The Fifth Bell Tolls But Not For Me

Two more demos down. One still to go. I have to say I've been impressed with my picking skills this time. No complete duffers yet. 

On the other hand, there's only one out of the five I've tired so far that I might buy when it comes out, that being Hawthorn. Not every game can be for everyone, even the good ones.

I was going to write up both demos in the one post but to no-one's surprise, I'm sure, I've ended up saying so much about the first there's no room for the second. It was pretty amazing I managed to cram the first three into one post on Thursday. I'll try and get the final pair into one more post next week.

The Fifth Bell  (38 minutes - Not Wishlisted)

Playing this was an interesting experience but precious little of that interest came from the story or the characters, which is a bit of a problem for a game that so obviously styles itself on the great narrative-led, character-driven point and clicks of the past, particularly a certain very well-remembered series from the 1990s. 

I'll start there. I'm used to adventure games wearing their influences proudly, as badges of respect and honor. There's nothing wrong with that at all. In the case of The Fifth Bell, though, it sometimes felt as if the primary influence might be  The Da Vinci Code, not Broken Sword. 

The demo begins where, I assume, where the full game will too, with the player character talking to himself as he rides his motorcycle down the hill into Strasbourg in a beautifully animated scene that really didn't work at all for me, for a couple of reasons. 

For a start, it's so lovely to look at and such a surprise at the very beginning of the demo that I didn't really take in anything I was being told. Consequently, when I arrived in the town square, I had no idea why my character was there or what he was supposed to do next.

Secondly, although I've never been to Strasbourg, I was under the impression it was a fairly large city. As you can see from the screenshot, the introduction makes it look like a small market town, as you approach it down an empty, country lane. It does look more like a city when you get there but it's still disconcerting.

All of which brings me naturally on to the graphics, which are by far the most striking and appealing thing about the game. That could also be a problem because this is one of nearly twenty per cent of all demos in the current Next Fest that come with an AI warning. In this case, AI was used "for the 2D background art, character sprites, and audio". All particularly problematic uses for many people.

I read an informative and revealing article on AI by Rob Fahey at GamesIndustry today. It makes a number of telling points about the dubious utility of AI, evidence against which is beginning to mount up now many companies, large and small, have had a year or two to try it. We're nearing the moment when the promises made for the technology are either going to be broken or fulfilled and it seems more likely to be the former than the latter. 

Fahey also observes that, even if the utility is there, it will come at a cost that might be more than most businesses will be willing to pay. Not only are the AI companies beginning to ramp up their charges in an attempt to claw back some of their vast investment but opposition from the end users, gamers, to any use of the technology at all seems to be both increasing and hardening. 

Those two factors combine to make the whole affair seem much less attractive than it did a year ago. Then, the worry would have been being left behind in the gold rush; now the safe option looks like sticking with the tried and true.

None of which necessarily impacts a game like The Fifth Bell, which looks to me as though it might be the work of a single developer. For someone making their own game, the attractions of automatically generated art and sound must seem extremely enticing. 

And the results are mildly encouraging, in a way. As I said above, the visuals are the best part of this game. The scenes are pretty to look at; well-composed and coherent. I suspect they're also mostly AI-generated, even in their final form.

The Steam AI proviso says they were "extensively edited, cropped, and manually integrated by hand", which initially makes it sound as though the end result was mostly the work of a human, until you realize what it actually means is that someone took the AI-generated output, tidied it up a bit, trimmed it to size and added it to the game. 

I'm not sure how rigorous the editing can have been, either. I spotted one error that certainly should have been caught in that editing process but wasn't. The game is set in 1994, when the currency in France would have been the Franc. The text of one puzzle correctly asks you to get hold of a one-Franc piece to make a call from a payphone but the menu boards inside and outside the cafes show all the prices in Euros, a currency not in use until almost a decade later. 

In fact, if I was going to be really picky, they also show what look very much like 2026 prices, not even the correct prices for the earliest date the Euro would have been use, namely 2002. It just shows how careful you have to be if you use generative AI and how much clean-up work you could end up doing.

Perhaps the most obvious warning sign, though, is that, as I suggested in a previous post, I could somehow sense the AI in the screenshots on the game's Steam page even before I read the disclaimer. Once I got into the game itself, that sense that something was subtly off intensified.

