This is the second of a two part guide on diagnosing and treating the factors that prevent you achieving clear, sharp photographs. The first covers blur and can be found here. NOISE Noise is a complex, technical field and it is easy to get bogged down in detail here. In this, I will attempt to simplify it to a usable level, at the risk of misrepresenting it. The image on your camera’s sensor is produced by particles of light (“photons”) hitting it in a patterned manner. This is the “signal”, or the useful information that your sensor is collecting to make a photographic image. The photons trigger the light-sensitive elements on the sensor, called the photocells, and this information is passed on to the processing part of the camera to piece together the image, like building a mosaic of light. However, as well as being hit by all the useful light coming in, the photocells can also be triggered randomly, such as by the stray photons that are inevitably flying around. This produces unwanted points of light in your image. This is called “noise”. Diagnosis Noise is light, sometimes coloured, specks on the image. It can also show as bands of light or colours. It is most commonly seen in dark areas of the image. This is a classic example of the effects of under exposure combined with a high ISO. The detail shows what the image looks like when brightened to approximate a normal exposure. The noise is so marked that it drowns out the useable information in the image. Cause(s) High ISO Increasing the ISO of your camera increases the sensitivity of the sensor to light. This is useful when there is not much light, enabling you activate enough photocells to get a shot. However, it has its own problems. Increasing the ISO not only increases the sensor’s sensitivity to useful light, it increases its sensitivity to those pesky, stray photons. This, then, leads to an increase in noise. Underexposure Underexposure is when there was too little light activating the photocells on the sensor, due to a fast shutter speed, a small aperture or a low ISO. The resulting image is therefore too dark. Underexposure in itself doesn’t cause noise. However, it does lead to any noise that might be present to become a problem. As stated above, the useful information in an image is produced by patterned light hitting the sensor. In the bright areas of an image, the amount of light hitting the sensor is great. There is therefore a lot of useful information. In the dark areas, the amount of useful light is little. There is therefore little useful information. As noise remains constant across all areas, the proportion of noise in comparison to actual, useful information in a dark area is substantially more than in a light one. Noise therefore becomes disproportionately represented in shadows. This is why the best way to look for noise problems is to peek in the shadows. With most images, this is not a substantial issue. There is enough useful information in the bright areas to make the image and a small amount of noise in the shadows isn’t going to lose anyone any sleep. However, it does become a problem when an underexposed image is lightened in post processing. When you brighten the image you are not only lightening the useful information, you are lightening the noise as well. As the ratio of signal to noise is low in these exposures, the brightened noise will become hideously noticeable. If you combine underexposure with a high ISO, you make the problem even worse. The result will be a speckled mess with no useful detail. Long exposure A long exposure allows more light to hit the sensor. In low light conditions, this is a useful alternative to increasing the ISO and thereby sensor’s responsiveness to random photons. However, it can itself be a cause of increased noise. This is due, in part, to the fact that, the longer the sensor is exposed, the more chance there is of random photons hitting coming in and producing noise. Remedies Increase the light The more light you have, the lower the ISO and shorter the exposure you will need to get a good shot. This therefore minimises the noise caused by these, two elements. Ways to increase light were covered in the guide on blur, so won’t be expanded on here. Expose correctly The best solution to the troublesome effects of underexposure is to “expose to the right”. This means setting your exposure as high as you can without blowing out the lightest areas of the image into glowing, white glare. This will give you maximum, useful signal, minimising the impact of any noise. You can do this by using the highlight alert on your camera, if it has one. This shows blown out highlights as flashing patches of colour when you view the image on the preview screen. Take a test shot and increase or reduce the exposure to the point just before things start to flash. This is easy to do if you are using manual exposure, by just popping the shutter speed up or down. Alternatively, you can use exposure compensation if you are using aperture or shutter priority exposure modes (check your camera manual for how to use this very useful facility). Watch your ISO Keep your ISO at or below its maximum, workable level. This involves getting to know the capacities and limitations of your own, particular camera. Most modern DSLRs can cope fine with ISOs of 400-800, without producing enough noise to be troublesome. When you know your camera’s maximum, useable ISO, you can set to this and work your shutter speed around it (your aperture size will depend on what depth of field you need for your particular shot). If yours is a popular model of camera, someone might already have done some tests on this, so Google around. If not, take some test shots at ISOs 100, 200, 400, 800 and determine the point at which the amount of noise becomes a nuisance. When you have identified it, don’t go beyond it unless there really is no option. Reduce noise in post processing Noise reduction software smoothes out the speckles produced by noise. It can be done in some cameras, but the most effective way to do it is in post-processing, by using a noise reduction filter. However, it is not possible for software to distinguish between random noise and fine detail, such as hair. Smoothing the noise therefore also diminishes the detail. When overdone, surfaces and edges can also look gain a distinctive “plastic” look. It is therefore a good idea to minimise the noise as you take the photo and to use any post processing noise reduction in moderation. The final word The amount of acceptable noise or blur is relative. If you are blowing your photo up to poster size or looking at it with a magnifying glass at three inches, any imperfections will be obvious and you would want to reduce noise and blur to an absolute minimum. However, a 600 x 400 image for the web will be quite forgiving. So, be realistic in your expectations as some of the solutions produce their own, intrinsic problems. Also, stressing too much over the technicalities of a shot takes your attention away from what should be a fun, creative process. Sometimes good enough really is good enough. Photo credits: Many thanks to feaydrak, yiuu, Jatzu, Balljointed Adventure, and Zoe for the use of their example photographs.
