3D Game Dev... World / Level Building

Yeah @angrypenguin did. Thanks though. The level is already built now. I used proTileMapEditor.

 

I thought initially you were just talking about level layout/design and I was going to suggest using orthographic view, prefabs and snapping as neginfinity suggested.

Level designers would stongly beg to differ. In RPGs, MMOs and FPS games there are armies of level designers setting up levels perfectly all day long for 2-3 years. First working with primitive objects then later swapping out actual geometry with built in logic (prefabs).
The main difference I see is you as a sole indie are attempting to do it all in one step instead of mimicking a team process by designing iteratively.
I've also not done final level design with logic but I think its advantageous to keep iterative design principals in mind even when prototyping.

A simple check the (I think) is upright box would solve the problem as well.

This thread is very informative. I never even heard of IInteractive until you discussed it - though its not a foreign design concept to me. Thanks.

 

Oh yeah I completely agree... the level design itself is extremely important. I was talking about the actual workflow of building the levels in the scene editor. I am sure there are many people who find working in that editor very intuitive and perhaps even cool. For me I have a real disconnect with it. Not sure why really but I just do. However, using the proTileMapper asset it makes level building actually enjoyable as it should be.

In my posts I know I come off in a way that is probably different than how I am actually thinking. Just the loss when going with text. Basically, I just continually try to improve all of the processes I am doing (in game development or even things off the computer). It's mainly because I am continually striving to maximize the amount of productivity in a given amount of time. So anything that I think is just wasting my time or fighting against me I look for a solution for it and if there is not one then I write my own.

So, it's not that I am sitting here screaming at the screen "I DON"T HAVE TIME TO WASTE HOURS ON THIS BUNCH OF CRAP!" LOL! It's more of just a statement of "ya know, I really don't want to spend my time fighting with this stuff when I already know how to do it in other ways OR I at least know there has to be a better way that someone else has already sorted out".

Hmm... well that would have been a good thing... that simple check on the "upright box". Honestly, I looked all over in the inspector and never noticed anything like that.

 

Without addons in order to work in level editors, I had to make several "tile" prefabs, place them all into level, configure grid snapping, and then just keep duplicating them. That way you can utilize grid snapping.

The biggest issue is that placing any object into viewport by dragging it into view ALWAYS results in random coordinates that are sorta located in in the center of screen, but aren't tied to grid coordinates or anything like that. Meaning you drag a box into viewport, and instead of placing at some sane position like (0;0;0), editor will locate it at 3.18473,-26.8888,146.091, just because it hates your guts or something.

Oh, also, light baking is major pain in the ***. If you quickly build, say, something city-block sized out of several boxes, you have possibility that enlighten will choke on some big object, and if it doesn't, baking will take half a day. It is nearly useless.

 

Just went to the asset store (again!) and on the front page they have TileWorldCreator which looks pretty cool.
Normally $60 but is on sale for $30.

 

They must be reading the forum & seeing how much you are spending

 

I've been looking at ObjGrid3D for a long time, because the demo video looks so good. Haven't broken down and bought it yet though.

Have you watched the videos for isoRPG TileEd? Looks interesting if you get the packs it is designed to work with. It gives you full control to layout the maps how you like, but also chooses the prefabs randomly.
https://www.assetstore.unity3d.com/en/#!/content/49094

My advice to a newbie is to go for assets with the highest ratings and highest number of users and check the comments are recent and are not complaining about developers disappearing. The only time I broke those rules, I ended up with dud assets. Also unless you are rich, wait for the sales. They come around quite often.

 

Any chance we'll get to know if you like it before the sale ends?

 

Ha ha! Yeah I have decided to stay away from the asset store for a while. It's not that bad though I spent a little over $200 the past day or so. But I also wouldn't have spent about $50 of that if I had known which dang asset to buy to begin with.

Overall, I think it is reasonable considering I got what I was after a much faster & simpler way to build areas. I also grabbed various terrain and character packs.

Then I decided to stash those away until later. I threw out my first map that was using several of those assets and made a new map using just cubes. I want to keep the presentation simple and focus on getting the systems in place for handling interactions and such.

