Comparison of various high dynamic range scenes processed with three methods.

Note: All the images on the page have been processed with a Gamma = 2.2.
Noise artifacts might be due to jpeg compression
We added a few images from Fredo Durand and Julie Dorsey "Fast Bilateral Filtering for the Display of High-Dynamic-Range Images" , and Raanan Fattal et al. "Gradient Domain High Dynamic Range Compression" both published at SIGGRAPH 2002. For these images we are not certain of the processing steps and correction used by their authors.

Column 1 -> Histogram Equalization. Greg Ward larson et al. "A Visibility Matching Tone Reproduction Operator for High Dynamic Range Scenes", 1997
We used pcond (part of Radiance) with the human sensitivity function (-s). The radiance values in some scenes are not scaled to real values (for example the scene "Belgium" has a maximum luminance of 0.6 cd/m^2) in which case the human sensitivity might not give an appropriate result.

Column 2 -> Adaptive Logmapping. Presented in this paper;
A default bias value of 0.85 was used for all images.

Column 3 -> Photographic Tone Mapping. Erik Reinhard et al. "Photographic Tone Reproduction for Digital Images". SIGGRAPH 2002
For every scene we used the local tone mapping operator with dodging and burning. The key value was automatically computed, following the recent paper "Parameter Estimation for Photographic Tone Reproduction", published in the Journal of Graphic tool. We used the source code provided on Erik Reinhard's website.

From "Gradient Domain High Dynamic Range Compression" Web site.

From "Fast Bilateral Filtering for the Display of High-Dynamic-Range Images" Web site.

From "Gradient Domain High Dynamic Range Compression" Web site.

From "Fast Bilateral Filtering for the Display of High-Dynamic-Range Images" Web site.