The Mérode Altarpiece, by the Master of the Flémalle (widely identified as Robert Campin) in the Cloisters Collection of the Metropolitan Museum of Art is the earliest work claimed by Hockney and Falco to have been executed by tracing a projected image.
The following is based on:
The first thing to note is that the perspective throughout the Altarpiece is wildly inconsistent: numerous perspective lines that should meet at a point do not do so—not even close. Take a moment with a ruler and prove to yourself this fact. Some of these lines (floor, ceiling, bench, etc.) extend deep and straight into the picture plane, clearly not executed by tracing an optical projection. These facts show that optics were not used in the left and center panels. Moreover, the tableau is manifestly indoors, precluding the direct sunlight Hockney and Falco admit is required for the optical projection theory. Thus we can be sure that no optical projections were used in these portions (neither, moreover, do Hockney nor Falco make such a claim). But as such, it would be odd indeed if Campin were to invent the world's most complicated optical system, take the bench and its trellis work out into the direct sunlight, all to trace simple checkerboard of the trellis—Hockney and Falco's optical claim.
The first evidence adduced by Hockney and Falco was that the perspective of the front portion of the trellis differs slightly from that of the back part of the trellis. (They later changed their explanation to involve three portions.) But a very simple geometric construction showed that such trellis-size changes in "perspective" can arise from a very simple geometrical construction. Hockney and Falco then pointed to teeny breaks breaks in the trellis bars saying that it is unlikely that this was drawn with a ruler or straightedge.
There are numerous problems and unexamined drawbacks with their analysis. The first is the manifestly straight sections, even between the breaks. It is exceedingly unlikely that an artist tracing an optically projected image could trace those sections so straight in the dim, blurry image. More precisely, it is far more likely that the artists got them straight because he used a section of a ruler, straightedge or painters mahl stick.
Ashutosh Kulkarni of Stanford thought to check the other set of slats, from LL-to-UR. He found several, very interesting things. First, these slats showed not kink whatsoever along the break identified by Hockney and Falco. This lack of kinks is extremely unlikely had Campin used optics. Second, he found a set of kinks going back into the depth of the painting. The locations of these kinks are incompatible with the depth-of-field explanation by Hockney and Falco, that is, they are incompatible with the optical projection theory. In short, evidence of the exact same form (kinks in the depicted slats) that Hockney and Falco use to support their optical claim just as strongly refute it.
So, what might be the source of the kinks in the slats? Ashutosh Kulkarni discovered that a simple slightly kinked straightedge or very slightly kinked mahl stick—both well known from that time, of course—explains the relevant image data, the slight kinks, their angles and relative separation. In short, Campin using such a straightedge or mahl stick to guide his brush could explain the image data far more simply than his using the world's most complicated optical system.