Here is my playfield in all its glory:



Looking forward to focusing on implementing some gameplay. Have some "bad blocks" sliding around, some collectible spheres and and so forth. Won't have much time the next few days but will work on it when I can.

Anyway, I need sleep. 3 AM here.

 

\Looking good, but a quick question, I'm assuming those are separate cubes, how do you define the mesh collider for these objects, or do you know care about this?

 

You can slap box collider on each cube OR just make one big invisible box at the bottom. Either way is not difficult.

 

Can you dynamcally create box colliders via code? And would the boxes have to be joined as one separate mesh for it to work.

 

Yes, you can. Just create BoxCollider component on some object via code.
However, usually you don't want to do that.
It is better idea to create colliders in editor, mark meshes as static and leave them be without moving them.

I do not understand the question.

Box collider is a box collider. You attach them to things as Components. If there's mesh on the "thing" you attach to, box collider will set its default size to match the mesh (IIRC only when you do that manually in editor). If there's no mesh, you'll get default sized box.

 

Wouldn't you want to do that if the whole level is dynamically generated though?

Maybe I worded it wrong. Perhaps it is a mesh collider you would need instead of a box collider, especially if you want to climb up the blocks like in the screenshot by garbenjamin.

 

Because occlusion baking, navmesh baking and lightmap baking goes out of the window in this case. Unless you "dynamically generate" the level via menu within editor and then bake.

Occlusion baking can produce siginficant speedup if you have a lot of objects.

You also want to mark anything that doesn't move as static - those objects have less overhead.
Also, moving static object is usually a bad idea (expensive operation), you'll see warnings in logfile if you do that.

Of course if you have very small scene, it doesn't matter, but if you start building castles with interiors, that'll be important.

You can have several children with colliders attached to object, you know that? Also those blocks might be just that - several blocks stacked on top of each other.

Mesh colliders are not a good idea, they're slower. If you use mesh collider, it is the best idea is to use convex hulls.

 

Yes sorry that's what I meant, the level is dynamically generated on the first frame, not every frame, on a separate note, does lightmap baking even work by work I mean usable, inside unity, I never tried it to say if it is reliable or workable.

I will remember not to use mesh colliders, thanks for this.

 

http://docs.unity3d.com/Manual/OcclusionCulling.html

Aw crap, I never even knew about occlusion baking, I've just been marking all objects in my scene to everything!

 

It sorta works, but it takes eternity and devours all the resources on the machine. Meaning you need monster of a computer to bake anything serious. Results - on static environments - can be quite good once baked, baked relatime lighting allows you, for example, change color of baked lights at runtime, and changes will propagate across the level - with light bouncing and all.

Dynamic objects in the scene require light probes, those are placed by hand and dealing with them is annoying.

 

Yeah, I haven't even tried area lights yet or baking, I don't want to lol. Slighly off topic, sorry garbenjamin, but why are shadows so poor in unity, I've turned to highest resolution, added soft shadows, gone into the graphics quality settings, I even added a DOF filter to try and fudge it.

What determines the quality, is it the graphics card, or what, I'm missing something I feel?

 

Well for the pro Tile 2 asset that Benjamin bought in this thread the same asset author has a mesh combiner asset too for almost $5.

 

Well from what I've seen there are a few things the asset either does or provides as options.

For example my entire area map is a 64x64 grid of course vertically there is no specification you just build up as desired.

You specify if a tile is to be static or dynamic.
I think I saw an option to generate colliders individually not sure but this can be done anyway whenever you like because all this is doing is just using prefabs as tiles. So you could just add a simple box (or other type of) collider directly to the prefabs you plan on using.

However... from what I see it also can auto generate mesh colliders for each batch.

Oh yeah, forgot to mention it has settings for optimization level: low, medium and high.
This setting basically combines the geometry into batches.
So my map in full contains something like 4,200 "tiles" and I ended up with 4 to 5 static batches and 7 draw calls.

Now, of course I'll make some things dynamic because it's always cool to be able to destroy stuff. And that is a nice benefit of a tiled approach. It's easy to destroy and otherwise change the environment.

Anyway, I finally got a chance to get back on the laptop. Gonna add in the jumping now that will give me an opportunity to sort out the environmental collision and interaction. Not sure what Unity will be doing for me. Probably nothing since I removed the rigidbody.

