useful in virtually all graphic scenarios. . . . Whether we are preparing for a photographic print, a large-format output device, a color laser or other digital proofer, a JPEGged file for Web use, or any type of print work, accurate USM is a big deal.”35 Now, unsharp masking was no longer about addressing specific problems or shortcomings of digital technology. It had become a tool for making pictures look sharper—meaning “better”—in all situations and for all possible use cases. According to Margulis, the contours and edges of objects in each and every image are too poorly defined because they are fundamentally—i.e., on the level of physical existence—narrower “possibly even than the film of the original photograph can resolve.”36 A few years later, he reiterated his position on sharpening in general and on unsharp masking in particular more bluntly: “Almost every picture needs it, and not because photographers don’t know how to focus their cameras.”37 Even if an image is perfectly in focus, it still needs to be digitally sharpened (fig. 6).

6. Unsharp mask giving a “better look” to faces

6. Unsharp mask giving a “better look” to faces

Illustration taken from Dan Margulis, “Professional Photoshop 5: The Classic Guide to Color Correction” (New York: Wiley Publishing, 1999), 100.

© Dan Margulis

30The same shift can be seen in Photoshop’s official documentation. As late as 2002, Adobe’s Classroom in a Book guide for the seventh version of the program described the feature as follows: “The Unsharp Mask filter corrects blurring introduced during photographing, scanning, resampling, or printing.”38 Other than the first, the very act of taking a picture, all the operations listed are processes of reproducing images. On the other hand, the current online help for sharpening filters in Photoshop available on the World Wide Web simply states: “Whether your images come from a digital camera or a scanner, most images can benefit from sharpening.”39

31More significant than the discursive normalization of unsharp masking seems to be its digital automation. The filter has long since become an integral part of the image-processing pipeline from the camera to the resulting image file. Whenever a camera outputs images in the JPEG format geared to end users, the corresponding files have already been algorithmically sharpened. Camera manufacturers take into account their customers’ expectations of “crisp” pictures:

  • 40 Rajeev Ramanath et al., “Color Image Processing Pipeline,” IEEE Signal Processing Magazine 22, no.  (...)

The human eye is known to be highly sensitive to sharp edges; we prefer sharp edges in a scene to blurred ones. Specifically, we are more sensitive to horizontal and vertical edges than diagonal ones, and even less sensitive to edges in other directions. Most camera manufacturers use an edge-enhancement step such as unsharp masking to make the image more appealing by reducing the low frequency content in the image.40

  • 41 Canon Inc., Canon EOS-1DX Instruction Manual (2013), 135–36.
  • 42 Wayne Lorimer, “Beware of the Worms: Sharpening Fuji Files in Lightroom,” October 14, 2019, http:// (...)

32Even on high-end cameras for professionals such as Canon’s EOS-1D X, the menu offers an option for “Sharpness,” which can be adjusted as a separate parameter under the “Picture Style” setting.41 And if you choose to “develop” the raw data coming from the camera’s sensor yourself, the software used for processing RAW image files such as Adobe Lightroom or Apple Aperture presents you with a preset for sharpening the picture (fig. 7). The camera system may even over-sharpen a RAW file, producing worm-like artifacts,42 before you get to make any changes to the image (fig. 8).

7. Till A. Heilmann, Sharpening slider in Adobe Lightroom

7. Till A. Heilmann, Sharpening slider in Adobe Lightroom

RAW image file shown by Timo Schemer-Reinhard, June 27, 2024, screenshot, Windows 11.

8. Wayne Lorimer, detail of Home Sweet Home zoomed in at 200%

8. Wayne Lorimer, detail of Home Sweet Home zoomed in at 200%

Photograph taken on a Canon 650D using a Canon EF-S lens.

© courtoisie de l’artiste, Wayne Lorimer.

33Indeed, even the most recent digital camera operates on principles akin to those of electronic scanners from the 1990s. It still needs to interpolate the color and brightness values of individual pixels, potentially resulting in soft edges.43 Consequently, the issues of poor or unfaithful reproduction, which have heavily informed the genealogy of unsharp masking, are inherently embedded in today’s prevailing image technology. However, it seems to be this very universalization of the original issue that has propelled the generalization of its “solution” as the norm in digital imaging. Algorithmically sharpened images have become the visual default of everyday photographic culture.

Top of page

Notes

1 See Andrey Ignatov et al., “Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 Challenge: Report,” in Computer Vision – ECCV 2022 Workshops, ed. Leonid Karlinsky, Tomer Michaeli, and Ko Nishino (Cham: Springer Nature Switzerland, 2023), 92–129.

2 See Wolfgang Ullrich, Die Geschichte der Unschärfe (Berlin: Klaus Wagenbach, 2002); Pauline Martin, Le flou et la photographie histoire d’une rencontre, 1676–1985 (Rennes: Presses Universitaires de Rennes, 2023).

3 Roger Cicala, “Have You Seen My Acutance?,” June 13, 2009, https://wordpress.lensrentals.com/blog/2009/06/have-you-seen-my-acutance/.

