I would argue that the real world isn't 2D raycast over a heightmap

I remember seeing a 486 DX4-100 doing a reasonable job at 320*200. However the problem faced by current StrongARM machines is that a large heightmap (and colour map for the colour of the terrain) will thrash the cache. Kinetic, Omega etc would really help here.

I've actually got a landscape running at about 1fps on my A4000 somewhere...I nicked the code from the PC version I saw

One idea I've had recently would be to split the height map up into a large grid, the contents of which would be the highest point in that area. That way the operation could be speeded up - if a ray is above the max height, jump it forward until it leaves that area or hits the top of it.

Um...yeah, isn't that what I just said?

The height map is a 2D grid (usually created by making a bitmap of a contour map in a landscaping package or generating it prceodurally). You cast a ray for each column on the screen from the viewers position until it either hits the landscape, at which point you draw a column, or until it moves beyond the view distance.

The only real problem with voxels is fitting polygon objects into them, although you could do a rough ray-trace, and map each point onto a series of polygons (giving the image depth), so then that data could be passed to a normal 3D engine. Of course detail would be lost, but that's the simplest thing I can think of at the moment.

It's probably easier to render the landscape into a z buffer and then render the polygon objects over it. Not efficient but dead easy.

And as for only being 2D environments, what about a double-layer voxel, e.g. for underground caves?

Take it further and create a full 3D voxel sprite....then you can have landscape deformation and everything.

As for speed, you can always store in words (e.g. load a whole word of colour from the approx location, and store onto the screen), although this is likely to make it look worse than full texture mapping at a lower frame rate.

And...if I understand you right, divide your horizontal resolution by four - that'll look awful.
On a similar idea you could raycast four screen columns at once and then as you draw the landscape you could combine the pixel colours into words. Of course this assumes 256 colours. It's irrelevant for 24bit colour.

One idea I've had recently would be to split the height map up into a large grid, the contents of which would be the highest point in that area. That way the operation could be speeded up - if a ray is above the max height, jump it forward until it leaves that area or hits the top of it.

Um...yeah, isn't that what I just said?

The height map is a 2D grid (usually created by making a bitmap of a contour map in a landscaping package or generating it prceodurally). You cast a ray for each column on the screen from the viewers position until it either hits the landscape, at which point you draw a column, or until it moves beyond the view distance.

I think what he was saying was that a lower-resolution grid should be used, with each element in the low-res grid containing the highest height value in the grid cell.
If the ray is above this hight, jump to the next cell in line. If not, start checking the ray against the actual heightmap, and plot where it intersects.
Is this right?

The only real problem with voxels is fitting polygon objects into them, although you could do a rough ray-trace, and map each point onto a series of polygons (giving the image depth), so then that data could be passed to a normal 3D engine. Of course detail would be lost, but that's the simplest thing I can think of at the moment.

It's probably easier to render the landscape into a z buffer and then render the polygon objects over it. Not efficient but dead easy.

IIRC the 3D engine published in Acorn User a while ago did this (I can't remember who wrote that one )

And as for only being 2D environments, what about a double-layer voxel, e.g. for underground caves?

Take it further and create a full 3D voxel sprite....then you can have landscape deformation and everything.

Hmm.. how much space whould this require - treating the play area as a simple 3D array is going to consume a lot of memory, I think
Of course, there is almost certainly a whay of cutting down the memory usage - subdividing the 3D space into large sections, and only storing the voxel info for sections that contain something, for example.

Anyway, back to Andrew's game, which I believe is a top down/isometric view - where we probably don't need a voxel engine (I still wonder why they bothered with a voxel engine in C&C2 - its not noticable in play IMHO )

I think what he was saying was that a lower-resolution grid should be used, with each element in the low-res grid containing the highest height value in the grid cell.
If the ray is above this hight, jump to the next cell in line. If not, start checking the ray against the actual heightmap, and plot where it intersects.
Is this right?

Ah right ok. Yeah that could work.

It's probably easier to render the landscape into a z buffer and then render the polygon objects over it. Not efficient but dead easy.

