The dynamic range of an image is the ratio between the maximum and the minimum luminance value.
The real world shows a vast range of light intensities ranging from star-lit scenes to white snow in sun-light.
Even within a single scene the range of luminances can span several orders of magnitude.
In such a case a photocamera is not able to recorder all the dynamic range, if you set the exposure for the highlights the shadows will be dark, if you properly expose for the shadows the highlights will be burnt. If a mean exposure is used, details are lost in shadows and highligths.



In this shot, as you can see, in order to properly expose the grass, the sky is completely overexposed.

A camera can produce only LDR (Low Dynamic range) images. Camera sensors are far away from human eyes. Monitors and printers are LDR devices too, so an HDR image must be compressed to a low dynamic range in order to be viewable.

What can we do to get a photo more similar to the world scene?
Is raw shoting a solution? Camera sensors are able to register 12 bit color depth, up to 4095 intensity values instead of 255 intensity values of a normal photo, jpg or tif.
Actually, 4095 is the upper limit, the maximum value depends on the sensor quality.
In any case is not so difficult to find scenes with a contrast of 1:10,000 or 1:100,000.

Anyway, let we suppose that our scene has a 1:4095 dynamic range and our camera can register this range.
In post-processing, we must fix the exposure to get an image wich is suitable for viewing or printing. But we can produce more images with different exposure values, someone good for shadows and someone good for highlights.
If we can blend these images saving the best of each one, we can get a result that is more similar to the real scene. We built an HDR image.

Raw shoting is not good for a very high contrast scene.
In this case, we can take multiple exposures.



HDR photo built in PhotoResampling using 5 shots at different exposure value





Following the 5 shots used to built the final image.



Using this technique in order to build the final image some constraints occur:

• Every shot must represent the same scene, the tripod use is suggested
  
  
• Scenes containing moving objects must be avoided, otherwise ghosting effects may be present in the final image
  
  
• The shutter speed must be modified, as aperure changing modify the depth of field too
  
  

Now we are: we got our images which all represent the same scene at a different exposure value. If we can blend these images saving the best of each one, we can get a result that is more similar to our eyes vision of the scene. We built an HDR image.

Our HDR image is ready, but how can we output it on paper or on a display? Monitors and printers are LDR devices!
We have to compress the dynamic range mapping the wide range to a restricted one. This compression is known as Tone mapping.
A good tone mapping algorithm converts the wide intensity range in the image to a lower range saving the original local contrast and details in shadows and highlights tones.


PhotoResampling is able to merge your multi-exposure LDR images to generate an HDR image for viewing or printing.