Photographic Alliance of Great Britain

Depository of the Technical Standards Committee

Snippets Section

This section contains observations which update individual sections of published reports.

 

Report Section
[‘3IR’ = Third Interim Report]

Topic

Snippet

3IR. Pages 29-30, Sections 10.5-10.6 

Errata

May-06.  Note that the subsections in 10.5 and 10.6 are misnumbered. Version 1.1.1 with corrections is on this site in Reports section.

3IR. Page 25. Section 9.3.
Also, PAGB Exec Minute 06.14.2(c)

Use of ppi when submitting digital images for projection

Jun-06.  Only the pixel dimensions are relevant when preparing an image for digital projection. The ppi setting stored within an image file is not used during projection.

The Exec Minute suggests that embedding 300ppi in the image file, while not being required for projection, would help an organiser preparing a printed catalogue from the submitted images, as 300ppi is a common printing industry standard. This is not correct.

When an organiser is compositing a catalogue, s/he will set up blank white catalogue pages of the required size and at, say, 300ppi. When each image is moved onto the catalogue page, its own ppi is ignored and the incoming image is always sized with the ppi of the receiving page.

The role of ppi is very commonly misunderstood, and needs to be emphasised to all authors and organisers.

3IR. Section 6.5

Projector luminance

May-06.  Individual users of slide and digital projectors have been invited to participate in a wide-area research project to document projector luminance on screen. The results will be collated and published when available.

It is possible that a wide range of luminance values, as measured by EV, will be regarded as acceptable. Ideally, it should be possible to narrow the range of the lumen rating which may be recommended for digital projection in varying situations.

See a summary of replies to date.

PAGB Exec Minute 06.14.3

Federation Strategies

Jun-06.  By a note from the Honorary Secretary, Federations have been asked for a short commentary on their policy or strategy for introducing projected digital image events at Federation level. The results may indicate a timescale within which PAGB needs to consider projected digital technology for one or more events of its own.

See the current log of replies to date.

3IR. Page 14, Section 6.1; and Page 18, Section 7.2

Research on theoretical requirements for projector resolution

Mar-06, Comments Jun-06.  Paper from XRR PS (CACC) on projector tests with special reference to the theoretical requirement for adequate resolution.Comments added by mbs. Also table of projector resolution modes with comments on use of portrait images and use of square format to accomodate landscape/portrait equally.

Download a copy here (PDF 67KB).

3IR. Page 15, Section 6.5. Also Page 35, Glossary.

Units of Luminance

Update Oct-06.  The use of the various units associated with illumination is fraught with argument. The units commonly encountered include the Lumen, the candela and the lux, where the former two describe sources, and the lux describes incident light. Lux is used in building standards, where some suggested values are General areas 100-200 lux; Average reading 300-500 lux; Assembly/manufacturing 500-1000 lux; Very fine inspection up to 5000 lux. A sunny day outdoors is about 100,000 lux. Lux is a linear scale, whereas photographers generally work in a logarithmic scale. The increase from 500 lux to 100,000 lux is just over 4 stops.

A projector rated at 2500 lumen, assuming this output and perfect efficiency (unlikely), projected onto a 6’ wide screen has an image size of 2.5 sqm receiving 1000 lux which, with perfect reflection, would return 320 cd/sqm. This is within the range of 250-500 cd/sqm quoted as typical emission from LCD monitors, so it should be suitable for projection even with a poor blackout. (LCD monitors are typically used in room lighting.) This would be a top estimate of illumination as lamps fade and projector calibration may mean underrunning. A good blackout should improve appearance, although it is not clear how much larger the screen could be for such a projector. A 10’ wide screen, for example, would have a top estimate of 345 lux or 110 cd/sqm.

For an exposure meter set at ISO100, these formulae are also published:

lux  =  2.5  x  2EV

lux  =  2.5  x  f2  /  time,  where f=aperture, time=shutter speed in seconds

Working backwards from acceptability to a working EV standard is the subject of the ongoing PAGB research project. To date, acceptable results have been obtained when meter readings have been no lower than 6.5-7.0EV ie, 220-320 lux.

