D
david stock
Guest
Here’s my take on the subject, with thanks and credit to the pioneers on the Small Sensor newsgroup:
Film grain and digital noise are fundamentally different. Film grain literally IS the image. Digital noise is visual static that overlays the image.
On the other hand, film grain and noise share important characteristics. Each of them can, beyond a certain limit, degrade the resolution and overall image quality of a photograph. Each of them is associated with smaller, portable cameras (often used in street photography). And each of them has been embraced by a subset of photographers in search of an impression of gritty realism and authenticity, or simply a particular “look” to their work.
Recent experience making black and white images with the small, high-quality Ricoh GRD2 camera has caused me to think more about grain and noise as photographic elements, and has forced me to alter my workflow. Here are a few observations:
Grain and noise don’t always degrade perceived sharpness. If they have the right relationship to what the lens resolves, they can actually increase the impression of sharpness. There are at least two ways this can happen. One is that visible grain or noise can add acutance, or edge sharpness, to the image. The “definiteness” of on/off, black/white elements makes visible detail look sharper. Experienced darkroom users take advantage of dilute Rodinal and other grain-enhancing developers for just this purpose.
A second sharpening effect, rooted in visual psychology, occurs when there is an even mist of grain or noise throughout a detailed image. Under the right circumstances, the brain “imagines” that there is more fine detail hiding behind the visible grain or noise. This effect only happens within a certain definite range, when there is a specific relationship between noise/grain and optical detail. This effect is exploited by advanced digital printmakers, who sometimes add small amounts of noise or grain to digital files to make them seem crisper and more detailed (as well as less “plasticky”) when printed.
Digital noise is worst when it is smeary. This kind of noise only serves to obscure detail and distort tonalities. Smeary noise is a problem that affects most current small-sensor noise-reduction technology and much of the current noise-reduction software. Fortunately, the GRD2 manages to produce “sharp noise” better than other cameras in its class.
In analog, traditional photography, the grain tends to be less visible in bright highlight and shadow areas (where it is either barely evident or, at the other extreme, concentrated into solid black). It is distributed predictably and naturally across the tonal scale. In contrast, digital noise is always most prominent in the shadow areas, especially in the capture of current small-sensor cameras at higher ISOs. Personally, I find this UNBALANCED distribution of noise more disturbing than the noise itself.
Moreover, our efforts to sharpen shadow noise (and prevent it from look-ing smeary) can result in an overlay of white speckles which seem to have no relationship to the optical image. A photograph with smooth clean highlights and midtones combined with smeary or speckled shadows isn’t what I’m aiming for.
There are many tactics for controlling noisy shadows, including avoiding sharpening altogether and careful use of noise reduction programs or the various Photoshop filters. My own approach is, first, to set a limit on acceptable shadow noise (avoiding smeary, detail-obscuring capture) and then to add noise to the rest of the image to match the noise texture of the shadows. Usually this involves making selections of tonal ranges and applying sharpening or adding noise as needed.
This is not necessarily better than other approaches. Some photographers accept or even intensify high-ISO noise for a look they prefer. Others use small sensor cameras only at low ISO’s to produce stunning image quality from the Ricoh’s small sensor. For the look of my black and white work, I have chosen a middle, rather traditional-looking path.
My current method of working with the GRD2 is to restrict it to ISO 400 or below unless absolutely necessary. I find that the high quality Ricoh lens and sensor combination maintains surprising quality up to that point. Also, since it is completely practical to shoot at f/2.4 or f2.8, the GRD2 at ISO 400 allows me significantly greater low-light flexibility than many other digital cameras.
I process my RAW files in ACR, paying careful attention to shadow noise. Al-though most of my current work is in black and white, the images still benefit from chroma noise reduction, as Sean Reid and others have shown. Working in Photoshop, I do my sharpening on a separate layer, working with layer effects to calibrate how much of the sharpening is applied to the shadows.
I do not avoid all shadow sharpening—in fact, smeary noise can sometimes be finessed with careful sharpening or the addition of a small amount of noise. But I personally avoid sharpening the shadow tones to the point of white speckling. In extreme cases, I do sharpen until white speckles appear, then knock them down with the Dust and Scratches filter and a high threshold.
After I finish sharpening, the “noise texture” of my middle tones and shadows matches fairly well. I then select the highlight tones and add a small amount of noise—rarely more than 3 or 4 units—with Photoshop’s Add Noise filter. For difficult images, I try different combinations: selective sharpening, adding noise, fading it, using Dust and Scratches. After some practice, this isn’t actually as complicated as it sounds; it becomes intuitive. My aim is to get an overall mist of noise, one that looks even, or at least “appropriate,” from the deepest shadows to the near-white highlights.
Again, I recognize that this approach isn’t for everybody. I’ve seen excellent small sensor images with white speckles in the shadows, and others where noise is suppressed by being pushed down into deep black shadows. In my own case, evening out the noise across the tonal scale has extended the useability of the Ricoh and allowed me to make friends with the small sensor.
Overall, I’m surprised at the quality that can be obtained from the GRD2. It is truly a serious photographers’ camera, capable of yielding good-sized high quality black and white prints.
Visible noise and grain are legitimate elements of photographic style. Em-bracing noise in small sensor cameras is one stylistic choice, as is managing noise.
Visual characteristic that were once mandatory, caused by technical limi-tations in photography, have a way of becoming optional “looks” as time goes on. Very grainy analog capture stopped being a sure-fire indicator of street cred years ago, as fast films became more and more fine-grained. Instead, gritty grain became largely an aesthetic choice. It will be interesting to see what happens as small sensors improve and shadow noise starts yielding to technology. Will we add it back in?
I haven’t figured out how to put photos in posts yet, but a few of my recent Ricoh black and white images are in the “new” portfolio on my website: david-stock.com.
