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

4 Discussion

4.1 Best capture method for the various categories of image

In pursuit of quality, represented by good spatial resolution, colour and contrast ratio, circumstances arose with the capture of particular image groups which would apply widely to the capture conditions surrounding any similar image. When considering the results the images divide naturally into 4 groups:
  1. Microscope slide and low magnification microscope images (both captured using microscopes and artificial light).
  2. Plant, soil profile, landscape and portrait (3 dimensional subjects captured using natural daylight)
  3. Skull (3 dimensional subject captured under studio conditions using artificial light)
  4. Map and line drawing (both flat art examples captured under studio conditions using artificial light)
It is suggested that the four groups of images identified comprise most of the image types likely to be encountered in a higher education context.

4.11 High and low magnification microscope images

Photomicrography is a specialised branch of photography and familiarity with the operation of the chosen microscope, light source(s) and selection of lenses proved to be important as the camera which simply records what it sees through the component parts of the microscope. All adjustments (e.g. composition, focus and colour temperature) are made to the microscope and its light source. Capture devices are required to react to these conditions.

Capturing the image of the source material with a colour slide or negative requires the photographer to wait until the film is developed for confirmation of success, and lack of experience can produce surprising results (see Table 3.2). In this project no suitable captured images were obtained using 35mm slide film. The most predictable and instantly accessible results were obtained when capturing with the Rostrum Camera 2A coupled to the microscope. The image can be viewed live on a video monitor in full colour and, provided relevant adjustments are available, the quality of the image displayed can be maximised before final capture. Success with this path is assured if there is confidence in the quality of image being displayed by the monitor. Slide film is not as tolerant of changes in colour temperature as the human eye or negative film, and if the microscope light source is an unknown quantity or the bulb is getting old results can be unpredictable. For this reason results obtained using the video rostrum camera on the microscope were preferable to those on 35mm slide. Although the lighting was perfectly acceptable to the Rostrum Camera 2A this was no guarantee of correct colour temperature for 35mm colour slide film.

As all adjustments to improve image quality are made to the microscope or its light source, skill in working with microscopes as well as photographic skills are required. Delly (Delly J.G. (1988) Photography throught the Microscope. Eastman Kodak Company (London) pp104) recommends not using negative film because of the difficulty of achieving faithful colour in a print if the technician does not have access to an original image for comparison, whereas slide film which only requires development and no manipulation should faithfully reproduce original colours provided correct filtration is established and used. However, in the results reported here greater success was achieved with the negative film.

Delly also suggests that there is more control of contrast inherent in black and white film than in colour. This could help to explain why the lighting setup for the barley apical dissection, which yielded good texture and contrast in the video rostrum version, gave very poor texture and contrast on a 35mm slide, rendering it unusable. Again the colour temperature of the light source needed to be a known quantity when capturing on slide film because of that mediumÆs lack of tolerance to fluctuations. Possibly because of the quality of the slide, or possibly because of the picture content (medium sized bright subject against a black background) the PhotoCD scanner 2000 for 35mm material would not focus on it and it had to be transferred using the larger format 4045 scanner, necessitating the purchase of another and more expensive Pro Photo CD Master Disc. It is not clear why the PhotoCD transfer from negative is so much worse than the rostrum transfer from the print which came from the same negative. Possibly it could be due to the fact that thePhotoCD operator who only has a very small preview window to work from on his monitor, making it difficult to see the detail required. This again indicates that the shortest number of copying stages does not necessarily mean the highest quality.

4.12 Plant / soil profile / landscape / portrait - images captured using daylight

These four images may be looked at as a group because they required the same general range of photographic and imaging skills to achieve a good result. The fact that PhotoCD versions of these originals came out on top in each case tends to bear out this view because no unusual specialist skills (like those called for by photomicrography) were required. Qualitative scores of 5 abounded because general photographic skills and an awareness of what things probably look like achieved acceptable versions of the same image. These, although not identical, when viewed singly would probably be acceptable to a viewer with anything like 'normal colour vision'. Putting all high or low budget versions of an image on screen together drew attention in a rigorous way to small differences in individual versions, and that is how these examples were assessed. It is stressed that the judgements made were technical and not designed to evaluate the images for their effectiveness in teaching.

Turning to characteristics of specific images, the soil profile was deliberately captured on an overcast day to minimise shadow distractions for the viewer, so contrast was deliberately minimised. Interestingly, most of the prints from 35mm negatives of the soil profile were unacceptable because with no colour reference, the print machine operator failed to achieve the rather unexpected pink colours in this particular quarry. The only prints that were acceptable came from a roll of film which also included a grey scale chart image exposed out of doors, and the operator had clearly set up on the grey scale before printing the soil profile. Full use should be made of the low cost Grey Scale reference cards (as produced by Kodak) to take on location to provide this form of reference, particularly for subjects with which the developer is unlikely to be familiar.

