|
Styles: High Dynamic Range (HDR) Photography
If you look at any of the photo gallery websites such as Flickr, you will see that HDR photography is quite popular at the moment.
Unlike the other styles of photography covered in this styles section of this website, HDR photography is a technique that requires you to carry out some image processing on your computer. Therefore this section could just as easily belong in the Digital Darkroom section of the site. But for now, I'll leave it here as it also requires some specific actions at the 'taking the picture' stage.
So first lets define what is meant by dynamic range.
Dynamic Range
Dynamic Range refers to the difference between the lightest part of a scene or image and the darkest part. For example a picture of some grey buildings on an overcast day might have a very small dynamic range - whereas a picture taken in woodland with the sun visible through the trees might have a very wide dynamic range. It is not uncommon for the brightest part of an image to be 100,000 times brighter than the darkest part.
The image sensor in a camera has a limited dynamic range (it's usually around 7 'stops' - you'll have an idea what that means if you've read the section on exposure). However the difference between the sky on a bright sunny day and shadow detail in the foreground of the same scene can be considerably greater than this (12 to 14 stops maybe). What this means is that when you take a picture of this scene, you must decide which bit of that 12-14 stops you want to expose for as you are only going to be able to capture about 7 stops worth in a single exposure.
You could 'expose for the highlights', which means you will get a nice sky, but your foreground detail will be all black. Or you could 'expose for the shadows' in which case you'll get some foreground shadow detail but your sky will just burn out to complete white.
So if your camera's sensor can't cope with the bright sky and the dark foreground, what can you do? The answer lies in the art of High Dynamic Range photography.
|
A source of confusion
Many people confuse dynamic range with colour depth (i.e. the number of bits used to represent the colours in each pixel). If you take a standard JPEG image from a camera (which by definition must be 8-bit), load it into photoshop and then save it as a 16-bit TIFF file, it has no more dynamic range than the original 8-bit image. All you have done is used more bits to represent the same information.
In a similar way, a lot of modern DSLR cameras support 14-bit RAW files compared to older models that support 12-bit RAW files. The 14-bit files have the potential to store a greater dynamic range, but in reality it won't as the dynamic range is a factor of the imaging sensor. In most cases, the 14-bit RAW will just store the same dynamic range as the 12-bit but do it more accurately.
Of course it is resonable to expect that a modern camera will have a better image sensor and therefore a higher dynamic range than an older model, but it would not be as much as the step from 12-bit to 14-bit might suggest. |
HDR Images and Tone Mapped Images
To truely represent the high dynamic range of a scene in a photograph, you should be able to store every bit of information captured (even if that capture process involves taking many different images at different exposures - see below). An 8-bit image (like JPEG) can only store 256 different levels - a 16-bit image can store 65,536 levels. However a HDR scene can often contain even more than this. So therefore you really need to use a 32-bit image file format. There are dedicated formats for this such as Radiance RGBE (.HDR) but these are not common, and you certainly can't take these files to your local Tesco/Wal-Mart to be printed.
So when you are browsing photo sites such as Flickr, and you see images described as HDR images, are they really HDR images? Well the answer is yes and no, depending on your definition.
Yes in that these images do contain the wider dynamic range of the original captured scene contained within a single image, but no in that it has been saved as an 8-bit image which is only capable of showing 256 levels per colour channel, and therefore does not contain all the original information that was captured. A lot of information has been thrown away to accommodate that wide dynamic range into the limits of an 8-bit format. Effectively the tonal range has been compressed. What you are looking at is a tone-mapped image.
Image used courtesy of Richard Cooper.
|
When is HDR not HDR?
The image above is very good example of a HDR image. However this image is an 8-bit JPEG file (a low dynamic range format). You are probably viewing it on an LCD screen (which is a low dynamic range device) or a printout (which has an even lower dynamic range).
Think about it for a second ... we've all experienced the vast dynamic range that occurs in nature. You are sitting in the shade of a tree with your head in a book and you suddenly look up to the bright sky. For a few seconds, your eyes can't deal with the brightness until the iris (the eye's aperture) has closed down.
When viewing a tone mapped (HDR) image on screen, you don't squint your eyes when you look at the sky - that's because the screen can't create anywhere near the same dynamic range as mother nature. The image contains elements of the wide dynamic range that was captured in the multiple photographs that went into making the HDR image, however it has been compressed into a much smaller dynamic range suitable for storing in a JPEG file and viewing on an LCD screen.
Seeing such a high dynamic range compressed and viewed on a low dynamic range device like your screen without your eye's having to adjust is what makes HDR images look so distinct and unnatural. |
Tone Mapping
Tone mapping is the name given to the process of tonal compression described above. It takes the original HDR data (32-bits per channel, with potentially a near unlimited number of light levels) and compresses these down to 256 levels (8-bit per channel, suitable for saving to a JPEG file).
Defining the term in just once sentence:
Tone Mapping is a way of reproducing the high dynamic range of a scene onto a low dynamic range device.
If you use the RAW format on your camera rather than JPEG, you have already been carrying out tone mapping without realising it. RAW files are usually 12 or 14 bit images, and therefore when you convert them to JPEG files, you are compressing the tonal range down. This compression is relatively simple and is generally applied as a global operation - i.e. the same tone mapping curve is applied to the whole image. However when converting from a true 32-bit HDR image more sophisticated algorithms are required that consider a pixel in relation to it's surrounding pixels. For example a pixel of a certain brightness will be mapped differently depending on whether it is in a bright part of the image or a dark part of the image.
