Qtpfsgui Reinhard '05 Notes

Test Image: Test1.hdr (Hass Burn images 357, 359, 361, April 2009 copyright DARW)

Initial Set-Up: As per Qtpfsgui Drago Notes.

Reinhard '05 Tone Mapping Operator

As in other notes, an empirical approach is adopted to study this TMO. The pre-gamma and Adjust Levels tools are available, as with other TMO's.

The user is presented with three slider parameters having selected the Reinhard '05 option from the drop-down menu. These are:

Brightness, default -10.00, range -35.00 - +10.00;
Chromatic Adaptation, default 1.00, range -1.70 – 1.30; and
Light Adaptation, default 0.00, range 0.00 – 1.00.

Applying default values renders a very dark image, clearly seen as stacked to the left-hand side in the histogram (available via the Adjust Levels command). Adjustment of the Brightness slider to about -2.0 pulls out the histogram to the right-hand side resulting in a natural looking image, although the dark tones are not well differentiated and saturation appears weak. At default values, adjusting pre-gamma to higher values (2.000) lightens the image, but contrast is not so good.

Using Brightness at -2.00, lowering or raising Chromatic Adaptation makes no changes to brightness, but there may be a slight change to colour balance.

Lowering the Light Adaptation slider to -1.00, blacks out all the light areas; raising it to +1.00, lowers the contrast.

Combining adjustments to the three sliders (Brightness -2.00, Chromatic Adaptation -0.60, and Light Adaptation 0.60) gives a moderately good realistic image which is enhanced by using the Adjust Levels tool to spread the histogram to the left-hand side.

Conclusions:

The Reinhard '05 TMO is fairly simple to use and can give reasonably realistic images. OTT or surreal images are not easy to produce, although it is easy to blow highlights etc. Typical starting values for realistic images would be a Brightness of about -2.00, a Chromatic Adaptation of about – 0.60, and a Light Adaptation value of about 0.60. The Adjust Levels tool is of more use here than pre-gamma adjustment. Note also that processing times appear to be fast on my system, a few seconds for 1024x1536 images.

Technical Points:

The Reinhard '05 TMO does not appear to be described as such in the usual sources (reference to Rienhard '04 and '02). It appears to be an update of a 2004 (hence the '04 term) publication by E Reinhard and K Devlin, “Dynamic Range Reduction Inspired by Photoreceptor Physiology”, allegedly in IEEE Transactions on Visualization and Computer Graphics, 2004, but actually published in vol. 11(1), 2005. Only the abstract of the publication is available free, and that is not very helpful. The Reinhard '04 algorithm had only two parameters, Brightness and Saturation.

Thus, apart from being related to the eye's light adaptation mechanism, I have no further knowledge of this algorithm.

Sources:

As per previous notes.

Qtpfsgui Notes - The Reinhard '05 TMO