4 Frederick Purves, ed., The Focal Encyclopedia of Photography (New York: Macmillan, 1956), 11.

5 Consequently, upscaling images with AI techniques like “super resolution” should not be regarded as sharpening in the traditional sense of the word.

6 Adobe Systems, Adobe Photoshop User Guide (Mountain View, CA: Adobe Systems, 1990), 306, https://archive.computerhistory.org/resources/access/text/2013/01/102640940-05-01-acc.pdf.

7 “Adjust Image Sharpness and Blur,” Adobe, last updated September 25, 2023, https://helpx.adobe.com/photoshop/using/adjusting-image-sharpness-blur.html.

8 Adobe Systems, Adobe Photoshop User Guide, 306.

9 Purves, Focal Encyclopedia of Photography, 705.

10 Purves, 706.

11 C. N. Nelson, “The Theory of Tone Reproduction,” in The Theory of the Photographic Process, ed. Charles Edward Kenneth Mees and Thomas Howard James, 3rd ed. (New York: Macmillan, 1966), 495–97.

12 Andreas Feininger, The Complete Photographer (Englewood Cliffs, NJ: Prentice-Hall, 1965), 158; Ansel Adams, The Print: Contact Printing and Enlarging (Hastings-on-Hudson, NY: Morgan Press, 1968), 95–97.

13 John A. C. Yule, “Unsharp Masks and a New Method of Increasing Definition in Prints,” The Photographic Journal 84 (November 1944): 321, https://archive.rps.org/archive/volume-84/737998.

14 Yule, “Unsharp Masks,” 321–24.

15 Yule, 321.

16 Yule, 322.

17 Yule, 327.

18 Werner Schmidt, “Gottfried Spiegler,” 2021, https://www.oegmp.at/wp-content/uploads/2021/12/Gottfried-Spiegler-Biographie-2021.pdf.

19 Gottfried Spiegler and Kalman Juris, “Ein neues Kopierverfahren zur Herstellung ideal harmonischer Kopien nach kontrastreichen Negativen,” Fortschritte auf dem Gebiet der Röntgenstrahlen 42 (1930): 509–18; Gottfried Spiegler and Kalman Juris, “Ein neues Verfahren zur Herstellung ausgeglichener Kopien nach besonders harten Originalaufnahmen,” Photographische Korrespondenz 67, no. 1 (1931): 4–9; Gottfried Spiegler and Kalman Juris, “Grundlagen des neuen Verfahrens und Vorrichtung zur Herstellung ausgeglichener Kopien,” Photographische Korrespondenz 69, no. 3 (1933): 36–41.

20 Spiegler and Juris, “Ein neues Kopierverfahren,” 518 [translated].

21 Mary Johnson, “Use of the Herschel Effect in Improving Aerial Photographs,” Journal of the Optical Society of America 41, no. 11 (1951): 748, https://opg.optica.org/abstract.cfm?URI=josa-41-11-748.

22 Johnson, “Use of the Herschel Effect”; E. Zieler and K. Westerkowsky, “The Ampliscope, an Experimental Apparatus for ‘Harmonizing’ X-Ray Images,” Philips Technical Review 24, no. 9 (1962–63): 286, https://pearl-hifi.com/06_Lit_Archive/02_PEARL_Arch/Vol_16/Sec_53/Philips_Tech_Review/PTechReview-24-1962_63-285.pdf.

23 Dwin R. Craig, “The LogEtron: Fully Automatic, Servo-Controlled Scanning Light Source for Printing,” Photographic Engineering 5, no. 4 (1954): 219–26; Zieler and Westerkowsky, “The Ampliscope”; Peter Diedrich and Robert Fuhrmann, “Digitale Nachbearbeitung von unter- und überbelichteten Röntgenfilmen mit einem Personal Computer,” Fortschritte der Kieferorthopädie 53, no. 1 (1992): 38–39, https://link.springer.com/article/10.1007/BF02168018.

24 Leslie S. G. Kovásznay and H. M. Joseph, “Image Processing,” Proceedings of the IRE 43, no. 5 (1955): 560, https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4055456.

25 Kovásznay and Joseph, “Image Processing,” 567.

26 Russell A. Kirsch et al., “Experiments in Processing Pictorial Information with a Digital Computer,” in Proceedings of the Eastern Joint Computer Conference (1957): 221–29.

27 David F. Malin, “Unsharp Masking,” AAS Photo-Bulletin 16 (1977): 10–13.

28 Alan Martin Gilkes, “Photograph Enhancement by Adaptive Digital Unsharp Masking” (master’s thesis, Department of Electrical Engineering, Massachusetts Institute of Technology, 1974), 7, https://dspace.mit.edu/handle/1721.1/37377.

29 Gilkes, “Photograph Enhancement,” 8.

30 Gilkes, 8–9.

31 Adobe Systems, Adobe Photoshop: Classroom in a Book (Carmel, IN: Hayden Books, 1993), 134.

32 Adobe Systems, Advanced Adobe Photoshop: Classroom in a Book (Indianapolis, IN: Hayden Books, 1994), 179.

33 Kim Baker and Sunny Baker, Color Publishing on the Macintosh: From Desktop to Print Shop (New York: Random House, 1992), 239.