IIRC the 3D engine published in Acorn User a while ago did this (I can't remember who wrote that one )

If you're thinking of Paul Thomsons engine then no it didn't. The landscape was polygon based, although it did use a coarse grained height map in the form of a 2D array of spot heights. The advantage of this approach is that the polygons are just included straight into the list of polygons to be rendered - you don't have to use a z buffer if you don't want to. Also you can apply 3D transformation to the polygons like any other polygon so you don't have to bodge things like rotation around the Z axis but rotating the screen.

Hmm.. how much space whould this require - treating the play area as a simple 3D array is going to consume a lot of memory, I think
Of course, there is almost certainly a whay of cutting down the memory usage - subdividing the 3D space into large sections, and only storing the voxel info for sections that contain something, for example.

It certainly will require a lot of memory. However John Carmack has been quoted as saying this is one strong possibility for the future as we reach the limits of triangle plotting and you have to take note of what that guy says...

It's another area entirely. Unfortunately it's one I really don't have much time to look into alongside other projects and the 'day-job'!!
VOTI intend that any engine they develop in the future will be available to members (maybe others?) for their projects so this should assist greatly in the long-term.

May I just clarify, if VOTI get to use any engine, if we make our own or we consult with someone else, we will try to make the engine available to other people to increase games development in the long-term.

We currently have no in-house 3D engine and there are currently none being developed that are in a decent state that we know of. Should we develop one in-house we would either sell people the right to use the engine (very small fee, this is RO you know or royalty based) or if we like you we'll come to some other arrangement.

But don't count your chickens.

p.s. the game models are taking shape nicely now

[Edited by 22 at 19:27, 14/12/2000]

For these "true 3D" games the only real difference (apart from the se of sprites) is that each "block" in a level is made from 1 polygon, made up from 6 planes whilst doom uses 1 plane per block. The only advantage i can see for using 6 planes would be from a mappers point of view where it is much easier to carve shapes, & prevent light leaks in the level. Apart from that, the use of 6 planes just wastes texture memory & slows down rendering.

No Doom and Wolf3D are not 3D. The renderer presents a 3D projection of an essentially 2D environment.

The concept of "true 3D" has staggering implications in research and visualisation. A lot of the tricks used to improve performance do so because they limit the environment in some way thus reducing realism.

The last I heard was that to qualify as "true 3D" you had to offer 6 Degrees Of Freedom. In Doom you can't even have bits of a map that go under other bits - how 3D is that?

However I do agree that the distinction is pointless if the end product is any good. Afterall surely it's the quality of the game playing experience that counts and not the actual technology used to deliver it.

Argue about realism, yes, but dimensionality? The concept is clearly defined, and hence Doom is 3D. Sorry to go all technical, but it does annoy me when people argue that because Doom didn't have much movement in the third dimension it wasn't 3D. This isn't so. A point is 0 dimensional, 2 points become 1 dimensional, a third point added not in line with these two points adds a second dimension, and the addition of a fourth point which is not coplanar with these three points adds the third dimension. End.

The fact that you cannot have a room on top of another room has nothing to do with whether a game is 3D or not. In Star Fighter 3000, a limitation of the renderer is that it fails inside objects or underground. This doesn't stop the game being 3D at all!!!

Also doom doesnt allow any form of Mouse Look

Back to the original question, I think (if possible) a none linear story line i.e the choice of where to go, things you do affect what happenes to you later on. Obviously this might be a bit kmuch ( I don't know, I don't do the games programming 'thing') but it might be nice.
Also, how about space combat ala sunburst? you know when in exodus you are about to warp out of a system and it goes 'Our sensors have detected an alien vessel yada yada yada' well, same sort of thing, except with a top down combat view such as that of sunburst. Which is cool.

Glad you liked it :-)

Recent games like Warlords Battlecry for the PC, an RTS, show how much the addition of character development can improve a game. Then of course there's Diablo and Diablo II which turn a mindlessly dull hack'n'slash into a brilliantly addictive game.

From the description, as far as I can tell your game seems to be an adventure game. Is there any way you could work in a system so that improvments you make to your character (by picking up items and choosing skills to get better at) impact how well you fare at the game?

