Reflection XXXVII

high noise / low noise

Noise happens. There is always some degree of noise present in any electronic device that receives or transmits a signal. Though noise is unavoidable, it can become so small relative to the signal that it is no longer visible. It’s all about a good signal to noise ration (SNR). The image is the signal. The capture and carrier mediums’ biproducts are the noise. (Film grain is noise.)The higher the SNR the more the image overpowers the noise; the lower the SNR the more the image becomes confused with noise. While some noise increases the apparent sharpness of images, the vast majority of noise found in images degrades quality. It’s best to avoid it.

There are a number of things you can do to reduce noise in your digital images at the point of capture.

1       Expose to the right

2       Use lower ISO settings

3       Use bigger sensors

4       Use faster exposure times

5       Keep equipment cool

Start here. Use software to reduce the appearance of noise only when needed. (Of course, reducing noise may not be the only deciding factor when selecting conditions, tools, and techniques. As with all things, you will have to balance many concerns simultaneously. Adhere to the above principles when practical.)

Knowing how noise is produced will help you avoid it.

histograms_zones_caution

avoid the red zone, caution in the yellow zone, go for the green zone

ETTR

Underexposure results in more visible noise. Darker regions contain more noise than lighter regions in digital capture; the opposite is true of film. This is because the darker regions are recorded with fewer photons and less bits of data. It stands to reason that darker regions of images are produced with fewer photons. But why are they recorded with less bits of data? Digital cameras record data in a linear progression. If a digital sensor is capable of recording 14 bits of data (or 16,384 shades of gray) with a dynamic range of 8 stops, the lightest stop contains half the data in the file; the next lightest stop contains half as much data; and so on; it generates this progression from dark to light – 32 / 64 / 128 / 512 / 1024 / 2048 / 4096 / 8,192. This means you want to expose to the right. Avoid the lower 2 stops whenever possible; they contain less than 1% of the total data in the file. With so little data the signal becomes confused with the noise; the signal to noise ratio is very low. What’s more, when these tonal regions are adjusted, brighter or contrastier, the noise contained there quickly becomes more pronounced and with so few bits of data it also has a tendency to posterize. In addition, significant under exposure greatly increases the chances of producing Bayer pattern noise, a type of noise that is seen in larger areas and is more challenging to remove in post-processing. You’re much better off making a light exposure and darkening the Raw file during post-processing; this way your shadows will be represented with much more data and contain less noise. When light becomes challenging, which should you choose – under expose (without clipping) or raise the ISO? The precise cut off point is different for every camera, but in general avoid the lower 20% of the histogram. You’ll get less noise if you boost ISO and move the histogram up.

ISO

Higher ISOs amplify noise. ISO (International Standards Organization) is a descriptor that signifies absolute sensitivity to light. A digital sensor has one native ISO. Higher ISO settings simply boost the resulting signal. This is useful, but not ideal. When the brightness of the image is boosted, the noise is too.

Use “Shorter” Exposure Times

Longer exposures generate more noise. Hot pixels become hotter. All sensors have a few pixels that heat up faster than others, producing brighter than expected values. Some even have a few dead pixels that never fire, producing only black pixels. During longer exposures hot pixels are given more opportunity to heat up, growing brighter still; slightly hot pixels not visible at shorter exposure times become visible. As digital sensors age, hot pixels may become hotter and more pixels may become hot. Hot pixels produce a consistent fixed pattern of noise that can be recorded for given exposure times, making it easy to reduce. There is a duration at which a sensor suddenly produces a lot more noise; in older models this could be seen in exposures lasting only one minute, but with newer sensors this is greatly extended and most users will not encounter this; if you’re making very long exposures it’s useful test this and find the duration at which this happens for a your specific camera.

Use Bigger Sensor Sites

When it comes to noise, bigger is better – theoretically. Bigger sensors have more light gathering capacity, producing a higher signal to noise ratio, or cleaner images. More isn’t necessarily better. Cameras with more photosites (yielding more megapixels) packed into smaller areas tend to produce a lower signal to noise ratio, or noisier images. That said, a stronger signal does not necessarily guarantee lower noise. It’s the relative amounts of signal to noise that determines how noisy an image appears. The way a camera processes the file it makes has a significant impact on the final quality. Consequently, many medium format cameras that produce beautiful files in daylight typically produce noisier files than DSLR’s with smaller sensors in low light.

