Adjusting Image Appearance based on Ambient Conditions

Content is mastered in a dim environment and reproduced on a display whose performance is measured in a dark environment. But a lot of content is watched in environments that are not dark or dim. As a result, images may not look like what the content creator intended. A paper presented by Praveen Cyriac of the Universitat Pompeu Fabra, discussed a new method to that can help predict how the gamma of the display can optimize the contrast based upon the peak luminance of the display and its ANSI contrast ratio. The goal is to provide the colorist with a fast and automated grade for use on-set or in a grading suite for HDR to SDR conversion.

It is well known that the brightness, contrast and colors on the display and the surrounding ambient luminance levels affect our perception of colors and contrast. These are described as color appearance phenomena and include Bezold-Brücke Hue Shift, Abney Effect, Helmholtz-Kohlrausch Effect, Hunt Effect, Simultaneous Contrast Crispening, Helson-Judd Effect, the Stevens Effect and more.

One method to adjust the image to varying ambient lighting conditions is to adjust the gamma. Cyriac and team created a model that allowed users to adjust this parameter until they had an image that they liked. This was then correlated with the display’s minimum and maximum luminance levels, surround luminance, ambient luminance, and ANSI contrast measured with a 4×4 checkerboard pattern.

The experiment consisted of two surround environments:

  • Office room: ambient illuminant of 47 cd/m² and average near surround luminance of 65 cd/m²
  • Dark room: ambient illuminant of 0.3 cd/m² and average near surround luminance ≈ 0 cd/m²

and three display types:

  • LCD: ASUS VS197D LCD monitor set to sRGB mode
  • OLED: Sony Trimaster PVM
  • HDR: SIM2 HDR47ES4MB monitor set to HDR mode

Twenty tone-mapped versions of HDR images over a range of image types were presented to subjects, who were asked to adjust the gamma non-linearity via a scroll bar to achieve a pleasing image. Analyzing the data revealed that the operator-adjusted gamma value can be predicted by knowing only the display’s ANSI contrast and maximum brightness level (Table 4).

Cyriac 1

The image below shows that a similar image can be produced on various displays in different lighting conditions using this method. The number in parenthesis is the ANSI contrast.

Cyriac thinks this method can be useful on-set by using the formula to adjust the gamma of on-set SDR monitors that are being used to preview HDR content using the RAW output format of the cameras. This might also be useful for realtime grading of HDR content end-user display tone mapping as well. – CC