All you need to know about HDRWelcome to the HDR tutorial. This tute will start with the basics concerning high dynamic range imaging, then move onto the process of actually creating a HDR image using either photomatix or photoshop 9 or better, before finishing with an explaination of the finer points of HDR image manipulation and creation.
There is some assumed knowledge though - Metering, Manual Camera Control, Stop (a meter of light), RAW processing, histograms ...
If any of these dont make sense, I would suggest brushing up on those skills before getting serious about HDR imaging.
H D What?Have you ever taken a photograph of a beautiful landscape, only to have those rocks near you totally black and the sky portrayed as bright white? Likewise, have you tried to take a photo of a friend indoors, without a flash, only to have their face dark while the rest of the rooms is bright?
The phenomenon in question is dynamic range. Take your eye for instance. As you drive along a road at night, you might notice that its very hard to see when there is another car coming. This is because your eye is capable of seeing a range of light in one sitting. If the guy driving towards you was doing the same in daytime, its not nearly as big a deal. The daylight is much closer in light intensity to the headlights.
Certain mechanisms for sensing light have different abilities to resolve detail across great differences in light intensity. Our eye, as mentioned above, has terrific dynamic range. Even when pressed hard, such as the headlights at night scenario, the eye can still resolve detail on the side of the road. (Incidently, the eye does not have spectacular resolution, especially compared to some animals such as birds of prey. However, most birds dont have the overall range that we have (the ability to see ok in the dark and at daytime)).
Film is pretty damn good too. Black and white film comes out on top, with dynamic range near that of the human eye. B+W slide film is slightly less, with colour film less again, with colour slides being further down the tree.
"But what about digital" i hear the kids cry! Digital sensors, particularly those that are used in cameras, are pretty poor in terms of dynamic range. If youre particularly interested in the phenomenon and intricacies of dynamic range and cameras, then hit google, there is a lot of info out there!
Thus, your next posit should be "so how does this affect me?". The dynamic range of your sensor determines how much of your image can be correctly exposed. If you had a sensor with an incredibly short dynamic range, you would get very little of an image correctly exposed. If you had a mythically perfect sensor, you wouldnt even need to think about shutter speed - everything would be perfectly exposed! Of course, such a sensor doesnt exist, so you have to learn to make do with what youve got.
As you learn your photography, you learn all about exposure and exposure compensation in an effort to get all your images as full of detail as possible. As you get into it more and more, you might find out that shooting in a raw format gives you a little more control again over your dynamic range. After that, you might ask...Then what?!?!
So...back on topic please...HDR, sometimes known as HDRi, is an acronym for High Dynamic Range imaging. It is the name for a process that gives an image more dynamic range than was able to be captured in a single exposure. In photography, there are two main methods of developing an HDR image.
The first is the most easily accomplished. "Pseudo-HDR's" are created from a single raw exposure. The file is edited 3 or more times to create shots of different exposures in .jpg form. The images are then assembled and the HDR is created from there.
The second is a little more intensive, but also more flexable. The camera is mounted on a tripod, multiple exposures are fired, each a certain Ev apart, until all the light in the scene has been exposed accurately. The images are then taken to a computer, assembled and developed into an HDR image.
Both of these methods will be desribed in more detail below.
Just a'standin at mah camera...IF you are going to use the single raw exposure method, there is still no excuse for not exposing correctly. Remember to use the exposure compensation all cameras have to tell the camera what it is looking at. If you are metering off a dark surface, compensate to the -ve and if youre looking at a bright scene, some +ve compensation is called for.
Take 1 image and go to step-the-next
IF you are using the multiple exposure method, then you have a couple of things to think about.
The first task is to compose your image. Ideally, you will be on a tripod or a very stable surface. If you are doing this by hand, youd better be more stable than
a bloody statue!Unlike the single exposure method, you dont need to think about exposure. It is a good idea to look through the viewfinder and ask the camera what it believes the single exposure to be. For example, your camera may suggest that 1/80s and F/4 is about right, then it is best to commit that to memory. It is best to do this step with the intended F stop, so apature priority mode would be suitable.
Now you are ready to begin exposures. Set the camera to full manual control and select an appropriate F stop (such as the one used above) and set the shutter speed to 3 or 4 stops lower than the idea exposure above. In our example, we will start at 1/10 of a second at F/4. Expose. Increase the shutter speed a whole stop to 1/20. Expose. Increase the shutter speed another stop to 1/40. Expose. Continue exposing for each stop of light until you reach 1/640s, which is exactly 3 stops +ve to the mean measure. In the end, youll have 9 exposures (in this example), 1/10, 1/20, 1/40,
1/80, 1/160, 1/320. 1/640s for each, all at the same apature. Pack up, head back to the computer, youre done!
