Testing the one-shot autofocus accuracy of the Sigma 50/1.4 versus the Canon 50/1.4 on a Canon 5D.

The testing consists of multiple shots of a flat target at various distances, with the Canon 5D
mounted on a tripod. The general-shooting shots in the Conclusions page were shot handheld.
In all of the tripod mounted shots, the autofocus was reactivated for every shot. The shot was
taken after the beep indicated AF had been achieved. No use was made of mirror lockup or
timer/cable shutter release, as this would have slowed down the testing and was unnecessary.
Camera shake is not visible in these shots.

The testing of the tripod mounted shots was carried out as follows to simulate the range of
demands in real world shooting:
Before firing a sequence of shots, the focus ring was rotated to its extreme end of travel
at the infinity focus end. The shutter button was then half-pressed and held until
the beep indicated autofocus was aquired. The shutter button was then further depressed to
trip the shutter and acquire the shot. Two more shots were acquired without touching the focus
ring; in each case the shutter was simply half-pressed to acquire focus and then depressed the
rest of the way to take the shot. The focus ring was then rotated to the opposite extreme end
of travel (at the minimum focus distance end of the scale), and the sequence of three shots
described above was taken. Generally, two sets of shots beginning from each end of the scale
were taken, making 12 shots in total for a given lens at a given distance from the target.

All shots were taken with center point focus only, in one-shot AF mode.

All tripod shots were taken at f/1.4 and a fixed (and fast) shutter speed with manual exposure,
ISO 100, and white balance set to Daylight or Shade, as appropriate for the scene.
Conversion from RAW to full size JPG and 8-bit TIF was done with Canon's DPP version 3.4.1.1
at the highest quality level setting. Further 100% cropping or downsizing with subsequent
conversion to JPG was done in Photoshop Elements 5.0 using JPG quality level 10.

The DPP conversions used default settings in Standard picture style: No increase in sharpening,
contrast, or saturation. The downsizing in Photoshop used the Bicubic algorithm and a subsequent
application of Unsharp Mask, with settings (100%, 0.3 pixels, 0 levels).

With the test now described, please view the galleries for comparisions.

As always, please select "original" size images to view. Other sizes could appear
unnecessarily soft.