3.5 Using hdri2tif on High Dynamic Range Images

hdri2tif uses a range compression algorithm(13) to convert high dynamic range images into displayable, low dynamic range TIFFs. All the important HDRI formats are recognized in input:

1. Floating point TIFFs as well as LogLuv encoded TIFFs.
2. Radiance files, uncompressed or RLE packed.
3. ILM's OpenEXR files.

Note that all formats must have 3 or 4 channels. The fourth channel is considered to be alpha information and stays untouched in the output image.

hdri2tif accepts the following options:

` -middle-gray n'

` -key n'

Sets the value of the middle-gray in the image, default is 0.18. Range is [0..1]. Specifying a value of 0 enables a histogram based automatic estimation (one can use `-verbose' to display the computed key value).

` -simple'

By default, hdri2tif uses a "dodging-and-burning" algorithm which is relatively costly(14). This option uses a simpler algorithm that is well suited for images with a medium dynamic range (lower than 10 "zones"). Note that this method may produce bad results on very high dynamic range images.

` -white n'

Sets the luminance value in the image that is mapped to pure white (1.0 in the low dynamic range output image). Setting this parameter to a smaller value produces more "burning". The default action is to set the white point to the maximum luminance in the image, which removes burning. This option is only meaningful when paired with the `-simple' option because the default algorithm automatically computes the white point. Specifying a value of 0 enables a histogram based automatic estimation (one can use `-verbose' to display the computed value).

` -gamma n'

Specifies a gamma correction to be applied on the produced image. Gamma correction is performed after the range compression algorithm.

` -nthreads n'

hdri2tif can run in a multi-threaded mode for increased performance. This option specifies how many threads to use. This option has no effect when `-simple' is used.

` -verbose'

Displays information while processing the image.

` -h'

Displays a brief help text.

The following options may be used to further control the "dodging-and-burning" algorithm. Modifying these options is not recommended.

` -nscales n'

Specifies the number of gaussian convolutions to use when computing luminance information for each pixel. More gaussians mean a more precise result but also a slower computation. Default is 8.

` -sharpness n'

Sets the sharpness parameter. Higher values mean sharper images. Default is 8.

The following options control the compression applied to the output:

` -lzw'

` -deflate'

` -packbits'

` -none'

Selects compression method to use for output TIFF. `-lzw' is enabled by default.

3Delight 10.0. Copyright 2000-2011 The 3Delight Team. All Rights Reserved.