Ever since the Olympus company announced that their new cameras could make images with more pixels in them than were on the sensor, I have been longing to try it out. The literature made it sound like they were lifting themselves up by their electronic bootstraps...
The trick needs a studio with constant lighting...which I can organise. It needs a camera on a sturdy tripod...my Gitzo Studex 5 certainly qualifies. It needs subjects that do not move at all...and my collection of miniature cars is ideal for this.
The function revolves around the fact that the Olympus engineers have made the image stabilisation mechanism on the sensor for the new OM-D E-M1 mkII move sideways and up and down by extremely small amounts - pixels - and with great precision.
This allows the camera to present a slightly different portion of the Bayer array to the incoming light over a series of exposures. Multiple images are put into the buffer with slightly different colours for each shot, plus a shot taken to allow the camera's computer to subtract electronic noise. Then they are recombined into one image that contains far more information than could be packed into it with one shot.
The multiple shots are not spaced out much in time. When I pressed the button down on the camera it was done in about a quarter of a second - the electronic shutter inside doing the 5 shots in that time. The lighting was simple tungsten modelling lamps from the Elinchrom studio lights and the camera was left to decide an effective white balance. After the shot went off it needed a couple of seconds to do the computer work inside.
Does it work? Yes it does. Here are two shots of the dashboard on a small Maisto model of a 1961 Volkswagen ( Yes children, dashboards were really that simple...no LCD screens and MP4 players... ) Shot A is the multiple detail shot while shot B is a standard exposure at 1/160 second.
One thing I did notice, though... when I shifted to the detail mode the manual aperture selection for the 25mm f:1.2 lens would not close down past f:8. Normally it can go to f:16, and this is desirable for the extra depth of field it gives. I suspect that it is deliberately limited to prevent diffraction effects through the lens masking the gains produced by the shifting mechanism.
This would make for a real balancing decision when doing close-up shots - they always need more depth of field. Fortunately Olympus also have another technical triumph in their camera to deal with that. Where the current pixel-shifting would be most use would be with setups at a moderate distance. or perhaps even with flat art copying of extreme detail.
Note: the only chance you are going to get to use this feature will be when you are using Pro-grade glass on the front.