Stay Cool

High temperatures exacerbate noise. Thermal energy (leakage current) in semiconductors can generate an electrical signal that is difficult to distinguish from the optical signal. Ambient temperature increases leakage current by a factor of 2 for every 8 degrees Centigrade. Whether due to ambient temperature (You might start to see some effect over 90 F / 32 C degrees.) or storage (Don’t leave your camera in the sun, especially on a car seat, for long periods of time.), your camera can get hot and this will increase noise, which you can reduce by cooling off and keeping your camera cooler.

The best way to avoid noise is to produce as little of it as possible during exposure. These five tips will help you do just that.

.
.

hdr_artifacts_425

Different HDR renderings accentuate different artifacts

HDR (high dynamic range) imaging captures extreme contrast ratios and subsequently renders them for LDR (low dynamic range) devices, monitors and / or prints. The very things that make HDR renderings appear natural can make them appear unnatural if taken too far.

Midtone compression

You can’t avoid midtone compression, they get caught in the middle when the relationships between highlights and shadows are compressed. But you can take steps to minimize it by being sensitive to this when choosing compression settings and amounts and by taking subsequent steps to expand it.

Tonal inversions

Some compression routines and settings can be so aggressive that they create inversions or solarizations of specific tonal relationships. Avoid this, there is no subsequent cure. If you like the overall effect of an aggressive setting and the inversion is contained to one area of an image you can render an image twice, once for the overall effect and once for a specific area, and then blend the two together using Photoshop’s layers and masks. 

Saturation Distortions

Saturation changes when lightness shifts but color stays the same. Because HDR produces effects that can be aggressive and localized to specific set of tones, the saturation shifts that accompany tonal compression often appear unnatural. Selectively adjusting the saturation of specific hues, with tools like the HSL panel in Lightroom or Camera Raw, can often convincingly cure a majority of these side effects and hide the rest.

Halos

HDR softwares help restore midtone contrast by accentuating contours. When used aggressively this edge contrast can produce halos.

Over the years, these algorithms have dramatically improved their ability to treat the halo (light line) separately from the line (dark line), suppressing the first more than the second. Sometimes, to avoid distracting halos at the border of skies, you may want to make a second rendering for the sky and blend it with another rendering using Photoshop’s layers.

Exaggerated Texture

The contour accentuation that HDR softwares use to help restore midtone separation can exaggerate texture, for better or worse. For this reason, HDR software routines can be excellent detail enhancers that produce different results than the standard image sharpening tools. Once again, the trick is not to overdo it.

Noise

Noise accentuation in HDR images is a bi-product of tonal compression and the routines designed to exaggerated micro detail. Just like LDR exposures, to avoid unnecessary noise use the lowest practical ISO setting. In addition, use exposures that are bracketed by two or fewer stops. Finally, plan to reduce noise after rendering; recent noise reduction algorithms are capable of producing remarkable results.

Banding

To avoid banding (most visible in smooth gradations) use exposures that are bracketed by two or fewer stops. Process files in a high (16) bit depth. When you see banding appear during rendering, change the settings that produce it; there is no satisfactory cure for banding in post-processing.

The vast majority of objections to HDR imaging come from observations of examples where practitioners were oblivious of or insensitive to these regularly occurring artifacts. When HDR renderings are made with these potential pitfalls in mind, the objections disappear. In many cases, viewers don’t realize HDR software was used in the production of a photograph. Some HDR renditions challenge classic photographic aesthetics and offer new ones. If viewers do recognize that these artifacts exist in images and they are appropriate for the visual statement being made, not unconscious bi-products but conscious choices, then viewers who are open minded may experience a sometimes surprising new window onto the world.

Read more on HDR techniques here.

Learn more in my digital photography and digital printing workshops.