3IR, page 22-23, Section 9.1

Viewing and Workflow software

[Update Oct ]. As a summary of types of available software, they may be divided into two major classes. Organisers need to decide which class of software they prefer, and then which item within class.

Class (1). These are commercial programs mainly designed for viewing and cataloguing numerous types of media file. They may contain many features for capture, image manipulation, and out put of images eg to slideshows and the web. As such, very few of these programs’ features are actually required to run a competition or exhibition event. Nothing in the programs is specific to a photographic competition, but they can be used to store a set of images to be shown as required.

  • IrfanView. Free. Recommended by PSSA. No colour space management.
  • ACDSee. Colour space management only in the Pro version.
  • CompuPicPro. No colour space management.
  • iViewMediaPro. Good use of ITPC fields for cataloguing and searching. Uses colour space management.
  • BreezeBrowser. Good cataloguing. Uses colour space management.
  • ThumbsPlus. Colour space management in all versions.
  • Pictures2exe. Not an album type program, but may be used to assemble a show for straight-through showing. Supposed not to be colour managed, but reports of good results anyway.
  • others

Class (2). These are hobbyist programs specifically designed for managing photo competitions. They have specific load and store functions for limited types of image file, and are optimised for showing an event with integrated scoring. Like all programs, they are ultimately inflexible, and there may be features of the organiser’s own events which are not accommodated by an individual program.

  • DigitalCompetitionPresenter (Farnborough). Requires MS.NET software. No colour space management, and Farnborough ‘rules’ expect image files to have no embedded colour space tag.
  • ImageCompPro (Maidstone). Resizes images. Colour space management is not based on any profile submitted with the image, but the HTML ‘colourfilter’ method can be used to configure Internet Explorer generally (used for display). Authors are recommended to submit in AdobeRGB space which would at least have all images with a common space matching a choice set by HTML in IE.
  • PhotoCompViewer (Marlow). Suite of programs managing authors as well as organiser. Resizes images. [Update Oct-06] Authors had been recommended to convert to sRGB before submission, as v3 (current) was thought not to be colour-aware, and colour space management was expected only in v4 (winter 2006/07). Recent information from a Microsoft source is that the display engine used by PhotoCompViewer means that v3 is already colour aware.
  • [Update Oct-06] CCComp (XRR). Suite of programs written in VB, including one to assist with submitting entries. Resizes larger images. Colour management not known.
  • others

Full colour space management seems to be limited to the more professional commercial products (ACDSeePro, iViewMediaPro, Breezebrowser, ThumbsPlus). These may be preferred for events above Club level where there is less direct connection between author and organiser, and the author may not be present at the event.

At Club level, organisers will have to consider whether authors can be standardised on colour space, and whether the organiser’s own training capacity justifies use of one of the integrated competition products. Organisers should also consider whether they need or want the complexity of integrated scoring.

3IR, pp22-23, Section 9.1

Additional information about colour aware software from Microsoft (report from Tony Riley)

[Oct-06].  Refer to PhotoCompViewer in the previous snippet. The same source from Microsoft has described a range of their software which has been made colour-aware. The added feature has never been clearly announced to general users, and a complete list is awaited.

This software, relevant to photographic imaging, is now known to be colour aware:

  • Windows Office 2003, including Powerpoint Viewer 2003. The status of earlier Office versions is not known, but Microsoft notes that the Windows operating system has contained colour management features since Win98..
  • Windows XP - Pictures and Fax Viewer

This software may be colour aware:

  • ‘View as a Slideshow’ from Windows Explorer when the folder is defined as pictures. (This would explain certain test results obtained with equipment at Harrow CC when no difference could be seen between files with sRGB, AdobeRGB and no embedded colour space tags.)

This software is not colour aware:

  • Thumbnail view in Windows Explorer
  • Windows Internet Explorer (except that its property for colour space expectation can be flipped with an HTM:L script - see above re ImageCompPro)

Microsoft’s policy is that a file without a specific colour tag is assumed to be sRGB. And, that colour management is not implemented where speed of processing is critical. Full colour management requires the whole image to be recalculated, hence omission from thumbnails, and from IE where a colour space tag is not expected in images from the web. Enter    color management windows    (note spelling of color) to Google search, to see an informative series of papers by Microsoft, including the extended Windows Color System (WCS) facilities expected in Windows Vista.

