2ne1

HDR Experiments With Free Software

Photographs can only capture a limited range of light levels, whereas the human eye can percieve a large range of light levels, i.e. they have a 'high dynamic range' (HDR).

If you want to capture a photograph of a scene with a large range of light levels, e.g. dark shadowy areas and bright lit areas then you need to take a bunch of photos at different settings. Then you use HDR imaging software (I'm using this free Qtpfsgui) to convert the extended dynamic range represented across the series of photos into a single HDR image.

Monitors, being 'normal' dynamic range things cannot display an HDR image so you then need to remap that HDR image to one you can display on normal displays. This is the clever stuff kicks in as there are many ways an HDR image can be converted into a normal image (called 'tone mapping').

Below are the original images I used. Note that no single photo captures the best detail in the building and the clouds.

There are many a lot of different algorithms available in Qtpfsgui and each has different parameters to play with so there is a lot of experimentation that you can do.

Below are two quite different results, the first using the Fattal algorithm with parameter settings Pre-gamma=0.636, Alpha=0.1, Beta=0.8, Saturation=1, Noise redux=0.099. The second is the result of the Mantiuk with parameter settings Pre-gamma=1, Contrast equalization=1.005, Saturation factor=2

The only retouching that these have had is a little contrast enhancement. The dithering artefacts on them are from the algorithms; I don't know if they can be removed through better parameter settings.

Fattal method HDR
Fattal method HDR

Fattal method HDR

Mantiuk method HDR
Mantiuk method HDR

Mantiuk method HDR