8 HDR Myths Debunked

December 12, 2016 | 2 Comments |

illuminationxxxviii_425

There are many misconceptions surrounding the practice of high dynamic range (HDR) photography. Here are eight – debunked.

HDR is new

Within the first five years of the invention of photography photographers began bracketing exposure to extend the dynamic range of photography. They used chemistry to process their negatives instead of software to process their files – but they still bracketed exposures to capture contrast ratios that exceeded paper, glass, and film.

HDR is hard

High dynamic range imaging has become so commonplace that cameras and software make it increasingly easy to practice HDR techniques – auto-bracketing, merging and rendering.

HDR requires the use of a tripod

While there are times when the use of a tripod is required, when exposures are long in duration, in a majority of cases current cameras’ auto-bracketing features and softwares’ image alignment algorithms make hand-held exposure bracketing highly practical.

HDR requires exposure bracketing with a fixed aperture

Aperture priority (fixing the f-stop) is recommended during bracketing to avoid changes in depth of field between exposures – and so the appearance of a loss of focus, which in extreme cases can make alignment challenging. However, it’s entirely possible to bracket exposure value by changing aperture, especially when depth of field issues are minimal. Shutter speed (motion), aperture (depth of field), and ISO (noise) can all be used to bracket; each has different consequences.

With HDR, you get better results with more exposures

Using too many exposures (half or quarter stops) can produce almost as many artifacts as using too fiew exposures (more than 2 stops); chiefly banding and excessive midtone compression.

HDR looks unnatural

The appearance of HDR images is flexible. What software you choose to render your image and how you choose to use it is up to you. You can produce a classical or a contemporary appearance with many possible variations in between. Whether the technique becomes invisible or obvious is a choice.

You can only choose the look of one HDR rendering

You can blend different renderings of the same image either globally or locally. (Use Photoshop’s layers and masks.) Again, how you choose to render your images is controled by you, not technology.

HDR is cheating

Really?

Read more on HDR techniques here.

Learn more in my digital photography and digital printing workshops.

canon_autobracket_back_425

back LCD menu

canon_autobracket_top_425

top LCD menu

It’s easy to set your camera to auto-bracket. The hardest part of this process is navigating a camera manufacturer’s menu. Once you find it – and do it several times – you won’t forget it.

Here’s how to do it on current Canon cameras – the steps are similar for other cameras but the buttons and menus vary.

First, set the number of frames made in each bracketed sequence. Press the Menu button. Use the main command dial (top) to cycle through the menus on the LCD screen (back) Go to the 4th tab (small camera) > 1st list and then the use the jog wheel (back) to select the 5th item. Press the set button to select it. Use the jog wheel to select the number of shots and press the set button once again. While 3 is the most commonly used, it’s not unusual to use 5 or even 7. Because 3 is the most commonly used number, it’s likely that once you set this, you’ll reset it infrequently.

Second, set the difference in exposure values between shots. Press the Menu button.

Use the main command dial (top) to cycle through the menus. Go to the 1st tab (large camera), 2nd list, 1st item. Use the set button to highlight this function and the command dial (top) to set the difference in exposure values, then press the set button again. Remember! Press the set button at the end of this – if you don’t these new settings won’t be saved.

Remember! Continue shooting in bracketed sequences as long as auto-bracketing is turned on. When you return to shooting single shots, turn auto bracketing off. If you don’t, one out of three exposures will be improperly exposed – because your camera is still bracketing exposure. The quickest way to turn auto bracketing off is to turn your camera off and on, which will turn auto-bracketing off on most cameras.

When auto bracketing is activated, on the display at the top of your camera you’ll see a series of bars indicating how many exposures will be made and at what exposure value.

What about exposure compensation? Use exposure compensation or not. Bracketing is extreme exposure compensation. All that matters is that you create multiple exposures that once combined render both excellent highlight and shadow detail.

For HDR exposures that are hand-held set your drive or burst mode to continuous; the fastest setting you have; One Shot is too slow.

For HDR exposures using a tripod, consider using a cable release and mirror lockup to further reduce camera motion.