I gotta say though... I knew 3D game dev had to be way simpler than I remember it from 15 years ago otherwise we wouldn't see all of these complete beginners making 3D games. It's kind of wild to think you can get all of this stuff in a scene and yet I don't have a single struct defined in code. No matrix multiplication and so forth. Of course, I am sure Unity is doing a ton of it. But yeah things have definitely improved considerably.

 

I just checked out the videos. Wow! Now see this is a very streamlined workflow for rapidly building levels. It says it is for top down or isometric worlds instead of first-person worlds though. Still would be cool to get at some point.

 

Ha ha! Well unfortunately, it has already went back up to $60. I just went to the page thinking "why not" after all it is important to find the best tool for the job. Of course, then I ended up browsing around and finding some more 3D tile editors. I was going to stay away from the Asset Store. I can't blame you though you only encouraged me and I went there of my own free will.

 

Of the new assets out lately for building art, I am most excited by Gaia

https://www.assetstore.unity3d.com/en/#!/content/42618

although whether to use Gaia or one of these tile builders depends a lot on the genre your game is. I bought it when it was on sale.

 

That kind of thing doesn't really do anything for me. I think it is because I've seen games for the past decade or more pushing for more and more realistic graphics. So when I see something that looks like basically what I can see when I walk out of the house it just seems boring to me.

I mean it looks good in the same way if I take a photo or video of a real life woods can look good. But in the end it's not something I spend my time doing. I mean surfing around looking for photos and videos of woods and plains and such. Now I do like to do some nature photography sometimes though. That is enjoyable but probably more because I am out of the house with camera around my neck.

In other news I just found out I have a Christmas Eve dinner to go to in like 10 minutes. Ha ha! Crazy thing about this time of year is even though you're off from work there is just a lot of stuff going on. Hopefully, I can get back to my project later.

 

Would it have to be restricted to realism though? I only watched half the video for the asset but it left me with the impression that it would be good for quickly painting and populating any terrain.

 

It has the realistic look because that's the assets it uses. The more painterly look is actually really hard to find as a publisher because you need a license that allows redistribution, and there is just very little of that look out there of any quality with an open source license. I'm building a demo for an asset I'm about to publish and I'd love to use something other then realistic, but I can't unless I want to go out and pay someone to create it.

 

There is that Painterly asset in the asset store, Von Gogh, Monet, or one can do or go for some of these syfy artistic renderings.

 

Ah well that is different then. And here I thought people were always striving to make breathtakingly realistic graphics like it is some kind of holy grail. Never even considered that maybe part of why we see so much of the stuff is because it may just be more readily available and easier to do in some ways. Well as far as the textures could basically be from photographs.

 

I was thinking more along the lines of cartoonish (ie Cartoon Forest Constructor) or stylized (ie Stylized Nature Pack) assets rather than simply applying a post process effect.

 

I don't know, it is "heightmap-based terrain" all over again. Would've been more interesting to look into world machine or voxel-based solution.

Steam has game named "Brothers: Tale of Two Sons". You could take a look at it (it is short but it is very good).
The game has "semi-realistic" art. Meaning that environments are somewhat detailed, but characters have somewhat cartoonish look. That works fairly well.

Then we have Guilty Gear Xrd Sign.
It is made in 3d but looks like an animated/sprite based game.

Both of those aren't unity titles, though.

 

Well, I've been wandering off exploring since I was very little so Gaia and those online repositories of maps are fascinating to me. Gaia goes beyond what you are thinking though in helping you make a terrain level.

As far as game style, I like the more realistic environment and more cartoonish character look too and that's the style I've done and plan. The way I see it you want to emphasize one or the other, but not both. You emphasize by making them appear cartoonish or out of the ordinary and the other realistic. Since it's usually characters you emphasize then I see them as natural for a cartoon look but I can see a game with a single protagonist character up against the wilds of nature where you give the protagonist a realistic look and the nature a cartoonish look but not both at the same time. You know what I mean, like a photographer having the discipline to focus on the human subject of the portrait they are taking as opposed to letting the camera focus the human subject and the Matterhorn in the background at the same time, even though it seems to violate common sense ignoring the one & only Matterhorn when you are trying to take an interesting picture.