34 See, for example, the advertisement for Agfa’s StudioScan in Popular Photography (October 1994): 123.

35 Dan Margulis, “Sharpening with a Stiletto,” Electronic Publishing, 1998, https://web.archive.org/web/20100911034407/https://www.ledet.com/margulis/Makeready/MA27-Sharpening_With_Stiletto.pdf.

36 Margulis, “Sharpening with a Stiletto.”

37 Dan Margulis, “Life on the Edge,” Electronic Publishing, 2005, https://web.archive.org/web/20100602094946/https://www.ledet.com/margulis/Makeready/MA69-Life_on_the_Edge.pdf.

38 Adobe Systems, Adobe Photoshop 7.0: Classroom in a Book (Berkeley, CA: Peachpit Press, 2002), 96.

39 https://helpx.adobe.com/photoshop/using/adjusting-image-sharpness-blur.html.

40 Rajeev Ramanath et al., “Color Image Processing Pipeline,” IEEE Signal Processing Magazine 22, no. 1 (2005): 41, https://doi.org/10.1109/MSP.2005.1407713.

41 Canon Inc., Canon EOS-1DX Instruction Manual (2013), 135–36.

42 Wayne Lorimer, “Beware of the Worms: Sharpening Fuji Files in Lightroom,” October 14, 2019, http://nzdigital.blogspot.com/2019/10/beware-of-worms-sharpening-fuji-files.html.

43 Thomas Fitzgerald, “Why You Need to Sharpen RAW Files,” November 20, 2018, https://blog.thomasfitzgeraldphotography.com/blog/2018/11/why-you-need-to-sharpen-raw-files.

Top of page

List of illustrations

Title 1. Sharpening filter on Instagram, January 31, 2024
Caption Screenshot, Apple iPhone SE (2022).
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-1.jpg
File image/jpeg, 1.0M
Title 2. Dialog box for Adobe Photoshop 1.0 “Unsharp Mask” filter, January 31, 2024
Caption Screenshot, Apple Macintosh Quadra 900 with operating system 7.5.3.
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-2.jpg
File image/jpeg, 1.9M
Title 3. Nevit Dilmen, unsharp mask, April 14, 2008
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-3.jpg
File image/jpeg, 425k
Title 4. Artistic effect of converting a photograph into an image that resembles a drawing using an unsharp mask
Caption Illustrations taken from John A. C. Yule, Unsharp Masks and a New Method of Increasing Definition in Prints, “The Photographic Journal” 84 (1944): 324.
Credits © Michael Pritchard / The Royal Photographic Society, Bristol, UK
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-4.jpg
File image/jpeg, 1.9M
Title 5. Mathematical and electronic execution of an unsharp mask
Caption Illustration taken from Leslie S. G. Kovásznay and H. M. Joseph, Image Processing, “Proceedings of the IRE” 43, no. 5 (1955): 565.
Credits © 1955, IEEE
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-5.jpg
File image/jpeg, 1.7M
Title 6. Unsharp mask giving a “better look” to faces
Caption Illustration taken from Dan Margulis, “Professional Photoshop 5: The Classic Guide to Color Correction” (New York: Wiley Publishing, 1999), 100.
Credits © Dan Margulis
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-6.jpg
File image/jpeg, 3.1M
Title 7. Till A. Heilmann, Sharpening slider in Adobe Lightroom
Caption RAW image file shown by Timo Schemer-Reinhard, June 27, 2024, screenshot, Windows 11.
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-7.jpg
File image/jpeg, 1.2M
Title 8. Wayne Lorimer, detail of Home Sweet Home zoomed in at 200%
Caption Photograph taken on a Canon 650D using a Canon EF-S lens.
Credits © courtoisie de l’artiste, Wayne Lorimer.
URL http://journals.openedition.org/transbordeur/docannexe/image/2389/img-8.jpg
File image/jpeg, 3.7M
Top of page

References

Electronic reference

Till A. Heilmann, Sharp Images and Unsharp MasksTransbordeur [Online], 9 | 2025, Online since 26 February 2025, connection on 13 January 2026. URL: http://journals.openedition.org/transbordeur/2389; DOI: https://doi.org/10.4000/13dwu

Top of page

About the author

Till A. Heilmann

Till A. Heilmann, PhD, is a researcher in the Department of Media Studies at Ruhr University Bochum with a focus on the history and theory of digital media. Prior positions include research associate and visiting scholar at the universities of Basel, Siegen, Bonn, and Iowa and acting professor at the University of Siegen.
Till A. Heilmann (PhD), est chercheur au sein du département des Media Studies de l’université de la Ruhr à Bochum, où il se spécialise dans l’histoire et la théorie des médias numériques. Il a notamment été chercheur associé et invité aux universités de Bâle, Siegen, Bonn et de l’Iowa, et professeur à l’université de Siegen.

Top of page

Copyright

CC-BY-NC-ND-4.0

The text only may be used under licence CC BY-NC-ND 4.0. All other elements (illustrations, imported files) may be subject to specific use terms.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search