Practice turning on and turning off auto-bracketing and it won’t seem nearly as complicated as the first time you do it and in time the habit will become so ingrained that you won’t have to think about it any more. But until turning auto-bracketing on and off becomes second nature, proceed carefully and methodically. The most common mistake is to turn it on and forget to turn it off.

Read more on HDR techniques here.

Learn more in my digital photography and digital printing workshops.

salinas1_425

Today’s cameras have the ability to generate HDR merges on the fly. The problem is that they produce JPEGs with a smaller gamut (lower saturation), lower bit depth (fewer shades of gray), and compression artifacts (noise and jagged edges) and they offer no control over the tone mapping process.

If you want a better HDR file, choose to make multiple bracketed Raw files, then merge and tone map them manually. Remember, aside from exposure settings, in camera settings that affect the look of your image have little or no affect on Raw files, which can be processed any way you want to process them.

In camera HDR JPEGs can offer a fast and convenient preview of potential HDR results. You can get the convenience of one and the quality of the other by setting your camera to produce both JPEG and Raw files simultaneously.

Read more on HDR techniques here.

Learn more in my digital photography and digital printing workshops.

Condensation X - Prelude

1 exposure

scenes like this won’t benefit from HDR

Incubation II

 3 exposures

scenes like this may be made with 1 exposure but benefit from more

Antarctica CXXX

5 exposures

scenes like this require HDR

How many exposures do you need for HDR images?

It depends.

It depends on the contrast ratio of the scene you’re photographing.

And, it depends on how many corners you’re willing to cut.

There is a case to be made for rendering all images with bracketed exposures. If the best data in a file is in the top stop of the dynamic range of a camera, then making one exposure per stop of dynamic range in the scene, each weighted to the top stop pf the camera (ETTR), will give you ideal data in every stop of the rendered scene, once the exposures are merged. So, ideally, you’d make one exposure per stop of dynamic range in the scene. In your first exposure place the shadows in the top stop of the histogram (to the right), without clipping. Then in subsequent exposures reduce exposure in one stop increments, making a new exposure each time, until the highlights are placed in the top stop without any clipping. Then stop.

Practically, you don’t need to make one exposure per stop of dynamic range in the scene because detail rendered by data in the middle of the histogram, though potentially not ideal, is perfectly usable. Here’s where theory can overtake practice. Test this yourself and find out whether using many more exposures is really worth the effort. In my own tests, I haven’t seen enough difference to make this a regular practice. If you can’t see a difference, then the difference doesn’t matter.

If data in or above the middle of the histogram is good enough, then some scenes will have so little dynamic range (one, two or three stops) that you only need one exposure. Remember to expose to the right to get the best data. Files like this will look over-exposed until they are processed. Once again, for these types of images, using more exposures and processing them with HDR software routines may actually cause problems like posterization.

Surprisingly, some images that look ‘normal’ and don’t seem to need HDR can actually benefit from additional exposures; images where shadows are close to the left of the histogram; the last three stops to the left have dramatically less data to render tonal values and are prone to excess noise and posterization.

For scenes that exceed the dynamic range of a camera sensor and can’t be captured without clipping in a single exposure, the number of exposures needed to render them depends on the contrast ratio of the scene – some need two, some three, some four, some five …

How about making more even exposures and bracketing in half stops? Theoretically this would work. In practice, you’re likely to find that using too many exposures causes some artifacts like excessive midtone compression and posterization.

The number of exposures you need also depends on the number of stops of difference you choose between exposures. Using exposures one and a half stops apart usually produces results just as good as those made with one stop apart. Using exposures with two or more stops of difference between them may cause posterization, which is most visible in smooth gradations, like skies. Using exposures less than one stop apart usually does not produce better results. In a few cases using too many exposures can produce artifacts like excessive midtone compression and sometimes posterization.

In the end the number of exposures you use matters because if exposures are too close or too far apart results will not be optimal, nevertheless the final number of exposures you need depends on both the scene you’re photographing and the camera you’re photographing it with.

Read more on HDR techniques here.

Learn more in my digital photography and digital printing workshops.


keep looking »

Subscribe

Get the RSS Feed  

Subscribe by Email