Skip to content. | Skip to navigation

Personal tools
Log in

WG1: Critical review and Insight into HDR Capture

Goal

To critically examine all aspects of capturing real world lighting and converting this to a digital format.

Objectives
To propose a file format for the most efficient representation for individual frames of real world lighting and to develop a quality metric for comparing the quality of the digital representation with the real world scene.

Methodology
Real world lighting may be “captured” as HDR content in one of four ways:

  1. Direct capture using specialized hardware: This hardware typically includes more than one CCD Sensor.
     
  2. Combining a number of LDR captures: Software is used to combine a number of LDR images (covering details from the darkest to the brightest parts of a scene) which have been captured by traditional camera technology.
     
  3. Virtually creating HDR content: Widely used in the computer games and entertainment industries, real world lighting may be simulated using physically-based computer graphics rendering algorithms.
     
  4. Expanding the dynamic range of LDR content: Software techniques are used to attempt to recreate the missing HDR content in LDR images. This is analogous to creating pseudo colour images from black & white ones and offers the significant potential of making the large amount of legacy LDR available to the HDR pipeline.

WG1 will analyse the limitations of each method in terms of digital accuracy (quantization), noise, dynamic range, and colour. The analysis of the required HDR data precision will be specific for each application. For example security or computer vision applications may require capturing contrast that is not perceivable by the human eye but useful for a given task. Many imaging applications, such as cinematographic production, require data whose precision is greater than what is normally visible to the human observer. Based on the capabilities of the capture process and the requirements of the applications, the Action will identify the optimum representation (colour space, colour gamut, maximum dynamic range, etc.) for high fidelity HDR imagery. This information will form a basis for a standardised file format or the extension of the existing format. If a single format cannot meet the requirements of all applications, a family of such formats will be specified.

Different capture techniques offer different precision and quality. It is desirable to have a common measure for the quality of the HDR capture process. The Action will develop the metric that will do this with respect to capture device (HDR camera) or computational algorithm producing HDR images.

Implementation
The major challenge for WG1 is to compare and contrast the HDR capture methods across a range of sectors and disciplines. The choice of partners in this WG will be crucial to ensure a sufficient coverage for each of the capture methods. The number of related research projects which this Action will coordinate is large and is thus a source of a large amount of key data. Other data sources will become apparent during the lifetime of the WG and through the Action’s Symposia and other activities.

The critical review and insight of capture methods will include:

  • Identifying a method and the field for which it was originally developed
  • Ascertaining the sectors the method is currently used in and their interdependency
  • Specifying the sector specific features of the method
  • Determining the stakeholders for the method
  • Specifying the method and its underlying theory
  • Fully evaluating the method with regard to its complexity, quality, data storage efficiency, range of lighting, colour, spectral information captured etc.

Expected Outcomes
WG1 will develop a systematically collected knowledge pool of existing HDR capture methods which clearly identifies:

  • The advantages and disadvantages of each method
  • The storage and representation requirements of each method and application, such as the required dynamic range, quantization error, colour gamut
  • The likely disruption that would arise if a method were to be replaced or improved

This knowledge pool will provide the key insights into what is required from a comprehensive file format for HDR image and video capture and whether an existing standard should be improved, or a new one proposed. It will also inform the development of the appropriate quality metric for HDR video. The knowledge pool will be available to all other WGs and will be made publically accessible to the wider HDR community.

Deliverables
WG1-D1: Proposed standard file format for HDR capture
WG1-D2: Quality metric for comparing digital representation of real world lighting with reality


INESC TEC - Laboratório Associado   European Partners