ne approach I like to use to extending the dynamic range of an image is the technique know as HDR (High Dynamic Range). This approach has earned a bad reputation, mainly because it has been over applied to to point of lurid surreal “Uncanny valley”-ness, but it doesn’t deserves this rap.
Behind the method is a simple idea, you can take a set of exposures (to capture the gamut of light and then combine the different exposures into one. This is not exactly new, some folk like Ansel Adams and other zone system proponents and the multi-exposure approaches like those of Gustave Le Gray, have been making composite and darkroom modified images back in the film, enlargers and wet chemical days. The difference now is most of this tonal manipulation is done in software. There is the conventional image merging facilities by way of layers in many packages but what is general called HDR is a more mathematical approach.
The methods begins by taking a series of different exposure of the same scene. In most DSLR camera this can be achieved by using a bracketing of EV values and probably three images will be taken together. (see the set taken for this series of posts) The intensities are combined pixel by pixel across these images.. Rather than just use and average intensity of the intensity of matching pixel, the simples approach might use the standard deviation of the nine adjacent pixels of which that pixel is the centre to adjust the average intensity.Thus making a better allowance of the local variation (ie sharp changes in intensity, or edges of objects) and tending to preserve local detail. (actually there are a real range of mathematical methods to do this “averaging”) The process doesn’t end there because trying to reproduce this wider gamut of intensities onto most forms of display or printing will involve reducing the intensity range again to suit that device or display. If those intensities are just arranged linearly back into the reduced range for reproduction, the image becomes flattened and “boring”, so there a number of methods, similar to the tonal adjustment curves in photoshop, lightroom and many other packages, that can be applied, These tone mapping algorithms take the original exposure range (and local adjustments, such as the standard deviation of intensities as described in my example) and refine the new intensify to be used for each pixels.Tone mapping can basically allows the highlights, midtones and shadows to be adjusted smoothly to a pleasing rendering of the image. Because there are a number of methods (and sliders to set the amount of adjustment) there is a lot of personal judgement that can be applied in the tone mapping step. This is why the “one setting suits all” and the “move all sliders to the right” approached tend to produce garish and disturbing images. Good tone mapping requires constraint and an understanding of how you want the image to look.
A lot of the photos posted in this blog are based on the HDR approach, and most folk don’t even notice, Used sensibly the tone mapping and HDR approach can bring out the detail and lift the photo to be closer to what you actually remember seeing.
The most conversational and probably the best introduction and advice of using HDR comes from +Trey Radcliff on his stuck in Customs website’.
(BTW You don’t have to buy photomatrix to do HDR, but you could if you want to follow Trey’s methods exactly. I’ve had the image above prepared for me by google+ AutoAwesome and will be discussing using Picturenaut, another “free” alternative in an upcoming post)
No comments:
Post a Comment