Organic Level Design
My current project is a large tree that is intended to be a "level" of a simple game. Your character is mouse sized, and the goal is to collect various shiny berries and what not on the way to the top of the tree. One issue I'm facing is how to design a tree in such a way that you can get to the top by walking along various branches and vines. That's the fun part, and while I'd appreciate advice I'm relatively certain I can work it out on my own.
The other issue I'm facing, which I think I CAN address but not necessarily in the most efficient manner, is how to design game assets for a very organic world. Most of the tutorials and examples I've seen focused on developing a bunch of architectural assets, or at least blocky rocks and stuff, to build a level out of. Whereas with a tree a lot of things need to naturally flow into each other in a variety of non-standard ways. (Branches seem fairly doable since they at least can be simplified into individual branch assets that plug into each other, but vines need to stretch on continuously in random directions without any obvious breaking points.)
I'm also a little unsure how to approach UVs, since the scale is so big. Seems like the best approach is have the bark texture tile across the main trunk and larger branches, and then the UVS gradually shrink as the bark would naturally be getting smaller near the top.
My current rough (rough) blockOut is here. The little guy on the left branch is the approximate size of the player.
The other issue I'm facing, which I think I CAN address but not necessarily in the most efficient manner, is how to design game assets for a very organic world. Most of the tutorials and examples I've seen focused on developing a bunch of architectural assets, or at least blocky rocks and stuff, to build a level out of. Whereas with a tree a lot of things need to naturally flow into each other in a variety of non-standard ways. (Branches seem fairly doable since they at least can be simplified into individual branch assets that plug into each other, but vines need to stretch on continuously in random directions without any obvious breaking points.)
I'm also a little unsure how to approach UVs, since the scale is so big. Seems like the best approach is have the bark texture tile across the main trunk and larger branches, and then the UVS gradually shrink as the bark would naturally be getting smaller near the top.
My current rough (rough) blockOut is here. The little guy on the left branch is the approximate size of the player.
Replies
(I've been having problems with the ZBrush demo I'm using though. After subdividing a few times the model sort of splits apart in a way that isn't visible until you try to export it. The best advice I got on the Zbrush forum is "buy the 3.5 version" which I can't afford at the moment and doesn't have a demo. But I don't think Zbrush would be necessary in the immediate future anyway.)
Your direction should be focused along; what will the player be doing, what game play elements will the mouse be facing, and how will you be solve these issue? Then building the level to look like a tree around the game.
From the perspective of the mouse, the full scope and reality of a tree's natural growth doesn't exist. I've done a game from a bug's perspective. Nothing should be treated as we would see it.
1. I want to model a tree.
2. I'd like to model the tree in such a way that you can play around on it.
3. Should I start by modeling a tree that looks good as a "normal" tree asset for a normal sized game, or start with assets that would be meaningful on a mouse-eye-view-scale?
4. I have no idea where to START with assets that would be meaningful on a mouse-eye-view-scale, so maybe I'll make my preliminary goal to just model a decent tree with some basic ideas on playability.
5. You know, I should probably ask the friendly people on the internet for advice before I jump into this.
Which brings us to now. You're probably right about the way I should be approaching it. I think the main mental obstacle was trying to figure out how to get the player from the "ground" (where a realistic tree would have few branches) to the upper areas where there's enough crisscrossing stuff going on that you can actually have an interesting level. So perhaps the best solution is to eliminate the ground section and have all the action take place at higher levels, where I can focus on branches and large leaves that are actually workable as traditional assets.
However, I'd be equally happy whether my short term goal was to produce a tree that looks good from a distance, a low-poly but complete level that plays well, or a few leaves and branches that look really good up close.
Idea: Make parts of the tree hollow here and there so the mouse can run along the branch inside to safety when an eagle swoops down to try and catch it. Also maybe include ants on the branches so that the mouse has to jump over to get to the goods, maybe the ants eat away the twigs so it's time based too?? More than one danger = more fun to me as this could get very repetetive very quickly.
Considering all the gameplay possibilites should help form the layout for the geometry.
Lots to think about here, texture size is one but before you consider that, is this going to be 1st person or 3rd person perspective? I'm presuming you want to go 3rd person as most people would like to see the mouse.
Try making a leaf on a branch next to the mouse, zoom all the way in till your inside the eye of the mouse looking at the leaf, start off with say a 128 texture size and keep upping it till any blurryness has gone. This should give you some basic maths to play with as you size up relative objects and texture sizes for twigs, berries and branches.
Also, afaik a mouse doesn't have very good long range vision, playing on that idea try depth of field blur for the tree trunk and any branches further than two jumps away and make the upclose shots crystal clear. This will help provide depth and a bit of mystery maybe, it could also help the game run more smoothly.
Good luck with this, it's a nice idea and could be alot of fun so I'll keep watching
The actual game concept was less that you were actually a mouse (I just used mouse for reference) and more that the tree was one of those mythological world trees, and you were actually a person who was, relatively, mouse sized. The idea I'm leaning towards was actually making the game set in the Garden of Eden and you're trying to reach the top of the Tree of Knowledge. I have some ideas on how to expand that into a more fully fledged, interesting game, but for now I'm deliberately trying to avoid getting carried away with complex game ideas.
The project is, for now, going to be in first person because third person adds a whole set of complexities I'm not ready for (camera controls, rigging, animation, etc). Once I've created an environment that looks impressive and is fun to run around in, I may go on to add creatures, game mechanics and the like.
