Illustrating eyewitness reports - a problem for cryptozoology

There are many famous sightings of "cryptids" whose fame is due largely to an impressive drawing of the creature. Some of these are made by witnesses themselves, while others are drawn under their supervision or even just from the text of an account. In addition to the skill of the artist, though, the accuracy of these images depends on how good a look the witness(es) got and how much is accurately remembered.
Mike Dash of Fortean Times just posted this blog, which I helped a bit on the research with, looking at several cases where an illustration, however well done, may not have been a very good match to what the witnesses saw.
I've worked with superb illustrators for my books, and they were excited by the challenge of drawing an animal based on text or old sketch. Some of the results, such as Bill Rebsamen's working of the Valhalla sea-serpent, have been superb and match other data very well. In the Valhalla case, two highly qualified witnesses watched the animal, wrote down a detailed description, and produced a very competent sketch Bill could work from. That's not always true.
Craig Gosling, who did excellent work for my first book, drew the two-finned cetacean reported by Quoy and Gaimard, but the witnesses never saw the animals' heads, and Craig had to make an educated guess at the head shape and the snout based on other cetaceans. The resulting drawing may (assuming the animal exists, of course) be very accurate or far off. If it's off, that's no detriment to Craig: he simply had to conjecture to fill in what was not reported.
On the other hand, Craig's drawing from that book of Steller's "sea monkey" has been licensed for two other publications and is probably the best we are going to get. Steller's description was a thorough one, written down by an experienced observer or marine life, so Craig's talents have likely given us something very much like what the naturalist reported.
I've rambled on here, but the point Dash makes is important. We have to consider not only the features shown in an illustration but what the illustration was based on. As Dash notes, Loch Ness witness Arthur Grant reported a weird animal crossing the road at night: his own drawings and those of others have gotten far more detailed over the years. So be careful. An illustration, however evocative and however well the illustrator knows animals and their anatomy, can be only as good as its origins.