I would suggest that unless you are using fully manual exposure, modifying aperture or shutter speed won't change your exposure just as adding light won't make your image brighter as your camera will adjust to compensate. To obtain a brighter or darker image you need to adjust your exposure offset to modify exposure.
Sorry, TomB. Not sure I get you. Is this with reference to the expose to the right recommendation? What I think you are saying is that your camera will automatically adjust shutter speed if you adjust aperture, and vice versa, keeping the overall exposure value constant. Yes, it will if you are shooting in aperture or shutter priority. If you are using one of those, you can increase or decrease exposure by using exposure compensation (which enables you to reduce or increase whichever of shutter speed or aperture that your camera is setting automatically). My example relates to shooting in manual, but the same would apply if shooting in aperture priority (which is the most useful mode if not in manual), just by using exposure compensation. However, it sounds as though this is not clear, and I can see that it could be confusing. I didn't want to get too bogged down with too much how-to-use-your-camera basics in this, but I will make that bit more explicit in the text. Thanks for the feedback.
I'm not sure I get me either. The first time I read your post I understood you to say that you could change exposure by modifying shutter speed or aperture. If you are in any exposure mode other than manual you need to ue exposure offset to change the amount of exposure. As I read it again I see that I apparently misread it initially. Apologies.
No problem. I did add to the explanation to make it clearer. It is very difficult to produce a guide that will be at once comprehensive and at the same some time accessible to serious beginner photographers, which is my aim. I am bound to get it wrong at times. I shoot and think in manual (because it is FAR easier than you could possibly imagine, if you have never used it) and need reminding that not everyone does. I do appreciate the input.
Exposure Compensation and auto ISO One gotcha to be aware of when using aperture or shutter priority mode with exposure compensation is that cameras in auto-ISO mode may also make adjustments to your ISO in these modes. Normally at a fixed ISO, adjusting exposure compensation in aperture priority mode will result in a shutter speed adjustment, but in some cameras with auto-ISO capability enabled, it might adjust ISO or both ISO and shutter speed. The camera has some internal logic whereby it decides which is the more sensible adjustment, and will often up your ISO before going to what it thinks could be a shutter speed too slow to hand hold. Exposing to the Right and Highlight Warnings One thing worth noting when trying to expose to the right is that a camera's highlight warnings (flash solid areas or zebra stripes) is typically showing you which areas will be rendered as pure white or pure black in a JPEG. If you shoot RAW, you have 1.5 to 2 stops of latitude on most DSLRs, such that areas that are just starting to flash are not actually clipping within the full range of bits saved in the RAW file. When shooting RAW, you will get the best exposure when you have just small areas of the image flashing the warning, and you'll still have the latitude to recover these areas fully during RAW conversion. With all current DSLRs, highlights are generally more salvageable than shadows, for the simple reason that the signal-to-noise ratio is better. This is why you expose to the right, rather than left. Err in the direction of overexposure. Other Tricks to Reduce Noise Some of the latest generation of DSLRs can do a neat trick when shooting with a tripod that is usable for still-life photography, which normally means product and landscape shots, whereby the camera takes a few shots in quick succession and then analyzes them to reduce noise, since noise is what, in theory, will be the only thing that differs in otherwise identical shots. This applicable to dolls since they don't generally move unless there is wind on their hair or clothing. My 6D does this, but it only works for JPEG shots. But, there is also software than can do multi-shot noise reduction as a post-processing task. Lightroom doesn't have this built-in yet, but it will in a future version. It's difficult to do by manually layering and averaging shots in Photoshop, because any sub-pixel differences in position will result in blur without sophisticated auto-correlation of the images (which is what the camera and software programs do). The noise reduction isn't a linear averaging either, it's more complicated than that, statistically, and not something you can actually do by, say, blending 3 images at 33% each, though this is better than nothing if they are aligned closely and you're going to shrink down the final image anyways. Shoot RAW If your camera supports it, shoot RAW files. In-camera JPEGs preserve only 8-bits of the 12 to 16 bits of dynamic range present in the RAW file. Cameras do compress the dynamic range when making the JPEG, so the overall range is preserved fairly well, but things get quantized or "posterized", which results in visible banding and blotchiness if you try to recover shadow areas, where there ought to be smooth gradients.
