You want to measure how sharp your camera/lens combination is to make sure it lives up to its specs. Or perhaps you’d like to compare how well one lens captures spatial resolution compared to another you own. Or perhaps again you are in the market for new equipment and would like to know what could be expected from the shortlist. Or an old faithful is not looking right and you’d like to check it out. So you decide to do some testing. Where to start?
First you start with images that have been captured with very good technique. There are many tutorials on achieving just that but let’s go over the basics: stable location, oversized tripod and head, delayed release, mirror up, electronic front curtain if available, Manual Exposure and ISO, Manual Focus, VR off, NR off, 14-bit Raw, proper, properly illuminated subject at least 30 times the focal length away, base ISO, stay away from the 1/10s to 1/300s shutter speed range, ETTR.
The ideal subject is a back-lit razor blade, see Frans van den Bergh‘s or Jim Kasson’s setups for that. But for non-absolute, relative purposes such as comparing the performance of your own equipment a good resolution chart that can be printed at home by a decent printer on decent paper will do. A recent laser printer at its highest resolution works well, as do pro-sumer inkjets. The blacks should be uniform and 100% black, no dark browns. The paper should provide a uniformly white background, so gloss types are typically not a good choice because of the specular highlights they inevitably reflect. Soft matte papers are not good either, especially with inkjets, because of their visible rugosity and because they smear edges. A good compromise is quality multipurpose paper often used in office copiers/printers. It usually weighs 80 g/m^2 or more with labels such as ‘silky touch’ or ‘ultra bright’.
Lighting is important. Illumination should ideally be balanced and uniform over the target. Use a light source that provides close to ‘daylight’ color temperatures (D50, D55 or so), diffusely. By uniform I mean that it should not create a gradient of decreasing or increasing intensity across the chart. Both of these requirements can be tested by taking a raw capture of the chart with the illumination as set up and gathering statistics off four patches of pure white at the cardinal points with software like RawDigger.
Tape the chart to the wall. The performance of your imaging system will vary with distance to subject so make sure that you are as far away as your intended application requires. If, like me, you do mainly landscapes the camera/lens should be at least 30 focal lengths away, aimed at the center of the chart and perpendicular to it. Make sure that the angle of incidence of the illuminant does not bounce close to the camera after hitting the target. In other words, watch specular reflections – ensuring that the camera is not inside the family of angles of the lighting. Reread the checklist at the beginning of the post and remember to Expose the whites of the chart To The Right: somewhere between 8000 and 12000 ADU are ideal in a 14-bit raw file. Make sure there is no clipping in the raw data around the edge to test.
Focus is critical and focus bracketing essential (LiveView only is a distant fall back). Ideally you will mount the camera on a rail, position it in its center, focus with LiveView, switch the camera to manual focus mode and bracket several shots by moving the camera forward and back on the rail. If you don’t have a rail you can try to focus bracket manually by keeping the camera in one place on the tripod, focus in LiveView, switch the camera to manual focus mode and turn the focus ring incremental minimal steps one way then the other trying to cover peak focus. It’s very sensitive and not easy to hit it exactly, so in this second case you will probably have to fit the resulting data to obtain the actual peak MTFs.
If you do not want to do it yourself you can outsource the picture taking to reputable review sites like DPReview by simply downloading files captured by them. Alas, they often also miss the absolute peak, but hey.
Once you have a picture of the chart the best place to measure the full, unretouched spatial resolution information captured by the camera and lens that took it is straight from the raw file as it rolls off the sensor and electronics. No demosaicing, no processing, no sharpening (well, assuming no subliminal tricks by manufacturers under the hood) – just what the hardware saw according to the laws of physics, in order to judge its capabilities and IQ.
What about measuring it off the image of the chart as rendered by a raw converter, a TIFF or, heaven forbid, a JPEG? Read on.