The part and the whole

The pictures on the radio, famously, can be more convincing than the pictures on the television. This is because they come out of our own heads: we supply what we think things should look like.

This module is based on something similar. Instead of supplying a whole, literal and frequently rather dull picture, we supply a mass of details, and leave the viewer to reassemble them into the whole. It's a bit like the archaeological reconstruction in 1066 and All That, where there purports to be a rather dull reconstruction of statue fragments by experts; a 'much better reconstruction by the authors' which is of course significantly funnier); and a blank space for a 'perfectly splendid reconstruction by the reader'.

The module was prompted by a rusty Citroen 2CV at (as mentioned in the free summary) a classic car rally. As usual, it was against a dull background. But the 'tin snail' is so well known that we only need a limited number of clues to imagine ourselves in deepest rural France, looking at one of these wonderful old vehicles.

Our first attempt

It's a simple enough technique, as the pictures above show. It depends mostly on just shooting a lot of pictures, preferably all the the same orientation, and enough of them to make a reasonable sized grid. You could probably get away with 2x2 (4 pictures) but more is better. We're not sure what the limit of the technique might be if you assembled all the pictures onto one sheet of paper. Probably with A3 (297x420mm) you could fit in as many as 20 or 24 images (4x5 or 6) or maybe even more. Here on the screen, we reckoned that 4x3 was about right; the images are 285 pixels wide. As a rough guide, we find we need to shoot 50 to 100 per cent more pictures of the subject than we are going to need for the final grid: it is equally easy to under-shoot and to over-shoot.

It cries out for digital, but equally, you could physically assemble a load of 10x15cm/4x6 inch or smaller machine prints, or of course make an enormous display from bigger prints; it's up to you. In fact, one thing we thought of doing was making a dozen small prints of the Citroen above and scattering them on the walls among all the other pictures in our séjour or living-room.

If you use machine prints, be very careful with your metering and ask the lab to lock the settings on the printer to match the best print. That way you should get even densities and the same colours. In the examples above, the large areas of blue and rust could provoke quite significant colour shifts, as the printer tries to balance the overall colour to 'average' -- which is not going to happen. For the same reason, use a fixed white balance with a digital camera, not auto white balance. Roger shot all the pictures in this module with a Leica M8, with various lenses. For the car shots, he used mostly a 50/1.5 Zeiss Sonnar, equivalent to 67mm on full-frame 35mm; with full-frame he'd have used 50mm or 75mm.

Our second attempt

Once we'd succeeded reasonably well with the car, we tried with a completely different kind of subject: the troglodyte village of Rochemenier.

This time we selected 20 pictures, 12 landscape (4x3, as for the car) and 8 portrait (2x4). The additional difficulties should be easy enough to spot. Colour balance is a lot harder to maintain, even outdoors where the sun was going in and out, never mind the mixed lighting of the interiors. We also made the mistake of cropping one of the pictures (top row centre, first set) so it didn't quite match. We've left the mistake in, just to show how and why it looks awkward. You'll see, though, that we have begun with outdoor pictures, trying to maintain some symmetry in colour and density, then moved on to a mixture of outdoor and semi-outdoor (the berries started out as a makeweight, but the colour worked rather well) before going indoors -- with one more outdoor in the middle, at the bottom, as a sort of reminder. You often have to play about with the order of the pictures quite a bit before they look right. It can be important, too, whether a particular picture element (a person, a tunnel) is going 'into' (towards) the centre of the matrix, or 'out of shot'. 




One of the great advantages of this scheme is that the person looking at the pictures is hit with so much information at once that they are usually less distracted by things which might stand out far worse in individual pictures: in this case, especially, the plethora of labels, especially in the bottom left-hand picture of the tools, above.

After making a considerable effort to get the colour balances to match, with signal lack of success, we adopted another approach, which was simply to make the colours look pleasing against one another, as in the bottom row above where we have the grey-green images on the outside and the warmer pictures on the inside. We also found that it was a good idea, in a given row, to keep the light as consistently to one side as possible, eg. from the left in the bottom row. This seems to be most important in the top and bottom rows -- though in a 2x4 matrix like the one above, you only have a top and bottom row. All of the pictures above are as shot, though there is an inevitable temptation to 'flip' the occasional picture (transpose it horizontally) to make it fit better.

An unsuccessful variant

The next experiment was another classic car rally, where we found that it is very easy to have too much of a good thing. We thought that by shooting details of a range of classic cars, we might create an impression of the rally. Well, we did -- sort of. The results are interesting enough, but as far as we are concerned, they lack the impact of the inital 2CV collection, simply because they are too varied. They might well appeal to a serious petrol-head, but they lack coherence as an impression of one thing. An interesting point here is that the order of the pictures, top right to bottom left, is the alphabetical order of the picture names (beginning with 'big headlight' and ending with 'vented bonnet').

We did however have enough pictures of the Jeep alone to make a 3x4 block. You can see that we tried a different sort of layout here, all 285 pixels wide, even the ones that are portrait format. We are not sure how well it works; we suspect that it would be better to stick with landscape format.

And then we thought: well, let's try a completely different approach, of colour for colour's sake (the subject of a paid module in its own right). We chose nine more car pictures that we had not used before, and...

This also gave us the idea of a grid of red only; or wheels only; or headlamps only; or car badges and labels only. It's interesting what you can do with a wretched excess of pictures.

The bottom line

There is little point in adding more examples, as it is a technique we are still developing ourselves, but we suspect that there is quite a lot of future in it, especially for travel photography on the web. But from sharing our successes and our failures, we hope that our readers may be inspired to try the same technique, and quite possibly do it better than we have. As we have said elsewhere, it's a poor teacher that doesn't want the pupil to do better than the teacher.

Go to the list of modules

or go to the home page

or support the site with a small donation.

© 2007 Roger W. Hicks