Superresolution — How to Quadruple Any Camera’s Pixel Count

Call it more of a marketing stunt — and Olympus is good at that. Fact is that the 40MP high-resolution mode of the new Olympus OM-D E-M5 Mark II seems to be a lot of fun to shoot with. If you’re care about maximum detail in landscape photography, this is a worthwhile alternative to a medium format high-resolution sensor that costs many times more. It’s the first time a consumer camera offers this mode. Here’s a tutorial how to mimic it with any camera, freeing any sensor from its inherent pixel count.

First off, DP Review explains the E-M5 Mark II’s multi-shot Hi Res mode in detail. If you want to mimic the technique, you’ll need a reasonably good camera, a reasonably steady hand (no tripod!) and Adobe Photoshop, to almost quadruple (!) a camera’s inherent resolution without softening really fine aliased details.

Superresolution -- an easy way to up-res any camera's pixel count by a factor of nearly 4. | Ian Norman / The Photon Collective
Superresolution — an easy way to up-res any camera’s pixel count by a factor of nearly 4. | Ian Norman / The Photon Collective
The class of techniques that enhance resolution of an imaging system, called superresolution, are not new. You may have heard of “over-/resampling” and “interpolation.” Some use PhotoAcute to increase resolution beyond a camera’s capabilities.

Photographer Ian Norman took the time to write an in-depth primer on superresolution that’s both a shooting technique and a post-processing method: Practical Superresolution Tutorial is a how-to guide that can keep you busy for many weekends to come.

Superresolution, says Norman, is actually a trick to maximize the amount of detail one can create with a sensor. It’s the future, says Norman. Resolution of a camera will no longer be determined by a sensor’s pixel count:

Whether it’s using sensor shifting, color filter array shifting, vectorized polygonal interpolation, some combination of these methods or others, superresolution will likely be implemented on every kind of camera from smartphones to DSLRs and compact system cameras. We’ll start seeing plenty of cameras that will be able to output images with more resolution than their sensor’s pixel count would otherwise indicate.

The main problem remains motion blur from moving subjects/objects. Faster processing performances will be an answer. And Norman voices the major caveat that often gets forgotten amongst our gear lust and resolution hunger:

Resolution, however is a single variable in the success of your image and in my opinion it’s a really low priority one. Lighting, composition and technique are all significantly more valuable to the success of a photograph than its pixel count. Just keep in mind that some extra detail on a roof almost 500 meters away doesn’t make your image a better photograph and in pretty much every contemporary medium on which you’ll display your photos short of extremely large prints, no one will notice the difference between a 12MP photo and a 100MP photo. Unless they look really really closely.

As always, one’s right camera is one’s right means to a right end. Shooting landscapes? You have every right to throw money into the camera you dream of. But try some superresolution stacks first. I myself am completely happy with my main, light-hungry 16MP camera. The other day I shot my son wakeboarding. Nice detail, even zoomed in.

Here is Norman’s video manual:

Must add that panoramas (GigaPan) would have the advantage of handling scenes with motion in them. As a welcome plus, superresolution elegantly circumvents lens diffraction, the physical limit of any lens. Even if you can resolve all the things on the sensor side, not every lens can.

(via The Photon Collective)
  • T N Args

    A fantastic promotion of what Olympus has done in-camera.

    • The technique will certainly improve within the foreseeable future, expect the next major camera releases to offer faster and more flexible Hi Res modes. The E-M5 Mark II is limited to a maximum of ISO 1,600 and a minimum aperture of F8 because
      beyond these points, the image becomes too blurred for the extra
      resolution to be particularly beneficial. But yes, again it’s innovator Olympus to introduce something that will be picked up by everyone else.

      • T N Args

        Unless they patented it. Wouldn’t that be awful. Also, it depends on a high-accuracy sensor position controller. Olympus have some patents on that technology too. Don’t count on this becoming a standard feature on many cameras at all.

        As for ‘faster and more versatile’, lets say in the next year or two they are able to do 10 frames in 1/60th sec (maybe jpeg at least), so you can hand-hold maybe a 24mm lens (to a fraction of a pixel for 1/60 sec? hmm, maybe not). The shutter speed will have to be *at least* 1/1000th sec, to handle 10 exposures in 1/60. Hmm, not so flexible after all, still need that tripod etc. So, if you like this, I suggest you buy it. There won’t be many clones, and Olympus won’t be advancing it hugely with faster processing etc.

  • I don’t think the Olympus way to increase resolution is the same as stitching multiple shots. On Em5II the sensor shifts one pixel left right up and down, and half a pixel towards the angles. It also eliminates moiree.
    Anyway, it seems that Olympus is innovating offering new features in camera.
    No doubt you can achieve similar results in post processing.
    Personally, once I’ll have it I’ll test it “against” my Dp2Quattro.

    • T N Args

      I think you can only achieve similar results post processing if you take multiple shots in the field. You wouldn’t want to give the impression that you could achieve similar results with a single 16 MP frame by post-processing it.