However, I do like the "hollow sections" idea of yours. I'm gonna play around with it somewhat after I get the other basic stuff done.
Raymond Arnold
raemon777@gmail.com
www.psycosmworlds.com
Moving around is standard WASD movement plus mouselook.
http://dl.dropbox.com/u/2000477/TreeWeb.html
-
My main concern, for now, is the obvious seams between branches. I've settled on getting the art done at a professional level as my immediate goal, and then finishing up the blockout of the level once I know for sure how the pieces are going to fit together.
So morale of my rant: Use textured quads with transparency for the leaves.
1. Building an interesting level for a game "design" that you haven't really thought out, is going to be very difficult...
2. If you don't want people to help you, why post your stuff? If you're doing this project as a way to build your portfolio to get into the industry, you should listen to the people who are telling you that you're going about it the wrong way.
3. Collaborate
4. 3rd person doesn't have that many problems... Just use the Gears camera or something... If your game design isn't complicated, the gears camera will work fine... and if it's not interesting, then it's going to be even less interesting if you can't see the main character...
5. If your level isn't fun to play because the game itself isn't fun, then you may as well have not designed a level for a game in the first place, and simply made a cool looking map in the UDK or something.
6. If you don't care, no one else can be expected to... and if you're not worried about making a fun experience, you shouldn't make games.
Just some points i wanted to touch on before going to sleep... i'm very tired, so if this is too blunt, my apologies...
eespecially the bowels
Fair enough, to some extent. I don't mind blunt, honest critiques (I prefer them actually). But I think that my third post clarified my goals somewhat and no one had given me any advice since then.
In the first few days I was trying out various different approaches because I wasn't sure what I was going for yet. I posted this thread before I quite knew what I was doing because I wanted some input on which direction to take it. Having gotten a better feel of the variables at hand, I think my initial post was poorly worded (and in some cases simply wrong). The main issue was calling it a "game" in the first place, when I've since made it clear I'm not really designing a game at all. (I am genuinely interested in game design and would love to sit down and plan out mechanics and what-not but once I start doing that I tend to lose focus on the graphics element, which is what I'm really trying to work on here. I may be overcompensating from my last project, but when I say portfolio I'm talking about graphics not level design).
The goal I've settled on (less ambitious but more appropriate given my timeframe) is to produce the assets that could be used to create a level/game/whatever, and to make a small walkable environment that simply showcases how those various assets might fit together in-game. Later I may go on and make an actual game or UDK level and if I do I'll be approaching it as an actual game or level as opposed to a bunch of art.
If people still feel this is a bad approach after that clarification, I'll reconsider, but it seems to me there are enough issues to warrant some experimentation before settling into an actual level design.
I will do that. Which brings up a few other questions. Assuming I was making an UDK map (I'm not, at least not for now, but I think it's a decent reference point) what's a good tri-on-screen-count to shoot for? Right now I'm taking the same branch assets and scaling them significantly in various directions, which results in much higher densities in some places than others, but I don't know what density I should be shooting for in the first place.
It's completely and totally dependent on the type of level, the type of assets, and the camera.
For instance -
if you have a first person shooter level, then you have a lot of reference to use from Unreal and the likes.
if you're doing a 3rd person camera game (ala Gears) you have a lot of reference there too...
If you're doing a heavily forested area, then the amount of overdraw and size of the triangles plays a big roll.
Basically... build something in the engine. Then figure it out from there. Post screenshots of your level and the framerate... if you start to see dips then you know where your problem areas are, and you can compare it to how different it is from the rest of the map.
It's a process...
Over the past few days I got some more concrete ideas on how an actual "game" (as opposed to just a navigable environment) might work. It's still intended to be very simple - you capture little glowing Will O' Wisps, and then use them to power magical abilities that help you progress through the level. (I went with Wisps instead of berries because they give me a lot more leeway on where to place them in the level. Making them mobile also increases the challenge of catching them).
Each level you can find a rune inscribed somewhere on the tree that shows you how to use a particular color of Wisp to do a particular spell, and then you'll have to use that spell to open up the next section of the game. For the first level (all I'm really concerned about getting done for now) the spell you learn is a simple light that you'll need to explore the inside of the tree.
I'm still focusing on more on developing the graphics than the actual game, but this does give me a better framework on how to design the environment. Some various technical things I learned/changed:
1. Used a much simpler leaf shape, with transparency around the edges, as was suggested.
2. Last week I was assuming I would want several different trunks with crisscrossing branches to maximize the player's ability to move. What I realized yesterday is that because the environment lends itself to a lot of clutter (and few big walls to occlude areas, hiding unused geometry), I should use one big central trunk that blocks about half the map at any time. I'm trying to make the explorable branches in a radial fashion that encourages the player to focus on smaller areas at a time. Right now the average polycount seems to be 30-50k, jumping up to 70k when you can see the whole tree, and works fine on my computer, but I haven't added everything I intend you yet. (Then again I also haven't implemented any way to reduce geometry at a distance either)
3. I'm still concerned about disorienting the player. The environment fog helps distinguish between near and far branches, but I still sometimes get confused about where things are (and I built the level, so I can only imagine what a new player would go through). I think a wider variety of textures on the branches will also help. In the meantime any other suggestions would be appreciated.
The current webplayer is below (again, does not work with Google Chrome yet. Sorry)
Unity Web Player - TreeWeb
Web Player is still available at:
http://dl.dropbox.com/u/2000477/TreeWeb.html