The idea I previously discussed about having encounters work "visually" (ie, the illustration shows a scene, and you pick a "response" -- the category you think will enable you to pass the challenge based on the illustration) didn't make a big splash with a certain influential playtester, and I think the most recent version will work fine with the simpler approach to challenges that we settled on. (The game is currently with that influential playtester, and we'll see what feedback his sessions produce).
However, in brainstorming this week, I had a thought for a micro-game concept with a Sherlock Holmes theme that could use this idea in a slightly different way. There would be three "decks" of cards (~4 cards in each), which show illustrations of three aspects of a crime -- e.g. the victim, the scene, and some locale that is in some way connected with the crime. To set up the game, you'd select one card from each deck. There are 8 "suspect" cards, and the game consists of examining the illustrations on the three "scene" cards, and deducing which of the suspects is responsible for the crime, what was his/her motive, and what weapon was used. This is done by identifying a common visual element that the three cards share, and connecting it to some aspect of one of the suspects. For example, maybe the three "scene" cards show fingerprints of soot, and upon examining one of the suspect cards, you see that he has soot on his hands, pinpointing him as the culprit.
The trick is that there's one and only common element for any three cards, so there's always a unique solution. And, the solution wouldn't always have to be as direct as having the same element on all three cards; there could be more difficult cases, perhaps, e.g. one card shows a train ticket, another shows a train schedule, and a third shows the train station, and you have to infer that these are all connected in a single chain.
To aid the investigation, something like "the interrogator" from this game would be used: it would be a "sleeve" (really two cards connected at their bottom edge), with an image of a magnifying glass on the top card with a hole in it; you place the magnifying glass over the item you want to examine, close the back card to make a "sandwich", then flip the whole thing over, and, through a hole in the back card of the sleeve, you get to read some text from the scene card describing in more detail what you see; e.g. it tells you the fingerprint that you're examining is made of soot, possibly suggesting the "chimney sweep" might be the suspect to focus on.
The game wouldn't need many rules, and probably would be reasonably quick to play, depending on the difficulty of the case.
Another idea that springboards off of this could be an actual Indy-themed microgame that's sort of a temple run, with the cards depicting obstacles in a lost temple. Maybe there are different pieces of equipment you can carry or acquire -- a pistol, a rope, a whip, etc -- and each has a sleeve. To attempt to solve a card, you select the equipment you want to use, take the corresponding sleeve, place the opening over the element on the card you want to target, then flip the whole thing over and the card tells you what happens. For example, targeting the sinister street tough with the pistol may work better than with the whip, whereas trying to cross the chasm by targeting the low-hanging branch with your whip will work better than the pistol. (And based on the construction of the sleeves, you could target the same feature with different cards and each can have a different effect).
I think microgames built around this visual mechanism may make more sense than trying to include them as an element of a bigger, more complex game. I think the latter would be easier for a non-artist like me to develop than the former, since the Holmes game depends so heavily on the artwork on the cards, and on the player being able to scrutinize it to solve the mystery. But I think that both could provide fun and unique gameplay experiences.