Basic Steps of HDR Photography
1. Instead of just taking one picture and having to choose whether to expose for highlights or shadows, take more than one picture of the scene - 3 or 5 pictures are quite common - each at a different exposure. Most cameras have a feature called 'auto exposure bracketing' to help you achieve this.
2. Combine these images on the computer to create a new single image taking the shadow detail from one image, the midtone detail from a second image, and the highlight detail from the third image. This can be done manually, using the HDR merge feature of your software, or for the best results, using dedicated HDR software such as photomatix. Here's a picture showing how three initial images (the 3 small ones on the left) have been merges to create a single HDR image.
This image was taken from the Photomatix website. We're sure
they won't mind as we've linked back to the site where you can
see it for yourself and if people like the look of the image, they
may even go out and buy the software.
Step 1: Taking the Pictures
The key points to taking the initial pictures for HDR photography are:
- Ensuring that the camera has captured the exact same scene (i.e. has not moved between taking the different exposures).
- Ensuring that the number of pictures you take cover the complete dynamic range of the scene.
To make sure you get the best result when merging the images together, the only thing that should differ between each photograph is the exposure. Therefore you should use a sturdy tripod when taking the images. It is possible to take the pictures without one as long as you camera has auto exposure bracketing which will take the 3 pictures in rapid succession with a single press of the shutter button. You should try and rest the camera on something and hold it as steady as possible while all 3 pictures are being taken.
For most pictures you will want to cover the complete dynamic range of the scene. For a typical landscape photography, this means getting a picture of the bright sky without any of the highlights burned out, and getting a picture of the foreground shadow detail without it being lost in the blackness.
Although you can take your initial shots by trial and error by leaving the camera to take a light reading and then just setting the AEB to the maximum, you can also follow these steps.
1. Put your camera into aperture priority mode and select some aperture (not really important at this stage).
2. Point the camera at the sky (make the sky fill the frame - zoom in if you have to)and take a light reading by half pressing the shutter button - make a note of the shutter speed (for example let's say it is 1/500th second)
3. Now point the camera so the darkest part of the scene fills the frame (make sure no bright sky is visible), and take a light reading again by half pressing the button. Make a note of this shutter speed (let's say it was 1/30th second)
4. Work out how many stops there is between the two values (see the page in the book: "Exposure (part 2)" ). In the example given above, the difference between 1/30th second and 1/500th second is 4 stops. Therefore you need to capture three images where the difference between the brightest and darkest is 4 stops.
5. Put your camera in manual exposure mode, and select an exposure mid-way between the brightest and darkest readings - in this case 1/125th second shutter speed with the same aperture setting, and set your auto exposure bracketing to plus/minus 2 stops. You will end up with three images:
- f/5.6 1/125s
- f/5.6 1/30s
- f/5.6 1/500s
Most cameras let you set the auto exposure bracketing up to +/- 2 stops (i.e. covering a 4 stop dynamic range), however it is possible that when you take your light readings of the brightest and darkest parts of the scene, you exceed this dynamic range. For example a sky of 1/1000s and a foreground of 1/15s is 6 stops apart. In this case it might be easiest to just use the manual mode without the auto exposure bracketing and set the values yourself starting with the darkest and moving up in steps of 1, 1.5 or 2 stops (don't jump more than 2 stops between exposures).
- f/5.6 1/15
- f/5.6 1/60
- f/5.6 1/250
- f/5.6 1/1000
|
If you can't be bothered with all that messing about but still want to have a go, just put the camera in auto mode and set the auto exposure bracketing to +/- 2 stops and take your three pictures. This will covered the dynamic range of the scene in most cases. However also read the next box ... |
|
A Word of Warning...
Never use shutter priority when taking images for HDR processing. The reason is that in this mode, the camera will need to adjust the aperture to create the three different exposures, and doing this will alter the depth of field in each of the 3 images. This will lead to a poorer result (less sharp) when the images are combined. |
Step 2: Combining the Images
Once you have your 2 or more images, you need to combine them into a single image. By far the easiest way to achieve this is to use dedicated HDR software such as Photomatix Pro. This is a very popular piece of software and can be used to create quite subtle effects, or the outrageous vivid coloured images often associated with HDR photography.
Photomatix is available as a free download so you can try it out before you buy it. Some of the features put a watermark across your image so I would recommend you having a play first, and then parting with a little of your hard earned cash if you think it will give you the results you are after.
HDR software usually supports different methods of combining images together - the two most common being tone mapping (mentioned above) and exposure blending.
Whereas tone mapping takes a merged 32-bit true-HDR image, exposure blending works in a much simpler way (but can still produce spectacular results). It works by simply taking the highlights from the underexposed image, the midtones, from the middle image, and the shadows from the overexposed image. If anyone has ever attempted to do HDR images manually in a program like Photoshop or Paint Shop Pro, you will have probably tried a technique very similar to exposure blending.
We will come back to manual HDR creation later on (probably in section 9 or 10 after we have covered working with layers in Photoshop).
|
