AGOCG logo
Graphics Multimedia VR Visualisation Contents
Training Reports Workshops Briefings Index
This report is also available as an Acrobat file.
 Back Next Contents


3.1 Quantitative results: test charts

The results of the quantitative assessment of the various image capture paths as measured by the Spatial Resolution (figure 2.2) and Grey Scale (figure 2.3) test charts are presented in Table 3.1. Table 3.1
Table 3.1 System Performance as measured by the Spatial Resolution and Grey Scale Test Charts

Unfortunately it was not possible to observe the spatial resolution chart with the Canon Ion Still Video camera because at 12" x 9" the chart was too small to fill the frame at the minimum focusing distance - an essential requirement for accurate quantitative results. The PhotoCD results are considered first and then the different equipment systems are reviewed.

Spatial resolutions achieved by PhotoCD transfers are misleadingly low where the whole of the test chart is displayed on the screen at one time (for a description of the PhotoCD system, see Appendix). The image derived from the print was marginally better than that achieved from the 35mm slide, 400 and 375 lines respectively, but significantly lower than the maximum of 800 TV lines provided by the chart. However, when the originals were examined the photographic print (and therefore the negative) and slide both resolved all 800 lines, and when the PhotoCD image was examined at its higher (photographic) resolution all 800 lines were present. Thus the limiting factor when viewing the whole chart image was the spatial resolving power of the display monitor.

PhotoCD images can be displayed in any one of 5 (sometimes 6) possible resolutions and if the whole image were to be viewed at the highest resolution at 1:1 magnification it would require an array of 16 (4 x 4) monitors to display it. Filling one screen with the complete image means that it is displayed at one sixteenth of its actual size, and at that size spatial resolution performance is limited by the resolution of the display monitor. However if the image is viewed at the PhotoCD photographic resolution (3072 x 2048 pixels) with only a small area of the actual image displayed on the monitor, maximum resolution is revealed. It was only when viewed at this higher resolution that the photographic versions (negative and slide) of the Spatial Resolution Chart revealed the maximum number of TV lines (800). PhotoCD versions of the Grey Scale Chart revealed the maximum number of half tones with no perceptible degradation.

For the various high cost and low cost system permutations the best results were consistently recorded by the Scanner 1 system which achieved 425 TV lines when capturing the Chart directly or a photographic print of it. Direct capture of the chart using Rostrum Camera 1 was slightly less satisfactory with only 375 lines resolved, whilst capturing a 35mm slide or print of the chart with the same system reduced the resolution to 350 and 325 lines respectively. The low cost systems delivered poorer spatial resolution from three of the six capture combinations (Chart - Rostrum - Digitiser; Chart - Neg. - Print - Rostrum - Digitiser; Chart - Scanner) and comparable results from the remainder (Chart - Slide - Rostrum - Digitiser; Chart - Neg. - Print - Scanner and Chart - Videotape Frame - Digitiser). The use of videotape frame capture pathways resulted in a significant loss of quality as measured in terms of spatial resolution compared with the other methods of capture.

The generally high results in Table 3.1 for the Grey Scale chart conceal the fact that while the number of half tones indicated were present as stated, they were often degraded. Progression along all capture paths did subjectively indicate a slowly worsening contrast ratio i.e. a loss of detail in either the very light or the very dark greys, or a loss of peak white or black. There were also cases where readings were less good at the top or bottom of a screen - an artefact not present in the original photographs of the charts which were all taken out of doors under an evenly bright cloudy sky. The higher cost systems all scored the maximum 11 half tones except for the scanned image of a print of the chart (10 half tones) and the still video and videotape frame (9 half tones). The low cost system tended to deliver fewer or degraded half tones when compared with the corresponding higher cost system, although for certain combinations of equipment the results were again similar.

3.2 Qualitative results: full colour images

3.21 Introduction

Two members of the research team worked together to agree upon a score which, given the variations that can exist in 'normal' colour vision, reduced the possibility for bias. When scoring the different versions of an image as many examples as possible were compared at the same time (see section 2.23). It was noted that when viewed singly many versions were quite acceptable, particularly for colour, whereas when compared as a group their individual shortcomings became apparent. A future development of the on-going research project will be to assess the various versions of the images in an educational context to see at which point quality falls to an unacceptable level.

Each image was rated on a five point scale (5 = Excellent) for sharpness (a subjective measurement of spatial resolution), colour (faithfulness to the original source material or analogue image of the source material when the original was not available) and contrast ratio (perceived range of half tones). An exception to this rule was the Soil Profile which was captured in lighting conditions deliberately designed to eliminate distracting contrast detail and therefore could not be rated for this attribute. The scores were then averaged and the results are presented in Table 3.2 of the qualitative scores for the 9 full colour images. Table 3.3 shows the breakdown of the individual scores. From left to right in each cell the figures indicate ratings for sharpness, colour and contrast ratio. The 35mm slides of the microscope slide did not achieve assessable quality and therefore there are no results present for this image in Tables 3.2 and 3.3.

