The observation was that in some photographs the fiducials would seem to disappear in bright areas of the image. The hoax theory was that this occurred when foreground objects were photographically composited over background plates that already had the fiducials on them. The bright parts of the foreground object seemed to obliterate the fiducial selectively, with adjacent dark areas (most often part of the background) retaining the fiducial.
Among the photographic explanations for this behavior were such things as bleed and halation. Bleed occurs when energy migrates sideways from highly exposed portions of the film to adjacent parts of the emulsion. Even if the image is optically in perfect focus, this tends to make small details disappear. Halation occurs when light reflects internally from the rear of the emulsion layers, from the base, or both. If the incident angle is acute, such as with the aperture wide open or near the edges of the frame, the light reflects back to one side or the other of the incident point. The layer is then effectively exposed from behind in a way that blurs detail.
What we discovered in testing was that the film was robustly resistant to both bleed and halation. This is because Kodak knows what they're doing. Although the fiducials are very fine, we were not able to make any of them disappear entirely via simple photographic exposure changes and ordinary processing. While some dimming occurred, we could not replicate the effect in the subject images by manipulating the variables of mere photochemical photography.
Since the subject images were digital, we then turned to likely digital image processing techniques to see if they had an effect. They did. Reducing the image size decreases the number of pixels that can span the thickness of a fiducial. This made them more likely to disappear in various color quantization and compression schemes. As expected, lossy JPEG compression could eliminate a fiducial entirely in a bright area. This is mathematically sound too. Depending on quality settings, the sharp transition from very light to very dark to light again simply didn't make the compression cut. It was too high frequency a change to be expressible in a discrete cosine transform with given parameters. This was not a problem when the transition was between various grays and the black fiducial.
The one photographic effect we did not test was contact printing such as would have been employed to produce duplication masters. Naturally prints and scans made for the general public were not taken from the original camera transparencies but from dupe masters gingerly prepared from the camera originals. I have seen very high quality scans of the camera originals, which are still being systematically prepared for all the missions. Naturally Roll 39 from Apollo 11 had priority. The difference between scans of the camera originals and the typical convenience downloads was stark. There was much greater tonal range in the originals. The reason we are concerned here is that contact printing can suffer fine distortion at the edges where the light strikes the transparency at an angle and transfers it at that same angle to the underlying receiving transparency. Since each color element has its own layer, you often get small amounts of chromatic aberration in contact dupes if the light source is not properly collimated. For a fine detail like a fiducial, that aberration might be enough to make it insufficiently distinct in subsequent reproductions.
The real silly part of the hoax theory, of course, is that the proposed compositing method is too stupid for any actual process photographer to have actually contemplated. The last step in this process is to rephotograph the composited scene using a process camera. The obvious way to apply the fiducials would have been to put a reseau plate in the process camera, not the camera allegedly used to take the background plates. The "anomaly" that arises in the hoax theory requires the conspirators to have been quite stupid. So stupid, in fact, that many theorists change horses and say this was a deliberate attempt at whistle-blowing. And yet somehow none of them got caught doing it at the time.