 

I've seen those and own many like those and have them on my wish list but I've found that the style matches my cartoon style characters I've licensed too much and so they sort of 'blend together' and I want the cartoon characters to be emphasized. For those assets I'd use realistic characters. Sort of like I talke in my previous post. Sort of an Alice in Wonderland Effect.

 

That Cartoon Forest Constructor looks awesome and seems like a great deal at only $14. Unfortunately (or perhaps fortunately since I've probably spent enough money on the Asset store this week) it says it requires Unity 5.2.1 or higher or I'd grab it.

In other news (that you all likely couldn't care less about) my belly is full mainly of Christmas ham. In about 30 mins I'll get back to work on my little project.

 

Nice points here.

I prefer different environments though (urban, man-made, steel+ stone, high verticality, etc, bonus points if abandoned or dilapidated), and a very specific art style, so things like Gaia have little appeal to me... but I see why you like that asset.

 

Code (csharp):

1. OnCollisionEnter (Collision other) {
2.   EnemyHealth eh = other.gameObject.GetComponent<EnemyHealth>();
3.   Destroy(this.gameObject);
4.   if (eh == null) return;
5.   eh.ChangeHP(-50);
6. }

Will hurt any enemies that have EnemyHealth component, will do nothing for others.

 

Nice and peaceful here. Great time to get back at this thing. So far so good! Haven't felt the need to write any custom code to handle this basic stuff yet.

I am sure many of you already know about it. Just thought I'd share for the people who may not know and wonder why they need to shoot out rays, boxes, spheres and so forth to check for collisions (which seems like overkill in many cases yet seems to be the "go to method" in the majority of videos and articles people post around the web)... check out OverlapSphere in 4.6+ or the other Overlap methods if you're using Unity 5.

It's just a straightforward check for colliders within a specific area. Casting just seems odd for many cases. For example, my use case is simply checking the area directly ahead of the player for moving forward and directly behind the player when moving backward. Sure I could do a cast but it just seems silly to do so when all I want is to check this specific area right here. And OverlapSphere does exactly that.

 

I don''t have any OnCollision code. My first couple of Unity projects I used those then switched over to casting for everything and then learned about the Overlap method(s) in 4.6 and 5 so I switched to using those for the majority of cases.

The collision event methods I just never cared for because they report the collision after the fact and I generally want to know about it before it occurs. So moving forward hey there is a wall ahead okay player cannot move forward any more.

 

Lock rotation over x and z axes of rigid body (it's under Freeze Rotation).

 

That's what I thought too but I already had those checked and it still fell over. But that was something I already sorted out. I just removed the rigidbody and it fixed that behavior. I plan on having no rigidbodies on anything. Which I think I read is "bad" when you are moving objects but I never had any problem with it in all of the little Unity games I made. It just makes things so much simpler to not have to fight with that stuff for me anyway.

 

One of standard unity packages has ThirdPersonController in it. Tear that one apart and see how it works. It uses physically driven capsule with attached rigidbody that doesn't fall over.

You might get some ideas on using 3d too, if you read the code of that class.

 

Yes I just learnt about the overlap sphere today, I was watching the tank tutorial.

https://unity3d.com/learn/tutorials/projects/tanks-tutorial/shell-creation?playlist=20081

 

Do you mean ideas for 3D graphics effects or something else? I basically see it like 3D is the same as 2D just there is that added dimension of going into the gameworld instead of just horizontally and vertically as in 2D. So programming-wise is the same thing just have more flexibility as far as level design and so forth.

I'm making good progress on the 3D game so far. However, I have not had much time to work on it due to the holiday festivities. Just got home from my third Christmas dinner. Tomorrow I have one final Christmas get together to go to and then I am done. Fortunately, I took some time off from work next week as well and there is nothing going on then so I should be able to really put some good time into completing my first little 3D game.

So far it has been dead easy. Again that's probably because for me I am doing things basically the same as in 2D just that now I have that 3rd dimension to play around in too.

 

I dealt with the same thing today and just set the rotation axis that I wanted to lock back to 0 in LateUpdate. It seems to work but might be a bad way of doing this. Better follow @neginfinity's advice and look at that class. I should do that myself soon.