Thanks for the input, Adam. As I said in the guide, I would urge people to get to know the optimal ISO for their own cameras and to set that themselves. I only use auto ISO when I am outdoors, shooting fast and in manual. In our studio setting, shutter speed is the obvious one to leave on auto (so, shooting in manual or aperture priority, with a known ISO) as, with a tripod, it is the one that will cause least damage. For the highlight warning, I personally prefer not to clip at all, even when shooting in raw. This is because, if you let it go and judge it wrong, you have blown out detail that you will never be able to retrieve. So, better to play it safe. Also, there is no need. There will be plenty of information available if you expose properly to the right without clipping highlights. Having said that, it is always advisable to shoot in raw anyway, as it does enable you to save a near-miss, should you produce one. I include shooting in raw in my long (website) version and talk about compression artefacts and noise. I decided to take it out of here as I had to prioritise to the points that people would be most likely to be able to use. Beginners generally have a mistrust of raw, which I can completely understand. There are various other, more technical causes and recommendations in the long version that, if people are sufficiently technically-minded to consider, they can. With the in-camera trick, do you mean multi shot noise reduction? I don't have that facility on my 60D, so can't offer any personal recommendations on it. Does it work better than using a good noise filter in post processing? I will look out some reviews. If it is relatively available and does a good job, I will add it to the long list. Thanks again for your input.
This is highly situational. If a scene has a considerably higher natural dynamic range than what your camera can record, as majority of outdoor natural-light scenes do, properly exposing for the subject (doll) will often result in other areas blowing out (sky, reflections off water, etc). This is especially true with backlit shots. If you teach people that the clipping warning is a universal sign of improper exposure, they will underexpose their subject by as much as several stops, which will result in a terribly noisy photo if they brighten it in post; generally completely unrecoverable if it's a JPEG. If you want everything properly exposed, you typically need to fill light your subject (flash or reflector), or shoot bracketed shots and process them into an HDR. Even the cheapest, most basic photo software supports RAW conversion these days, and memory cards are cheap. 10 years ago, dealing with RAW was more of a challenge due to file size, and understanding of more technically-oriented conversion tools and plug-ins. Now with Lightroom, it's as easy as JPEG for the most part. Shooting RAW is the one thing that is going to give most beginner DSLR users the largest increase in image quality--not as a crutch to compensate for getting exposure right in camera, but for being able to do retouching corrections and creative treatments to the image without artifacts like banded gradients. Yes, multi-shot NR. It's a lot better than frequency-based filtering when applicable, because it removes noise without affecting sharpness, color or detail. As you correctly noted, filtering a single image doesn't have enough information to know what is really random noise and what are noise-like textures or high-frequency details actually present on the subject. When shooting real people, for example, even modest noise reduction can remove skin detail like pores, fine hairs and freckles, which doesn't always look natural. Multi-shot techniques do have enough information (multiple statistical samples) needed to figure this out. That said, this really only comes into place when shooting at extreme ISO or long exposures. It's not of great concern to most people shooting reasonably well-lit doll photos. Whenever possible, adding light is usually better and easier.
OK. We are getting into the nitty gritties of exposure now, which is going slightly off piste for the noise and blur guide. But given that it refers to some of my recommendations, I will spend some time on it. In your high dynamic range situation of dark foreground and bright sky, exposing for the sky will indeed lead to an underexposed figure and the risk of the underexposure noise I discuss in the guide. However, exposing for the figure will lead to a blown out sky, with no detail. So, neither situation is ideal (I would personally use exposure stacking in a situation such as this) and I don't think it reduces the value of recommending exposing to the right being the generally best way to avoid underexposure noise. Yes, most cameras will shoot raw, but most beginners still won't use it. In these guides, it is therefore best to go with what people will actually use. As I have already said, I do discuss it in the long version, so it is there for the more technically confident or for when that beginner is ready to move onto the next step. There are other, less daunting solutions that can be tried before raw becomes necessary. Interesting on the multi shot noise reduction. I will certainly consider that as an additional to the long guide.