Would it have put me off playing, had the story grabbed me more firmly than it did? No, I can't say it would. The pictures might feel a bit bland but they're not unattractive. Plenty of hand-drawn games have art that looks a bit wonky to me so it's not an aesthetic deal-breaker. If anything, found the odd, sidling, diagonal movement of the main character, presumably not the result of AI, more disturbing than the slightly flat backgrounds.

  

Leaving the visuals for a moment, what about the sound, for which AI was also at least partly responsible? Here I found the artificiality harder to ignore and less worthy of a pass. 

One of the core requirements of an adventure game of this stripe is convincing, engaging voice work. The Broken Sword titles are the gold standard. I can hear George and Nico's voices in my head even now and Mrs Bhagpuss, who hasn't heard them since the 90s, still occasionally imitates them in conversation for comic effect. The voice acting in that series, and in several other adventure games I've played, often does as much of the heavy lifting as the plot or the puzzles.

In The Fifth Bell, the dialog isn't all that inspiring to begin with but the vocal interpretation sometimes drags it down a little further. It's not bad, as some human voice acting I've heard in games has been. It's mostly just a bit flat and unconvincing.

The thing is... the voice-work here isn't very good AI. There were a handful of minor line misreadings that I would say were typical of AI, which I would have thought, once again, should have been dealt with in the edit. And it wouldn't have been hard. I've heard - and indeed created - more convincing speeches generated by free online resources. 

After I'd finished playing, I copy-pasted a chunk of my own prose into Suno and had it create a spoken-word version, just to see if I was being over-critical. Suno did a better job on the first attempt and a much better job once I'd tweaked it a bit.

That only took me about a quarter of an hour, most of which was spent listening to the output. I don't think it would be hard to produce some convincing voiceover for an adventure game using AI. On the other hand, I'm sure I could do a better job myself, just reading it aloud, and so could most people, I'd have thought. I'm not sure voice acting is a part of the creative process that really needs much automation.

The parts of the game that apparently don't have AI at the back of them are the story, the gameplay and the mechanics. The last of those is easy to dispose of: the mechanics are solid. Nothing much wrong with them at all. Everything works, nothing is more awkward than the average adventure game, which admittedly isn't saying a lot because the entire genre is generically fiddly. I didn't come across any bugs or glitches.

Gameplay is absolutely traditional for the genre. Walk around, inspect things, pick up anything that isn't nailed down, talk to anyone who'll talk to you, do whatever they want you to do, solve problems and remove obstacles by using Item A on Item B... We all know the drill. 

I found all the puzzles reasonably easy to solve without a walkthrough. Most of the solutions were at least semi-rational although I think it's fair to say no-one in any adventure game ever made has ever behaved entirely rationally. The characters were quite engaging for the most part. The builder was amusingly aggressive, the girl and her disturbingly photo-realistic dog were charming, the waitress was suitably harried and irritable...

The plot is mildly involving. During renovations, someone discovered a modern cassette tape, hidden impossibly in a medieval wall in Strasbourg cathedral. On the tape was some kind of dire warning about not allowing a fifth bell to ring. I was never very clear what would happen if it did or why the character I was playing, Evan Marek, an archivist, was involved in trying to find out but I was willing to go along with it.

The biggest problem is the sudden start I alluded to earlier. Most adventure games begin with a fairly lengthy, slow set-up, during which you get to know the characters as they're slowly drawn into some kind of mystery. The Fifth Bell, in contrast, begins with a short voice-over and then there you are in the cathedral square in Strasbourg with not much of a clue why. I didn't find it to be a start that engendered much commitment.

The writing itself is a bit of a mix. As I said, the character dialog can be entertaining but the item and location descriptions are workmanlike at best. In general, it all feels a little perfunctory except when people are talking, at which point it sometimes seems like the writer might be having too much of a good time.

All in all, I didn't think it was a bad demo or that it's likely to be a bad game but even as a fan of the genre, I wasn't motivated to add The Fifth Bell to my wishlist. I already have a few point and click adventures on there and a couple more in my Steam library that I haven't gotten around to playing yet. All would seem to have more going for them than this one.

That said, based on the demo, I'd say it will probably be perfectly fine. If you're an adventure gamer who can't get enough of the genre and you're sanguine about AI usage, I'm sure you could do worse. That's damning it with faint praise but the demo, which I haven't finished and most likely won't, makes it feel like that sort of game.