For instance at the moment you might have a sequence in the game where you must find a key to get to the next area (one of the 'puzzles'). But what about if you've chosen to develop your lock picking skills?

Of course there needs to be a mechanism for which by doing things well you're rewarded with the possiblity of upgrading skills. Killing lots of monsters is the obvious choice, but unless I've got the wrong idea, yours isn't a game with a (significant) monster killing aspect. The other way is quest completion, but is your story structure amenable to quests? Ie. is there only one main story that must be followed (quests not possible), or can side stories be taken in order to improve skills that will help in the main story?

Anyway, if its not to late to implement role-playing elements to the game (that's if they're not allready present), and you're interested, I can try and describe more what a good implemenation of character development might be.

Andrew

Forever Story has a more traditional RPG element to it (so character development is based on levelling up, improving or acquiring additional skills/spells etc.). OTOH, FS will have fewer puzzle style elements than Overcast, and is more monster killing and narrative based. So far, I've only come up with 8 or so character stats, but that should be enuff to be getting on with

Side stories will form a fairly major part of the FS gameplay, but will not all provide "material" rewards.

I would also be interested in ideas on character development.

Anyhow, sorry for posting on your thread Andrew!

First off, shane: Yay! May your game be good and strong. Clearly Shanes game is more what I'm looking for, but that doesn't mean an adventure game still wouldn't benefit greatly from role-playing aspects, so my comments should hopefully still be relevent to this, Andrews thread.

On background story: that's all very well, and undoubtedly vital to a good adventure story, but just to make the point, writing background information on things, even on who the character player is supposed to be, has nothing to do with the character development rpg element I'm wishing for.

There are two ways of applying background information in an effective character based game. Method 1: Provide no information what so ever. Alternatively Method 2: give a whole, detailed life history (perhaps revealed during the game).
Method one is probably the most difficult to do effectively in terms of programming, while method two requires inspired fiction writing ability.

Providing no information allows the player to be who he wants to be. The scary thing for the programmer will be the knowledge that the player in this situation demands freedom of expression. If he's not told who he is the player needs to be able to carve out his own particular unique character and will become deeply frustrated if the interface and options available prevent him. Or more likely, the player will just become bored or dissatisfied and you may as well not have included the character development feature at all.
Of course this doesn't mean you have to include every possible option of character improvement possible - that's _im_possible. You've just got to be clever with what you do, carefully molding the players expectations of what might be possible.
For example, don't show or describe a character (an npc or enemy) that is similar to the player character in some ways (is human for instance) but unattainably different in others (the enemy is a dual sword-wielding bad-ass, while the player can only ever wield one sword, no matter how well he plays the game).
If you tell the player he can be a fighter, the player might want to be a super strong fighter or a super quick fighter. Or one that uses mainly risky special techniques or etc. etc. It's easy not to provide for all these possibilites. It's also easy to make a bad game. Instead mold the players expectations: your essential game story describes your world as 'set in the dark ages with primitive weapons'. The player now won't aspire to be a japanese kensai sword saint and won't notice that he can't choose to specialise in speed but only strength.

Now with method two you can take these ideas and define exactly what the character will be and therefore exactly what skills will be learned during the game. There must still be choice of course, but you can get away with far far fewer options. The only problem is that the player must want to be the person you're telling him he is. Given the choice the player might have chosen to play a warrior, but your detailed story and character description has him playing a wizard. As long as the wizard is a deeply interesting and special person, the player won't mind playing him. He needs hints from the outset as to how special this wizard might be.

Uggh, this is too long and I haven't even started on my ideas of a good implemenation yet

I'll save those for another day if anyone's still interested.

I'n using a mixture of the two in Forever Story. The game starts of by defining pretty much what kind of character you are (well, characters really, there are several playable characters in the game). You can then customise these characters to varying degrees throughout the game - you can decide what kind or even if each character learns magic for example, and you can increase certain abilities permanently by use of items you find throughout the game (eg, a Strength potion/pill/amulet to increase physical combat done by a character), and you can give characters certain abilities through the use of items, such as the ability to use two weapons or a weapon two handed etc.