A little option for those who have cameras that support it is to use the "Auto-Braketing" feature found on some bodies. This can be especially helpful for situations that are time critical as it can reduce the time taken to adjust settings. However, the downside of this method is that you can only get 3 shots to work with, and are limited to a range of stops (such as the 5 stop range on Canons).
Step-The-NextNow is the time to plug the camera / memory card into the computer and download you images. Get a cup of coffee / tea / pan galactic gargle blaster handy, this might take some time...
Im a Photoshop userIf youre a PS user, and you have PS 9 CS2 or 10 CS3, now would be a nice time to fire it up. While you're waiting, it would be a good idea to note which image / images youll be using. Better yet, put them in a different folder.
(for the single exposure method) Get your raw image and adjust the required settings EXCEPT FOR EXPOSURE. When you are happy with the settings, adjust the exposure compensation to -2ev, save a jpg copy, adjust to 0ev, save a jpg copy, +2ev and save a jpg copy. You will have 3 images of 3 different exposures. Put the raw file somewhere safe and remember where the 3 jpgs are.
(for ALL methods) Go to "file > Automate > Merge to HDR...". You will be prompted to locate the folder where the images live, or the images themselves. You will also be presented with the option of "Attempt to align source images". You can tick this to allow PS to align the images to correct for tripod movement. If you have created the three images from a raw file, it is totally unecesarry. This process is VERY processor intensive and will take time. Once you have selected the appropriate images and options, hit OK and wait.
PS 10 CS3 may give you a warning prompt at this stage - just hit OK, its not important.
Eventually, you will be presented with a new window. On the left is a list of the images used in the image, in the middle is a preview pane and on the right are the compilation options.
Firslty, on the left, if you can see an image that is totally white or totally black, you can drop it from the compilation by unticking it's box. Once you have a range of shots selected, it is time to set the exposure cut-off.
The histogram on the right represents the tonal values for ALL compiled images. If there is a spike on the right, it means your exposures were not fast enough, and if you have a spike on the left hand edge of the histogram, then you have not shot enough images on the slow end of the HDR.
Underneath the graph is a slider. This slider indicates where the new white-point will be (256 on all colours in the .jpg format), so if you wish to include any blown areas in the image, then slide the slider down until the appropriate areas are cropped from the histogram. The response curves section is available if you have a file detailing your camera's response curve, otherwise the automatic setting is available (and works just fine).
Above all, dont be too miffed if the image looks flat, oddly exposed or overly grey. We will get to that shortly. Hit OK when you are ready and have another wait.
You will eventually be presented with the HDR image, in a 32bit form. You are able to perform limited tasks. If you are planning a B+W conversion through channel mixing, now would be a good time to perform that operation, but be aware that the image should still look flat and grey.
Once you have tuned the image in the available methods to your satisfaction, it is time to convert the file to a format that is useable. Go to "Image > Mode > 8 bits per channel." Photoshop will then present you with a dialogue box for processing the 32bit file.
I thought I already exposed it?Photoshop will give you 4 options for processing the image. These are explained below.
Exposure and GammaThe exposure slider adjusts the apparent brightness of the image, giving bias to certain samples of data over others to make the image brighter or darker. Adjust until the image appears correctly exposed given the contents.
Gamma is the adjustment of luminosity of middle tones in relation to highlights and blacks. The 0 and 256 values of brightness are left alone, but the middle values are biased darker or lighter to give an even and appropriate adjustment of mid-tone rendition.
Both of these sliders are set by feel more than any scientific method.
Highlight CompressionThis process alters the response curve in a manner such that the brighter or darker the tone, the more it is biased towards a middle grey tone. As a result, this processing method can looks very flat and grey, especially in images that were very contrasty to begin with.
Equalise HistogramThis process takes the tones avaiable and distributes them such that each tone from 0 to 256 is represented according to the original historgram. This results in an image that uses the full dynamic range avaliable, but might not be correctly exposed or suitable for the image - this can do bizzare things to the image.
Local AdaptationThis is the most powerful method, but also the hardest to develop. Each tone is defined by the apparent intensity of tones nearby.
The options avaiable are radius - which will define the area in which each tone will be compared to in order to be defined. A small radius can result in halos (which some people dont mind), a large radius will make the image appear more "normal". The threshold variable defines how much difference (in stops of light) are taken into account before the value is ignored.
Photoshop TroubleshootingWhy is my image so grey / lack so much contrast? - HDR images take a wide range of tonal values, then squish them into a smaller tonal range in order to include more detail. The idea behind HDR is for you to have more control over the tones in the image and, ideally, the subject. Dont be afraid to "discard" data from the image in order to draw attention to the subject.
