I'm playing Alundra (1997, originally for PSX). Bear with me as I jot down some disorganized notes, more for my own benefit than anyone else's (sorry!)
This video is from cubex55 longplay PSX series
I've heard people discuss Landstalker quite a bit as one of the better Zelda-likes from the 90's. As I was playing Alundra I oddly kept thinking about Landstalker, I think it must be because of the 3D navigation system which Landstalker seems to be famous for. The engine is graphically 2D tile-based (and has amazing pixel art) but the gameplay takes place on a proper 3D "big voxel" physics map.
This means you can go behind and on top of objects when ordinarily they would represent an immutable barrier.
It turns out that Alundra is a spiritual successor to Landstalker (again, which I haven't played) created by a lot of the same people-- interesting to me, particularly since I hadn't even played Landstalker but kept thinking "this is quite a bit how I imagined Landstalker!"
The 3D "big voxel" engine really is central to the way this game works, and it's a very nice technical feat. Even more so to consider that Landstalker works the same way and is on Genesis!
It's not hard to understand the developers' motivation for this. Most top-down 2D games at the time used a strictly 2D map for navigation, sometimes with hacks for traveling in layers.
Zelda: Link to the Past is a good example of a game using such a "layer hack" that you might be familiar with. In Zelda there is an element of 3D traversal, introduced really in the first dungeon and used nicely in the water temple (where the water level varies between each layer) but it can feel artificial.
Even with a layering system, the limitations of a 2D navigation system are quite noticable to players, who quickly learn you can't go behind certain structures where it looks like you should be able to.
I imagine this frustration with most games at the time was maybe what inspired the developers to build a proper 3D map system. It would have stood out and been appreciated I think by many people as a definite step up and something interesting and more true-to-life.
With Alundra, I feel they lean on this technical feature too heavily.
On the one hand, it's very satisfying to be able to properly navigate in 3D, for instance, being able to jump up on a table rather than walk around. But by making the game areas more complex much of this satisfaction is negatively compensated for.
A 2D navigation system is awkward because things that look like you should be able to go behind or on top of, you can't. But in implementing the 3D navigation system, the developers then created areas that were much more elaborate than you would see in other 2D games. Ultimately it's often hard to know apart from trial and error where you can or cannot go.
Incredibly, it's almost exactly the same problem!
There are quite a few jumping and other "tricky navigation" style of puzzles which rely on the 3D navigation feature. These are (to be fair) usually managable but not really satisfying, because of the amount of trial and error involved. Since the rendering is strictly 2D, there just isn't enough information by looking whether certain jumps can be made.
What might have been ultimately more satisfying would have been to structure things still in 2D, and have the excellent 3D engine be more of a salient point. In other words, have map complexity similar to existing 2D games, but allow the 3D navigation system to stand on it's own as an elegant and satisfying thing.
This is something I've been thinking about a lot lately; taking something very strong technically about a game and then leaning on it too heavily. It's tempting when you have something in place that stands out to try and build an entire game around it, but it requires clear thinking to see where to draw the line.
This could be true for any aspect of a game (or other creative form, for instance the author who leans too heaviy on vocabulary) but it's a trap that I think I fall into with respect to technical features.
I know in The Real Texas there are a few places I relied too heavily on the "big voxel" nature of the game engine in much the same was as Alundra (although there is no jumping in Texas.) (Aside: one of the reasons it's so interesting to play Alundra is how similar it is to The Real Texas in this respect.) I think overall I avoided this becoming too big of a problem btu the main offenders are surely certain puzzles in the garbadge mine.
Make sure the technical features of your game serve the greater idea, rather than distort it. If you have a fantastic physics engine, it doesn't mean that your game needs to be about that. Instead, exercise restraint and don't let one aspect get blown out of proportion.
Something I figured out while making The Real Texas (unfortunately only about half way through) is that maps based on simple geometry are more satisfying because they are easier to understand. It's true that a real-life path in the woods rarely runs straight, but having a straight path in a game map can be more useful because it better draws attention to those places where it diverges. A fork in a straight path could feel more important and interesting than one in a crooked path, because it stands out more.
Another area of map complexity to watch out for is density. Comparatively "empty" maps may seem like a bad idea-- but they aren't necessarily. If you overwhelm the player with too much detail, you make the world difficult to comprehend.
When you do art for a platformer, you typically want to make foreground and background elements stand out so the player understands what is a platform and what is just scenery. But you may also want to apply this technique conceptually.
Alundra's maps are very dense, and while they are interesting I don't really like the rhythm of them very well. I always feel like I'm navigating a complex maze, for instance even the main town has a very complex structure (fortunately well-connected so it doesn't become too burdensom.) This is exacerbated by the developers' decision to use the 3D big voxel engine in almost every situation.
As a broad point, I really believe you don't want to overload the player cognitively.
Instead, pace things out so that you have areas which are difficult to navigate interspersed with straightforward ones.
Also, think in terms not only of space and filling it, but of the rhythm of moving through it. If your character moves more quickly, then your maps may need to be larger.
One reason Castlevania: Symphony of the Night works so well is the sheer number of plain hallways. Platforming sections (again, often geometric) and in particular vertical ones that rely on jumping and jump-slashing are often connected by longer hallways that don't have much vertical movement. This creates variety and allows the focus to change from combat to jumping and back again.
In a top-down game, if the player constantly has to make short steps left or right along a simple pathway, it quickly becomes tiring. Instead make a straight pathway and put decorations or other points of interest alongside it for the player to explore if she wants to.
Please don't read this as some kind of review of Alundra! It's a great game, filled with magic and fantastic ideas, an original story and beautiful artwork. My purpose in this "Game Diary" is really just to jot down some observations, in case I want to come back and try and remember what I was thinking in the future.
If you found it helpful, that's great :) I've certainly appreciated being able to play it. I do reccommend it if you get the chance!