The portrait was designed to test specific aspects of image copying. No fill light or reflector was used to relieve shadow areas on the face. The intention here was to show up a poor contrast ratio and in fact all the paths performed quite well in this respect. In addition the background chosen was deliberately textured and monochrome to show up any colour shifts more obviously than might appear on the (full colour) face itself. This facet proved to be more of a test than the lack of fill light, particularly with the frame of videotape. Because of the texture in the background the multi-colour shift on the videotape may be connected with moiré patterning.

4.13 Skull - image captured under artificial light

This subject was fundamentally different from any of the others in that it was three dimensional and captured indoors under multiple artificial lights. Conventional lighting units mounted on a rostrum stand for flat art subjects are not appropriately sited for the modelling required by a three dimensional subject. In addition to the capture methods established for other subjects, the rostrum camera was taken off its stand and the skull experimentally lit by a variety of different methods. The benefit of using a rostrum camera for this was the ability to white balance (line the camera up for colour) to the light source in use at the time.

One lighting method involved the use of two 800 watt æRedheadÆ tungsten lights giving a known colour temperature output of about 3,200 degrees Kelvin. In the small studio these lights were too powerful as they were, so one thickness of Neutral Density 6 filter was used on them to cut down their output by 2 stops viewed through the camera lens. The slides when they came back showed that a considerable green cast had been provided by the Neutral Density filter which proved to be too great to eradicate from reversal film examples, but was treatable in negative examples. Slide versions of this subject exhibited a range of lighting variations from very warm colour temperature using anglepoise lamps with domestic 60 watt bulbs, through green from the Redheads with (alleged) neutral density filters on them, to bland overexposure and lack of modelling from the anglepoises with Photofloods. At the PhotoCD transfer stage the colour temperature latitude of the 35mm Ektar negative film stock proved predictably more flexible than the 35mm Kodachrome 64 reversal slide.

Eventually the most flexible light source for modelling, as the image was not too large, turned out to be two anglepoise desk lamps with domestic 60 watt bulbs. The result of capture on the KY-F30B was the best version for contrast ratio, best equal for colour and an image which was superficially very good for resolution, but on comparison with the PhotoCD transfer of a 35mm negative it marginally lost out because its sharpness looked artificially enhanced - a not uncommon problem with electronically generated images. These two versions, the video rostrum camera and the PhotoCD of a negative, turned out to be significant because they illustrated the strengths of two good capture paths, at least for this class of image.

4.14 Map / line drawing images

These two images can be taken together because of their flat art similarities. Both were captured using the longer capture paths used for many of the other images, but in addition shortened paths were taken directly via rostrum camera and via scanner, bypassing the various photographic options.

With the map, colour and resolution proved to be particularly taxing, given the nature of this particular example. All paths coped at least adequately with the subtle variations of colour required but overall results were variable, largely because of colour temperature problems mentioned in Section 4.13. Post digitisation processing through Photoshop could cope with minor variations, but major colour shifts made it difficult or impossible to recover original hues downstream. The immediacy of being able to look at the original and adjust the quality of the displayed image before capture is likely to be a preferred option in most cases. However, this measure of control provided by the rostrum camera was slightly tempered by its innate lack of spatial resolution power and a surprising lack of saturation in some colours, particularly blues. Scores were particularly close between pathways for this image class, but two capture devices were fundamentally unsuitable: the still video camera would not focus to fill the frame with a small map detail and would not have been able to resolve its fine detail anyway, and the names on the map would not have been legible at the selected magnification when recovered from the videotape.

Similar reasons precluded the use of these capture devices for the line drawing. Here, possibly because of the presence of a known white paper background, the green tinge provided by the filtered lighting source proved obtrusive and insurmountable in the slide, whereas the flexibility inherent in the negative made the same lighting condition correctable.

The other interesting version of this image was provided by the Scanman Colour scanner. Opinion was divided on spatial resolution, one reviewer feeling it merited 5 rather than 4, but both agreed that the script printing in the lower left corner of the picture was clearer than any version apart from 35mm negative and PhotoCD. The Logitech scanner clearly outperformed the high budget HP Scanjet IIc on this feature.

4.2 Making a choice of capture system

From Figure 2.1 it is possible to identify 9 possible image capture paths, which vary in the number of equipment components required:
  • Equipment-light paths
    • Subject - 35mm Slide - PhotoCD - Storage
    • Subject - 35mm Negative - PhotoCD - Storage
  • Equipment-rich paths:
    • Subject - 35mm Slide - Rostrum Camera - Digitiser Board - Storage
    • Subject - 35mm Negative - Print - Rostrum Camera - Digitiser - Storage
    • Subject - 35mm Negative - Print - Scanner - Storage
    • Subject - Still Video - Digitiser - Storage
    • Subject - Videotape Frame - Digitiser - Storage
    • Subject - Rostrum Camera - Digitiser - Storage
    • Subject - Scanner - Storage
The equipment-light paths would require spending decisions based largely on a materials budget whilst the equipment-rich paths require investment based on predominantly capital expenditure. The latter keeps control within the hands of the university or department; the former may relinquish some control by utilising outside expertise for CD production while calling for heavier recurrent expenditure on per-image charges.

