Re: Lens test charts for digicams


"Don Stauffer" <stauffer@xxxxxxxxxxxx> wrote in message news:Uf7%f.7$lb6.3295@xxxxxxxxxxxxxxxxxx
SNIP
ISO resolution chart

http://www.graphics.cornell.edu/~westin/misc/res-chart.html
SNIP

Wow! Thanks. The ISO chart is just what I was looking for. I had never seen a free site on it before, I had just looked at ISO's own site and sure didn't want to pay 93 bucks for it!

A word of caution though...

Most people don't use/interpret the chart elements as intended. The bi-tonal chart elements, e.g. the hyperbolic 9 bar elements, are not very suited for quantification of resolution of sampling systems like digicams. And the are not meant to, they're there just for a visual clue (and aliasing detection). The sharp edges of the elements also skew the quantification due to their high spatial frequency components.

The accidental alignment of the vertical/horizontal elements will skew the results, and a fraction of millimetre offset will in that case change the result (especially if a camera uses a mild anti-aliasing filter). The chart is also mostly testing horizontal/vertical resolution which would give e.g. Fuji sensors a benefit due to their 45 degree rotated sensor layout.

Finally, the chart needs to be shot at fairly short distance, when most lenses are designed for longer distance performance.

IMHO a much better target to shoot is a sinusoidal grating. To overcome the limitation of magnification factor dependency, I created a version that's insensitive to shooting distance and allows to measure at many rotation angles in a single shot. It also reveals lens aberrations and camera shake if present, and it is a test of your printer capabilities, which would go unnoticed when printing the ISO target.

You can make your own target from the following files at home with a decent inkjet printer.
For HP/Canon inkjet printers (3.8MB):
<http://www.xs4all.nl/~bvdwolf/main/downloads/Jtf60cy-100mm_600ppi.gif>
For Epson inkjet printers (5.3MB):
<http://www.xs4all.nl/~bvdwolf/main/downloads/Jtf60cy-100mm_720ppi.gif>

Print it at the indicated ppi without printer enhancements on glossy Photopaper which should produce a 100x100mm target, and shoot it with your (digi)cam from a (non-critical) distance like between 25-50x the focal length, or more. This will probably ensure that the target will 'out-resolve' the capabilities of the lens, and thus easily those of the AA-filtered sensor.

The resulting "blur"centre diameter is a measure of "on-sensor resolution" of the whole optical chain (lens+AA-filter+sensor), and can be quantified as cy/mm after calculating "(60/pi)/diameter". The diameter can be expressed as number of pixels multiplied by the pixel pitch.
If you want to compare to different sized sensor arrays, all you need to do is relate it to physical sensor size, i.o.w. normalize magnification differences to same sized output.

It will also reveal certain Raw converter issues (e.g. false-color aliasing artifacts) as it approaches the limiting resolution.

Bart


.

Relevant Pages

  • Basic newbie question on cmos sensor and optics
    ... Sensor A: ... Pixel size: 5.6um x 5.6um ... Given the same target illumination, and given the same area of view (roughly ... Obviously using two different custom designed optics. ...
    (sci.optics)
  • Re: FIR filtering in the Fourier domain
    ... he had "tracked" a stationary targed from a moving sensor. ... There's your position of the target. ... He gets several independent measurements, ... In the stationary sensor scenario, on the other hand, nothing ...
    (comp.dsp)
  • Re: Erwin Puts on AA filters (in his M8 review)
    ... for Bayer-pattern reconstruction algorithms? ... Unfortunately his version had a wrong gamma, and the centre of the pattern was not avaiable for torture. ... For HP/Canon inkjet printers: ... The benefit of a star target is that it also allows to determine the limiting resolution of lens+sensor, ...
    (rec.photo.digital)
  • Re: RS232 read with xpc target
    ... The sensor sends back ascii data terminating with carriage ... connection between the target pc and the serial device ... received when you purchased xPC Target has both 9 and 25 pin ...
    (comp.soft-sys.matlab)
  • Re: Mini Attack Swarms
    ... I'd question what a swarm system could do that existing weapons ... If you can't see a target you've got to sense it. ... Having a mobile sensor may be useful, but it may be easier just to ... you can hit it with existing weapons. ...
    (sci.military.moderated)