However, you will have to decide on your own character development. You *can* make a character who starts off as a poor magic user into a good one, by using several items on him. But those items would be better used on a good magic user to make him an excellent one. Certain abilities are character specific and cannot learned by other characters, but others are generic.

Hoepfully, all the characters will be deeply interesting and special people

Anyway, Doom provides a 3d representation of a world which is not specified completely in 3 dimensions. By this logic, Doom would fall around 2.5d which is far closer to the truth in my opinion.

After all, if you have a 2d game game with parallax scrolling, is that 3d? By your logic, it has vertical and horizontal scrolling, and the parallax offers a depth effect (in the way Doom offers a height effect) your 'world' appears in 3 dimensions. Yay! Predator on the BBC was 3d...

[Edited by not_ginger_matt at 02:07, 20/1/2001]

It can be fine if by 'define what kind of character you are' means simply to say something along the lines of 'you are member of society p in the middle of the wobble wars'. This is meaningless characterisation (as opposed to description of character), so effectively counts as my method 1 of saying nothing at all.

Saying, however, that the player character is 'good', but then allowing the player to choose to be evil is clearly broken. It completely devalues the background writing and leaves the player unsure if he should be trying to play his own or the game writers character. By defining the player character, you provide focus and an interesting goal for the player.

And all the characters _shouldn't_ be deeply interesting and special people, at least not in the way I meant it. I meant that the player character should be the most interesting person in the game. So if the player likes being a fighter and sees really cool enemy fighters in the game, the only way he won't be unhappy about not being able to be a fighter (beause your story has him as a wizard) is if your description of the player wizard hints that his magics can become truly spectacular and will be able to defeat any fighter in the game. With enough detailed description of this wizard, the player can be focused into investing thought into truly playing the role of this wizard (working out what the wizard might do... role playing!), and also have the exciting goal of developing those magics to defeat the fighters the player otherwise might want to be.

Again, I haven't time to get into my ideas on implementation proper, but just to address one other thing you mentioned:

The idea of using items to increase skills that you previously had to work hard killing monsters or doing quests is a dangerous one. If you just find some item lying on the ground that, when used, does the work of two hours of careful fighting, you'll completely devalue the players hard work. Suddenly he doesn't even need skill to do what he's been doing. Simply having these items at the end of hard quests doesn't solve the problem per se.
The way to do it is to set up these items as very special goals - the very point of doing certain quests as opposed to the suprise reward of doing them. So you say that 'in the cave of bubbles there is a mystical bottle of pills'. You can still keep the surprise of what the pills do, but now the player sets out to the cave of bubbles and all the hard work he does getting there counts directly (as far as the player is concerned) to getting that extra point of strength from the pill. This is very different to going to save the princess in the cave of bubbles where the player will think all his hard work is going to level himself up in the normal way and get an extra strength point in the nomal way. If the princess then gives him a strength pill the player feels cheated.

Also, I don't think it's appropriate that an item should give you the ability to use two weapons. It doesn't make any sense. That should be the reward of being a good fighter and a difficult choice the player must make - I've just spent the last 10hours fighting monsters, now I have enough points to choose between fighting with two weapons or better using a two handed weapon. The player enters role-playing mode and thinks about the sort of character he's player - a big brute that would use a large sword, or a skilled technition that would use two light swords. He makes his choice, wonderful. But if you can just happen across an item in the course of the game that says 'use two weapons' it's more like 'errr, ok, whatever', even if he stops himself using two handed weapons in the future.

To sort of summarise the whole item issue: Reward the player for doing what he likes doing by letting him do it even better. So if he likes to swing a sword, once he's swung the sword enough times, reward him by letting him swing two swords. Don't tie the reward to something unrelated to what the player likes doing - don't tie it to 'picking up items'. If there's a picking up items skill you could reward a player that spends most of the time picking up items by doubling his chances of finding items on the ground, for instance.
Anything else, IMHO, devalues the players experience.

[Edited by sendu at 10:24, 20/1/2001]