Can AI vision models process "Magic Eye" images? the demo video reminds me of text in SIRDS images.