Yeah, I get. It wasn't my goal to present things that contradict the general advice of the guy (which is sound), just to add some footnotes about caveats and potential pitfalls. So, to play devil's advocate here, my thought would be that anyone who has taken the step to buy a RAW-capable camera with the intent to produce artwork, and who is interested enough in getting the best results to even be reading a guide like this, is already at the point where they should not be shooting in-camera JPEG or encouraged to do so, as it's at odds with what they hope to get from such a guide. Shooting manual exposure, exposing to the right, analyzing histograms and running post-processing noise-reduction filters... these are the techniques of someone looking to get the best quality from their images, and all of them work best when utilized in a RAW workflow. I feel like suggesting that it's OK for beginners to avoid shooting RAW just keeps perpetuating the misconception that it's somehow complicated or requires specialized knowledge to get started doing. The bare minimum it usually requires is taking a few seconds to navigate the menu system of your camera to set the file format to RAW instead of JPEG, something I feel that DSLR users should be doing at unboxing time unless they have a specific reason not to (which mostly applies to sports photographers and photojournalists, not so much anyone here). For some people, especially those with newer cameras, it may require bringing your computer's OS or your photo editing software up to the latest version if you're way behind--something you should do anyways in order to use current techniques and tutorials. A program like Lightroom manages and edits RAW files the same way it does JPEGs, and you can dive into the fine tuning of the conversion process at your own pace.
I need to be realistic, and I am aware that a serious beginner is more likely to buy a tripod than shoot in raw. It involves additional processing and software that many beginner photographers are not yet ready to learn. And, if you take your shot carefully, JPEG does a perfectly adequate job. I would place raw in the same category as mirror lock up, for those who have reached the point of proficiency in photography that they are aspiring to pin sharp. For that reason, it is included in the long guide.
I don't have the statistical data to confirm or deny that, but my point was that this is a self-fulfilling condition. If you keep suggesting that a RAW workflow is more complicated or involved than a JPEG workflow, people will believe it and it will continue to be true. Again, true 10 years ago, not now. Most people who have reached the point of deciding to buy a DSLR or mirrorless camera and are endeavoring to learn photography are using either the software supplied with the camera, or more likely Adobe Lightroom, because it's $10/month to license Photoshop and Lightroom vs. $900+ to purchase them outright prior to Creative Cloud licensing. So in reality, they're not getting the most out of their camera OR the software they're paying for. If you are shooting JPEG because you are uploading shots right from your camera, well, there is nothing wrong with this if it gets you images you're happy with, but it's not really the starting point for taking your photography to the next level though the use of guides like this and chances are good that you already have all the software you need to import and develop a RAW file to a superior JPEG end product. I disagree. Everyone with a DSLR camera can get very tangible and immediate benefits from shooting RAW files, in most cases with little to no additional work or software. You can work the same way, and things like making exposure adjustments or white balance changes will just work better. More importantly, the benefits are retroactive: shots you took months or even years ago and love can be revisited and improved upon with your latest techniques, software and knowledge, because you have good source files to work from. Want to do a black and white rendering... no problem. Forgot and left your camera in an inappropriate white balance mode, no problem. Underexposed an amazing shot.. much more salvageable. People who move to a RAW workflow invariably have that Ah Ha! moment when they realize that their camera has been doing all manner of auto-leveling and color adjustments to their images, and giving them just one of an infinite number of possible renderings of the data and that many of their favorite shots could have been made so much better if they'd only known. I normally tell people, even if you are not ready to embrace RAW workflow, but you're shooting a once-in-a-lifetime trip or event, shoot JPEG+RAW. Get an extra memory card or external drive if you need to, just do it. Mirror lockup on the other hand is a very specialized feature intended for high-magnification photography, and if you find that you need to use it for general purpose photography it's an indication that it's time to buy a tripod that actually works. The original intention of the feature was not to compensate for the mirror setting wobbly tripods oscillating, but that's why most people who see benefits from it are.
I'm with Adam here. I'm a huge advocate of RAW image capture. The more color depth you start with the less risk of banding and other image degradation as a result of editing. If you are editing your image you will always get a better result if you start with a RAW image.
I am too but, for the reasons given a few times up there (hence why I stopped. No point saying the same thing over and over, ad infinitum) I didn't include it in this, basic version of the tut. It is included in the full tut on my website.