One of the paintings adduced by Hockney in support of his optical projection theory is Hans Memling's Flower still-life in the Thyssen-Bornemisza Museum in Madrid. The following is based on:
Hockney does a perspective analysis of the front of the carpet and of the back half of the carpet and notes the central vanishing points differ. He attributes this to Memling building a projector, projecting the front half of the carpet onto his canvas, tracing this projected image, then refocussing the projector (to overcome its limited depth of field) and tracing the back half.
Note first that this painting does not have the "optical look" touted by Hockney—no heightened realism.
There are several other simple geometric tests that a passage—here the front of the carpet, and the back of the carpet—is in perspective, known to everyone who has studied the subject. It is quite surprising that the proponents never applied these tests to the painting. Specifically, parallel lines on the carpet, when extended, should meet at a vanishing point on the horizon. Well, such tests on the carpet show, with no doubt whatsoever, that such lines do not meet at a vanishing point on the horizon—not even close. This proves that the front of the carpet is not in perspective; likewise the back half of the carpet is not in perspective. Or of course the carpet itself is asymmetric, in which case the perspective tests of the central vanishing point are meaningless.
The proponents' lack of testing of these basic perspective lines is another example of selecting which evidence to consider in order to confirm a pre-determined "conclusion," rather than evaluating all the evidence, even if it means some of the evidence does not conform to one's pre-determined stance. This is a specific example of the more general drawback in the Hockney/Falco approach, where they claim an artist would have traced some portions of a projected image and not others but have never given any objective or principled method to determine which are which all based on the artist's unknowable and untestable "artistic vision." Thus if some portion of an image happens to fit an optical projection (which of course will happen much of the time even for paintings executed without optics), they can claim that this portion was done using optics. Likewise, if some portion of an image happens not to fit an optical projection (which of course will happen much of the time even for paintings executed without optics), they can claim this portion was not done using optics.