 

About colour management. By Tony Kaye at Good Picture 2006 (RPS Imaging Science conference)

Capture devices may embed colour space information into image files in two ways. EXIF data includes a flag field to show colour space eg, sRGB or AdobeRGB. ICC profile is a complete specification of the capture. Either or both of these may be present in the image file, and if both then the software will give precedence to one of them.
A review of some cameras and of the software provided by those camera manufacturers showed that EXIF and ICC may be conflicting in a given capture file, and that manufacturer’s software may incorrectly process the colour space information and may output an incorrect flag value.
These are unlikely to be a problem for capture and downstream processing genuinely limited to sRGB. But users of AdobeRGB colour space during capture need to be aware that either their camera or their software will fail to recognise and/or process their files correctly. The astute user must become familiar with their equipment and software, and provide a correct workaround to ensure AdobeRGB space information is preserved if required.

The problem of failure to recognise and process for AdobeRGB is present in almost all photofinishing laboratories producing prints. Anyone using such services is advised to convert files to sRGB before submission.

[Note by MBS:  Using a Canon Pro1 and selecting AdobeRGB capture, files are named with a leading underscore  ‘_MG’ which distinguishes them from sRGB files named ‘IMG’. AdobeRGB files appear to have correct EXIF date but always open in Photoshop claiming to have no colour space setting. Workaround is to assign AdobeRGB on first opening in Photoshop.]

 

About image processing. By Elizabeth Allen at Good Picture 2006 (RPS Imaging Science conference)

Digital imaging files are extensively processed - sometimes deliberatately, but very often automatically. The user may be unaware of automatic processing. Examples include Interpolation (demosaic from detectors to RGB values; zoom/resize; any geometric change); Levels/Curves/etc (clipping; rounding; profiling for individual devices including display); and Filters (eg, noise reduction, sharpening).
Many/most of these processes involve some image data destruction with consequent loss of image quality eg by blurring or aliasing.
In principle, data destruction is never reversible, although some subjective improvement can be made eg, by unsharp mask.

Standards for Events

Submitting monochrome images

May-07.  Noted from the requirements published for the RPS International Projected Images Exhibition. Monochrome images must be submitted in the RGB colour model, just as colour images. Note that the FIAP monochrome rule does not require that the image be black/white: only a single tone/white.

This requirement needs to be added to the Guidance for standard B.01 in the PAGB Standards for Events.

Mar-08.  Suggested that Grayscale displays the same as RGB. Experiments to test this.

A colour image in RGB/sRGB (Mode/Space) was modified by Desaturation, and saved as jpg. This was the master for comparison with various Grayscale conversions.

The same image was converted to Grayscale, taking the author’s default Grayscale space of DotGain 15%, and saved as jpg. Further image files were saved as jpg using other spaces of DotGain 10%, DotGain 20%, DotGain25%, DotGain 30%, Gamma 1.8, Gamma 2.2. Side by side, all these image versions appeared the same within Photoshop.

Images were displayed using Thumbnail view (not colour managed), Windows Slideshow (colour managed), loaded onto an HTML page (not colour managed), and via ImageCompProv15.1 (not colour managed). Using all these display methods, and regardless of whether colour managed or not, all the Grayscale versions differed from the RGB version. The closest matches were DotGain 30% and Gamma 2.2. All others were significantly darker.

Examining a common test area within each image file showed that the K value reduced from 65% (DotGain 10% file) to 45% (DotGain 30%), when the RGB file showed 136 (of 255) equivalent to 46% density. Gamma 2.2 was nearly the same as DotGain 30%. Gamma 1.8 was nearly the same as DotGain 20%.

The experiment was repeated using tif format files, with the same results.

Conclusions: Grayscale and its associated spaces may, like CMYK mode be primarily intended for commercial print output. Although the space is encoded and readable by Photoshop, typical DPI display software, whether colour managed or not, does not seem to recognise the monochrome spaces and merely decodes the image data unmanaged. Image files for digital projection should always be in RGB mode, and Grayscale should be prohibited.   (mbs)