Stories inferred from pictures or videos are not implicitly using language. Language comes into the picture only when a human viewer interprets what is happening and describes it in words.
E.g. if we watch an unnarrated nature video of a squirrel gathering nuts into its burrow, we can describe it with words and sentences. But those sentences don't actually occur in the video, implicitly or otherwise, and can say misleading things, like that the squirrel is thinking forward to next winter, and planning for having enough to eat when foraging becomes difficult.
The pictures in Xu Bing's book would have to be logograms rather than pictograms in order for the book to contain language.
A sequence of pictograms could be used to record a story (say of events shared by a group of people). Those pictograms could be used to stir their accurate recollection of the events, which they could talk about in language. But that's not the same thing as actually recording that language.