 

I know it seems to be just me but this is the kind of stuff I just find odd. For me it just makes more sense to not even use these things if you have to piddle around trying multiple ways just to make these simplest of things work as expected. I'd not have even thought of going into the LateUpdate and locking them in code. It just seems to me like you end up applying "patch after patch" to get this stuff to work as you want. Or as I want anyway.

I've got my player wandering around the stage, movement is blocked when they reach an obstacle as well as identifying exactly what the object is ("You found a medium size rock!", "You found a tree!", etc) and am now about to implement jumping. Been looking forward to getting back at it all day!

But again to me this is no different in 3D as it is in 2D. I'll just check for a jump request and if requested and can jump (standing on ground and nothing directly over their heads) change the state to jumping and off they'll go. Then when gravity (the constant force I define) has exhausted their vertical lift they'll change to falling state and so on.

 

I mean ideas for dealing with animation controllers and physics subroutines.
The class should give you a nice rough picture of what you can do and how in 3d space.

One extra dimension can make a big difference. Compare 3d rotation with 2d rotation, for example.
------

That's why I recommend to take a look at ThirdPerson controller. You ALREADY have gravity in the scene. By defining your own physics force you're reinventing the wheel.

Also I sincerely hope that you're using deltaTime in your updates and not doing anything movement related in FixedUpdate. That functions in the same fashion in 2D, but just in case.

And it makes sense to familiarize yourself with this graph (if you haven't seen it already):
http://docs.unity3d.com/Manual/ExecutionOrder.html

 

Oh yeah. I have been using Unity for about 2 years now. I just never made anything 3D with it because I mainly prefer retro style games. So I focused on 2D. But there are retro 3D style games that I like as well so figured I'd spend some time working on 3D projects this coming year. Just getting a jump start on it.

Yeah I know there is gravity and the normal physics stuff. I was going to use that stuff and decided why bother. If I need to waste loads of time just getting things to work as I want them to I might as well do what I did with my 2D projects and just handle that stuff myself.

At the end of the day I just want to get things done and make games not spend a ton of time on all of this stuff that doesn't really provide any value to me. Again to me. I know a lot of people like the stuff and find it useful.

But this stuff is so simple. I mean just define a gravity constant. Assign some velocity for movement, jumping, etc and then just adjust the velocities each frame according to the forces in play.

Of course, I am using Time.DeltaTime. The same as my little 2D Christmas game in Monkey X is using millisecs passed per frame for frame-rate independent timing. That's just standard stuff that has been around for decades. lol

 

...and then you end up reinventing the wheel and doing more work than you needed, because you just reimplemented a feature someone already wrote for you.

Seriously, just take a look at how ThirdPersonController works.

 

I use two usefull packages for non terrain 3D games.

Rotorz
https://www.assetstore.unity3d.com/en/#!/content/3344
This one allows me to make dungeons almost automatically because you setup your 3D tiles rules like walls , corners etc ... Rotorz place them automatically when you pain the level. You can have multiple grids on a level for different heights, or you can use different tiles libraries on a level and also create tile libraries for decoration objects that will align to floor or walls.

Probuilder
https://www.assetstore.unity3d.com/en/#!/content/15447
I use this one to make quickly the basic layout of game levels, or to make simple game assets sometimes, than i just place detailled models on top of the level coming from Blender or Maya.
With this tool i prototype gameplay very quickly as i don't leave Unity editor.

That's it, as i say often just pick up the tool that you like using the most and that suits your needs.

 

I'll definitely check it out as soon as I get this first little game done. I do want to see what everyone else is doing. Well I mean see what the people who are using those things are doing. There seem to be many people who are not using the physics systems or these character controllers based on random snippets of code I have seen posted around the web from time to time. So it's not like I am the only person just that we are probably a small percentage.

Anyway, I will definitely check it out. The main thing is right now I just want to complete this little game and am making great progress as it is. It seems very silly to me to stop making progress and spend time studying code for how to do things another way when I am already making great progress as it is. I mean if I had no clue what I was doing, no experience in game dev at all then yes I would certainly be studying the code to figure it out. That is not the case though.

In other news, I just got back from my 4th and final Christmas celebration. We did it this evening instead of tomorrow so that tomorrow I will have a full day off to just do whatever!