The colours don't look quite right - Each tone in the image has been created from a number of pixels from each of the component images. Sometimes an HDR can accentuate digital noise or reduce it. The circumstances change for each image.
Feel free to post your questions below and ill amend this section of the tutorial.
Im a Photomatix userIn case youve never heard of photomatix, it is a little piece of software that was designed for some of the more specialist tasks in photographic development, including the processing of HDR's.
Photomatix has the ability to process raw (at least the canon raw I use). If you used the single raw exposure method for gathering images, then there is an entirely different process you can use (which i will explain on another day), although you can process the raw into 3 jpgs as in PS and use the method described below. If you took multiple exposures, then you are already set to begin.
Open the program up and go "HDR > Generate...". You will be presented with a dialogue box asking what images you will use in the HDR. Browse and select the images, either in raw format or in .jpg and hit OK. It will then present a dialogue asking for the options of HDR creation. Again, aligning sorce images will be a computationally heavy task, but can really help results. You can also ask photomatix to attempt to detect and remove moving objects (such as people or clouds). This is sometimes successful, sometimes not. It is worth a try. The curve profile is also available, but the first option will deliver results just fine.
Once you hit OK, it will begin to assemble the image. Toilet break time!
When you return, youll be looking at the HDR image. Before you exclaim how miserable it looks, realise that it is only showing 8 bit's worth of data from the middle of the exposure. The image is still a 32bit depth HDR image.
Now we can hop straight into the tone mapping process, which reduces the 32bit width down to 8bit for conversion into a jpg. Go "HDR > Tone Mapping". Here, you will be presented the options for converting the image back to an 8bit file.
Options options options!!!The following are for the "Details Enhancer" method.
Strength - Controls how hard contrast is obtained in relation to other pixels in the area.
Colour Saturation - Of course, the saturation of colours. Makes sense right?
Light Smoothing - How savagly light values are averaged in small areas. Can soften details and hard edges.
Luminosity - How bright the overall image will be.
Microcontrast - Controls how much contrast is boosted in local areas.
Microsmoothing - How much tones are smoothed out in local areas. Can work against mircocontrast or with it to create somewhat tonally soft images.
White Clip - How much detail is discarded from the white end of the histogram.
Black Clip - How much detail is discarded from the black end of the histogram.
Gamma - How bright or dark midtones are in comparison to the white and black points.
You can also save image preferences to use again on another HDR.
The following settings are for use in the "Tone Compressor" method.
Brightness - Obviously, the apparent brightness of the image. It does not shift the histogram a-la PS9 and previous, it compresses and biases the histogram so as to not loose data.
Tonal Range Compression - How much data is compressed into the selected brightness level. Lots of compression yeilds great details at the sacrifice of contrast, low compression has higher contrast imaging but sacrifices detail.
Contrast Adaptation - How much of the tones are biased towards the central grey on the historgram.
White clip and Black clip are as above.
The finer PointsOnce youve completed your conversion back to 8 bit, either through photomatix or photoshop, you can continue working on the image as you see fit. As you complete your processing, both during and after the conversion process, keep some of these points in mind.
- In photomatix and in photoshop you have the option of setting an upper (and in photomatix alone, a lower limit) to the dynamic range of the image, through the black / white clip (photomatix) or the slider in the HDR image compiler in PS. I would recommend ignoring this function at this early stage of the processing. After the image is converted back to an 8bit file, you can edit the black and white clip in either the levels or curves with slightly less accuracy but with much more range and, through curves especially, more control.
- In an attempt to maintain as much control over detail as possible, I tend to keep the image as close as possible to an ideal exposure. Remember that the histogram should reflect what you are looking at - a dark image will dominate the left of the histogram, and light image will dominate the right. If the exposure of the image needs adapting, doing it earlier in the process, while the image is still in 32bit form, will give you more data to accomodate changes in each pixel's definition, thus holding onto more detail and accuracy.
- You might hear people talking of "savage" and "natural" HDR's. A savage HDR is one that develops its tone difference in small areas, so that areas that are normally pretty high contrast will end up having halos around them as the dynamic range is adapted to each area. Natural HDR's appear somewhat as they did to the eye. Objects tend not to have a halo, but a more accurate representation of their relative brightness relative the the rest of the image. An example of a natural HDR would be
this and a savage one can be seen
here.If youve read thus far, I thankyou. There is a certain feeling of satisfaction when you get to divulge your knowledge for others. If you have created a HDR image, feel free to post below, I for one would love to see your stuff. Also, if you feel Ive missed any points, or have any questions, feel free to post below as well.
Enjoy!
HDR How To! {Beginner}, All you need to know about HDR
nice tutorial very helpful, alot of hdr tutorials normally just tell you how to take a few different exposures or combine them but not both, nice one 