It is not possible to say that one system is the best for all applications. For example photomicrography benefits from the immediacy of a rostrum camera but this is not the preferred capture device for all other images. However, the system which provides consistently high quality for both the test charts and for a general range of subjects is PhotoCD, provided that original material is first captured on good quality slide or negative. There will be individuals who will prefer the control afforded by their own rostrum environment, but if use is made of the range of PhotoCD resolutions and advantage is taken of the large storage capacity of the discs, there is little that competes with it for quality and convenience at the present time. Kodak originally said that they would be introducing a scanner to capture hard copy (reflected light) images, and although prospective units have been tested none has yet been introduced into service. Kodak's expressed view is that the quality is not yet good enough. Currently the PhotoCD process is only able to cope with transparencies and negatives (transmitted light) from 35mm up to 4in x 5in.

Results from PhotoCD transfer, undertaken by a local company which leases the full range of the Kodak equipment, were predictably outstanding with some qualifications. It was found that being able to sit with the operator of the equipment and talk him through the transfers image by image was very beneficial. The operator himself suggested that he finds it easier to work with negatives than slides because of inherent contrast and colour dye flexibility in negative film stocks. It is unlikely that many potential users will have the luxury of working directly with the operator either because it is against company policy or because the cost of travelling is too great. Instead they will have to send their images away for transfer and thereby lose control of their valuable source material and leave quality issues in the hands of the transfer company.

PhotoCD could provide the answer to future proofing looked for by developers because there is more resolution in the file than the monitor can currently display. Creatively it is possible to consider comprehensive display of an image through the use of wide shot and any number of details with no loss of resolution. This course of action is also possible with a scanner, but there are implications for storage of large files, probably involving compression. PhotoCD has the benefit of requiring no extra compression beyond its own visually lossless routine. With a video rostrum camera flexibility is limited as the system requires the user to decide on composition of the image details at the capture stage. Zooming in after digitisation to recompose results in pixelation - proportional to the change in size.

Of the 5 different disc media used by Kodak for PhotoCD, the PhotoCD Master Disc for 35mm originals mastered on the Kodak Scanner 2000 can hold up to 100 images (some storage space is lost by transferring at different sessions) and the Pro PhotoCD Master Disc for 35mm up to 4ins x 5ins mastered on the Scanner 4045 stores between 30 and 100 images, depending on chosen image resolution and physical size of source image.

Comparing the results for the test charts in Table 3.1, of the serious contenders for digitisation (rostrum cameras, scanners and PhotoCD), the scanners significantly outperformed their respective rostrum cameras in terms of spatial resolution but less significantly in terms of contrast ratio. However when the general range of full colour images were assessed qualitatively in Table 3.2 and 3.3, these performance differentials were not maintained. The choice between scanner and rostrum camera is not so clear cut. What is clear is that PhotoCD is capable of at least equalling the best points of both alternatives. Often PhotoCD will have extra resolution in hand for display of image detail which might not be possible via rostrum camera or scanner.

Fortunately the high quality achieved by PhotoCD coupled with its ease of use makes it an ideal capture path for both experienced and inexperienced developers alike.

4.3 Critique of high and low budget equipment

4.31 Rostrum cameras plus their respective capture boards

Results from both rostrum cameras and their associated digitiser boards justified their selection. Control of composition, aperture, colour quality and contrast ratio were as expected at the camera end. With push button white balancing in the KY-F30B camera (camera 1) and simple but effective hue controls in the TK-1280E (camera 2) there was no need for extra colour correction, and enhancement would have been possible downstream using a software package. Experience showed that it was better to ignore the settings for saturation, contrast, brightness and red, blue and green offered by the Screen Machine II board. Standardisation on 51% for contrast, brightness, red, blue and green gave the benefits of a predictable starting point from which to set off into Photoshop. It was found necessary to increase saturation to 79% - again a standardised setting. These settings were achieved by trial and error since it was not possible to predict from the computer monitor screen what the outcome would be after digitisation. No such setting options were provided on the Captivator board. Instant access to results and no recurring costs per image complete the resources picture for rostrum camera operations. In Table 3.2 it can be seen that the results from Rostrum Camera 1 were not consistently better than those for Rostrum Camera 2 and consequently the large investment could not be justified on results alone. The benefits of Rostrum Camera 1 are ease of use, automatic line-up and controls designed for regular use compared with flimsy controls and time consuming set-up procedures with Rostrum Camera 2. Good results from Rostrum Camera 2 are heavily dependent on the skill and experience of the operator.

4.32 Scanners

The handheld Logitech Scanman was predictably more difficult to operate than the Hewlett Packard, requiring some dexterity and space to manoeuvre it. With some parts of images, particularly the black background to the dissection under the low powered microscope, the joins between scans were clearly visible. With some images, particularly the test charts, it either made a mess of stitching its passes together or was unable to find enough points of contact to do so.
Back Next Contents