Film grain and digital noise are fundamentally different. Film grain literally IS the image. Digital noise is visual static that overlays the image.
On the other hand, film grain and noise share important characteristics. Each of them can, beyond a certain limit, degrade the resolution and overall image quality of a photograph. Each of them is associated with smaller, portable cameras (often used in street photography). And each of them has been embraced by a subset of photographers in search of an impression of gritty realism and authenticity, or simply a particular “look” to their work.
Recent experience making black and white images with the small, high-quality Ricoh GRD2 camera has caused me to think more about grain and noise as photographic elements, and has forced me to alter my workflow. Here are a few observations:
Grain and noise don’t always degrade perceived sharpness. If they have the right relationship to what the lens resolves, they can actually increase the impression of sharpness. There are at least two ways this can happen. One is that visible grain or noise can add acutance, or edge sharpness, to the image. The “definiteness” of on/off, black/white elements makes visible detail look sharper. Experienced darkroom users take advantage of dilute Rodinal and other grain-enhancing developers for just this purpose.
A second sharpening effect, rooted in visual psychology, occurs when there is an even mist of grain or noise throughout a detailed image. Under the right circumstances, the brain “imagines” that there is more fine detail hiding behind the visible grain or noise. This effect only happens within a certain definite range, when there is a specific relationship between noise/grain and optical detail. This effect is exploited by advanced digital printmakers, who sometimes add small amounts of noise or grain to digital files to make them seem crisper and more detailed (as well as less “plasticky”) when printed.
Digital noise is worst when it is smeary. This kind of noise only serves to obscure detail and distort tonalities. Smeary noise is a problem that affects most current small-sensor noise-reduction technology and much of the current noise-reduction software. Fortunately, the GRD2 manages to produce “sharp noise” better than other cameras in its class.
In analog, traditional photography, the grain tends to be less visible in bright highlight and shadow areas (where it is either barely evident or, at the other extreme, concentrated into solid black). It is distributed predictably and naturally across the tonal scale. In contrast, digital noise is always most prominent in the shadow areas, especially in the capture of current small-sensor cameras at higher ISOs. Personally, I find this UNBALANCED distribution of noise more disturbing than the noise itself.
Moreover, our efforts to sharpen shadow noise (and prevent it from look-ing smeary) can result in an overlay of white speckles which seem to have no relationship to the optical image. A photograph with smooth clean highlights and midtones combined with smeary or speckled shadows isn’t what I’m aiming for.
There are many tactics for controlling noisy shadows, including avoiding sharpening altogether and careful use of noise reduction programs or the various Photoshop filters. My own approach is, first, to set a limit on acceptable shadow noise (avoiding smeary, detail-obscuring capture) and then to add noise to the rest of the image to match the noise texture of the shadows. Usually this involves making selections of tonal ranges and applying sharpening or adding noise as needed.
This is not necessarily better than other approaches. Some photographers accept or even intensify high-ISO noise for a look they prefer. Others use small sensor cameras only at low ISO’s to produce stunning image quality from the Ricoh’s small sensor. For the look of my black and white work, I have chosen a middle, rather traditional-looking path.
My current method of working with the GRD2 is to restrict it to ISO 400 or below unless absolutely necessary. I find that the high quality Ricoh lens and sensor combination maintains surprising quality up to that point. Also, since it is completely practical to shoot at f/2.4 or f2.8, the GRD2 at ISO 400 allows me significantly greater low-light flexibility than many other digital cameras.
I process my RAW files in ACR, paying careful attention to shadow noise. Al-though most of my current work is in black and white, the images still benefit from chroma noise reduction, as Sean Reid and others have shown. Working in Photoshop, I do my sharpening on a separate layer, working with layer effects to calibrate how much of the sharpening is applied to the shadows.
I do not avoid all shadow sharpening—in fact, smeary noise can sometimes be finessed with careful sharpening or the addition of a small amount of noise. But I personally avoid sharpening the shadow tones to the point of white speckling. In extreme cases, I do sharpen until white speckles appear, then knock them down with the Dust and Scratches filter and a high threshold.
After I finish sharpening, the “noise texture” of my middle tones and shadows matches fairly well. I then select the highlight tones and add a small amount of noise—rarely more than 3 or 4 units—with Photoshop’s Add Noise filter. For difficult images, I try different combinations: selective sharpening, adding noise, fading it, using Dust and Scratches. After some practice, this isn’t actually as complicated as it sounds; it becomes intuitive. My aim is to get an overall mist of noise, one that looks even, or at least “appropriate,” from the deepest shadows to the near-white highlights.
Again, I recognize that this approach isn’t for everybody. I’ve seen excellent small sensor images with white speckles in the shadows, and others where noise is suppressed by being pushed down into deep black shadows. In my own case, evening out the noise across the tonal scale has extended the useability of the Ricoh and allowed me to make friends with the small sensor.
Overall, I’m surprised at the quality that can be obtained from the GRD2. It is truly a serious photographers’ camera, capable of yielding good-sized high quality black and white prints.
Visible noise and grain are legitimate elements of photographic style. Em-bracing noise in small sensor cameras is one stylistic choice, as is managing noise.
Visual characteristic that were once mandatory, caused by technical limi-tations in photography, have a way of becoming optional “looks” as time goes on. Very grainy analog capture stopped being a sure-fire indicator of street cred years ago, as fast films became more and more fine-grained. Instead, gritty grain became largely an aesthetic choice. It will be interesting to see what happens as small sensors improve and shadow noise starts yielding to technology. Will we add it back in?
I haven’t figured out how to put photos in posts yet, but a few of my recent Ricoh black and white images are in the “new” portfolio on my website: david-stock.com.