Table 3.2
Table 3.2 Averaged qualitative scores for the 9 full colour images (^ average of 2 results, * images which failed to achieve assesable quality, # examples which were not technically possible) Table 3.3
Table 3.2 Individual qualitative scores for the 9 full colour images (^ average of 2 results, * images which failed to achieve assesable quality, # examples which were not technically possible)

3.22 Comparison of capture systems

It can be seen from Table 3.2 that, with the exception of the Microscope images, all the PhotoCD transfers equalled or exceeded in quality the images captured by other systems. From Table 3.3 it is evident that in general terms PhotoCD scores for sharpness, colour and contrast ratio slightly exceeded the results from the scanners and rostrum cameras. In terms of contrast ratio, the PhotoCD system was able to capitalise fully on whatever information was present in images and particularly the extended shadow and highlight details afforded by negative versions, e.g. the landscape.

The comparision of the results in Table 3.2 for the rostrum camera systems with those for the scanner systems did not conclusively indicate whether one system was preferable to the other, although the balance was in favour of the rostrum camera systems. While qualitative comparisons of sharpness in Table 3.3 were not as conclusive as the quantitative spatial resolution measurements in Table 3.1, there were examples which reflected the test chart results. The map, when captured by rostrum camera and scanner directly from a photographic print, came out sharper from the scanner than the rostrum camera. The map detail was specifically chosen for its ability to test spatial resolution with its sharp edged detail and lettering, whereas other images like the landscape had no such well-defined hard edge content.

One strange result is worth noting individually. When the line drawing was captured directly by the scanners - a short and theoretically high quality path judging from the quantitative resolution chart result - the images were poor for sharpness and contrast ratio. On examination it was found that sharpness was there, but masked by the poor contrast. This was possibly caused by the extremely fine lines in the original artwork compared to the lines on the resolution chart. Traditionally television systems have had difficulty displaying very thin lines, which this subject had in abundance. When this particular image was looked at in Photoshop and its contrast greatly increased without changing the brightness, it was found that the lines were clearly there but that they would not appear without greatly exaggerating the contrast, and this severely compromised the content of the rest of the image.

Quality of those images captured using an early example of new technology (still video camera), and older technology (single videotape frame) were particularly poor. The Canon Ion camera is an early attempt to capture still images directly onto floppy disc with no intervening hard copy photographic process involved. In theory this short processing path should lead to good results. In practice the technical shortcomings of this new camera have still to be overcome. Sharpness, colour and contrast ratio all leave a great deal to be desired in comparison with more established media even with general subjects which were not intended to be specifically challenging.

The other method of capture which failed to come up to scratch was the single frame of videotape which with hindsight was pressed into service in an area for which it was not intended. In comparison with other versions of the images, for instance those captured on negative or slide film and transferred via PhotoCD, this one was poor in all the identifiable criteria, as was the Still Video version. However taken by itself, and considering the physical dimensions of the videotape used (8mm wide) for this analogue signal, the quality of the resulting images was surprisingly high. Inevitably there was a grainy look to the end result which may not be as noticeable with a continuously moving image. This has implications for the re-purposing of videotape material, and it does not necessarily apply to all videotape. Beta SP tape, for instance, may yield better results than Hi8 and some moving images may be more blurred than others, also lenses designed for the capture of moving images are often designed to lower toerances. Examples have to be assessed individually on their merits.

3.23 Comparison of high and low cost equipment

In isolated instances the better scores achieved with the low budget equipment exceeded the worst scores achieved with the high budget packages. As high and low budget versions were not initially compared together (see section 2.23) examples were re-examined to ensure that consistent standards were being employed and the original assessments were confirmed. Overall average figures for the capture of slides via Rostrum 1 were better than the figures for Rostrum 2. However a comparison of the results between Rostrum 1 and Rostrum 2, or between Scanner 1 and Scanner 2, did not conclusively indicate whether one cost system was preferable to another. Comparison of output of the two scanners may be questionable because the default resolution for screen display of the HP Scanjet IIc is 75 dots per inch (DPI) and for the Logitech Scanman Colour it is 100 DPI. The plant taken from the photographic print scored higher for colour quality on the low cost rostrum camera than on the higher cost system. The soil profile print was judged to be better in colour quality when captured by Rostrum 2 than Rostrum 1. Soil profiles call for very subtle colour rendition and it may be that the automatic white balance on the Rostrum 1 camera was not as comprehensive on this occasion as a manual colour balance for this particular image on Rostrum 2.

When the low budget Rostrum Camera 2 was carefully lined up by hand for a single image the quality was better in all three criteria - colour, contrast ratio and sharpness - than the output of the Canon Ion, which is an automatic still video camera for which no operational line up is possible. The results from the Canon Ion and the Videotape Frames were consistently deemed to be less good and would probably be unacceptable when quality was a factor compared to the other systems.

Back Next Contents