Canon's white paper claims that AF accuracy under artificial light was improved in the 50D.
Improved precision over previous AF systems stems from each camera’s ability to adjust for different light sources while calculating focusing distance. Here’s how it works: during Phase Detection AF, the AF engine rapidly measures the density of horizontal and vertical details in a scene and the camera processor uses this data to determine where the sharpest edges and details are located. When found, the AF is locked on target. But under fluorescent lighting and some other artificial lighting sources, a rapid, imperceptible flickering of the light occurs along with sudden color temperature shifts. These tend to throw off the density measurements and therefore the accuracy of the AF calculations. In both the EOS 50D and EOS 5D Mark II, the type of lighting is taken into effect, and readings are averaged and processed with the help of the super-fast DIGIC 4 Image Processor. The result is improved AF accuracy and speed when shooting under lighting conditions that could fool the AF systems found on earlier Canon and competitive models alike.
I have a 40D and while I'm generally very pleased with it (AF included), I have noticed some AF inaccuracies under artificial light (especially fluorescent) where AF point was at one point but the actual focus point was elsewhere.
My question is simple: Has anybody noticed such improvement?