• Advertisement
  • entries
    35
  • comments
    145
  • views
    30830

Level of detail vs. Level of Distraction

Sign in to follow this  

2021 views

Whelp.... My last attempt at spurring some conversation was a bit of a bust. Over 1000 views in 24 hours, which is amazing to me, but despite strongly encouraging people to say hello, not a single comment. I'm not angry or sad (okay, a teeny bit sad), but it does seem to indicate that people both here and elsewhere (I have tried the same in other fora) are passive consumers, rather than active participants. To me, that is a bit disheartening, because I love to hear other people's thoughts. Seriously, I would not be able to handle telepathy in a mature, respectful way, no sirree. But alas, I will poke people no more to speak their minds. Well, for now, at least. Cab driver, on to other topics!!

So, you may have noticed the small, discrete image at the entry to this post. That's a test image from my game. The top image is from the surface of a celestial body (technically, a star, but the same code should be able to generate planets, moons, whathaveyou), the bottom one is the same celestial body at a distance. The big deal here is the weird pattern of ever-shrinking polygons on the thing. They are, of course, the product of my new favorite word: Tesselation. Just say the word, feel it, it is goddamn beautiful. For the uninitiated, it simply means that when you close in on a polygon, it becomes more polygons, adding dynamic details as you get closer (and removing them again when you go farther away). Combined with my logarithmic layers of scale, this should allow me to create a graphics engine that can handle anything from subatomic particles in a fly's wing to galactic superclusters in one and the same scene. As stated in a previous blog entry of mine, one use is to create Shadow of the Colossus style kaiju monsters in the game. Or to go one step further and create intergalactic monsters like in Lovecraftian horror or comicbook ultra-supervillains, or even planet-eating mega-robots voiced by Orson Welles, without adding undue strain to the computer (it's running on an 8GB laptop at the moment). The current work-name for the project is the SDNM engine: Size Does Not Matter.

One thing that dawned on me in this was the question of "when does detail actually matter?". I like detail. If I could, I would make everything so detailed you could pick the little hairs off the edge of a blade of grass (go look closely at a big blade of grass, closely. Yes, you'll look like an idiot, but that stuff is complex!!). But that doesn't mean the details actually matter. Minecraft got along fine just on blocks, and retro or fake-retro graphics styles are popping up as a counterweight to the graphics overload of quote-unquote "tripple A" games. Detailed graphics rarely affect story, either. So why have them?

At this point, the argument of "well, it doesn't harm anyone" could be made. True, a game can fail because of overdependence on snazzy graphics and lacking priority on story and gameplay (No Man's Sky, many brown-n-grey shooters, etc.), but if the game is solid, the graphics are no harm, right? Most of the time, true. But there comes a point when details become distractions. Imagine if Super Mario had a busy city in the background, plummers and goombas running around to shop and socialize. The idea might be appealing, it would add a whole new layer of visual mood to an otherwise fairly simplistic game style. But the big advantage of Super Mario is the simplistic game style. You know where to go, when to jump, and anything unusual either sticks out clearly or is so truly hidden that you do not blame yourself for missing it. No distractions, just the game.

Push that to more modern games. Many games have whole towns to walk around, or even city districts (think Deus Ex), but most is just for show. we know, as players, not to expect to interact with much of it, at least not in a meaningful, productive way. But that means that game developers have to counter this level of detail with in-game symbolism to show what can be used, and what is just for show. If we increase that level of detail, at what point do the cardboard setpieces outshine the actual, useful game content? When will half a game involve running through a city of pretty-but-useless background fluff, just to find the one shop or contact that actually has game value?

I'm not at that level yet, not even anywhere near it. But I'm already having similar concerns. My SDNM engine structure lets me create millimeter detail on anything the player can see close enough (technically, it can do it for distant objects, as well, but that is of course pointless). But how much of that would be nothing but distraction? Like the procedural universe of No Man's Sky, how long would it take for that level of detail to become the gimmick that drowns out any work towards creating gameplay? I am sure most of us face a similar problem, in that we let the giddiness of coding something unusual cloud our minds and make us forget to actually make a game. But even if we make the game, with great gameplay, could the level of detail distract players so much that it drowns out the game itself? Where does the line go between level of detail and level of distraction?

I have no answer. Honestly, it bothers me. If anyone has a stance, an argument, some random thought on the matter, please, do let me know (yes, I am trying to make people comment, again, but this is not the same pressure as before). If nothing crashes horrifically for me over the next few weeks, I will most likely be facing the question of "how much detail?" very soon, and having some things to keep in mind might not be all that bad....

Cheers, and don't be a stranger ;)

Sign in to follow this  


27 Comments




Recommended Comments

Interesting thoughts, I had somewhat similar ideas for a game. It was a fantasy setting, open world etc but with large scale battles in first person as part of the gameplay. I looked in to how you could have 10-20k characters fighting at the same time, including large beasts and war structures, and I realised that most of those characters would fit in to a few categories- 1) out of sight 2) far away 3) blocked by others 4) in need of detailed rendering Using these ideas I was able to have a few thousand characters on screen without really pushing the detail as low as I could have. I even changed combat mechanics based on distance too, to just basic dice rolls when out of sight. Basically i just wanted to take LOD to another level.

Share this comment


Link to comment

It's actually nice that you talk about that right now. I'm currently working on a game with a small team of devs, and being in charge of the level design, it's something I'm slightly worried about. The game we're making would be an open-world isometric stealth game, between Thief and Divinity: OS, and my big concern right now is making a believable city, a sizeable one, with different types of architectural inspirations.

Creatively, it's absolutely blissful. Being kept away from the programmng concerns make it so that I can really focus all my energy on the city, and oh boy do I love the european setting we got there. I can draw inspiration from almost everywhere in Europe, make claustrophobic vertical mazes, Santorini-like staired districts, rich, huge avenues surrounded by mansions, banks and whatnot. It's fantastic.

But then, I wonder if what I *want* is what the game *needs*. You should see the place I've made last month, codename "The staired world of Sapienzetta" (and before you ask, yes, those are Dark Souls and Hitman references, sue me). The place is huge, inspired by cities like Santorini, packed with crossing paths, verticality and hidden areas.

The problem is then asking myself, what good is it to create environments that'll have to be fully textured, detailed and decorated, but with a hundred different ways to navigate through them? How do I *not* screw up the player's navigation skills, how do I make my city compelling, intriguing and worthy of the exploration of my player without making it unclear what's pretty makeup and what's clever design? You talk about the use of symbolism in the navigational system, and they're indeed needed, but a problem I've always had with these is about their (devilish) efficiency.

Look at Mirror's Edge. When it came out, the universe's style was very pretty, and detailed, with complex environments that almost everyone completely overlooked because of the red environmental markers telling you what's interactive and what's not. At some point, you cease to look for anything else than these red elements, since it helps navigation so much (you can still turn it off, but a very small minority of players will actually do that). What this does is basically throw out of the window half of the environment design to accomodate the player's experience. It's great on a level-design standpoint. It's really, really shitty when you want to create a believable, interesting environment that'll draw the eye of your player, since you'll have conditioned him/her to overlook everything that *isn't* a symbolical navigational help. Splinter Cell Conviction did that too, the beautiful, delicate coloring of the environments made a leap of faith in the trash compactor whenever you were in the shadows and the black and white kicked in. And, you know, infiltration games and shadows... yeah. You spend a lot of time in them.

The good thing of isometric games is that you can keep the level of details to a minimum while still having a very pretty final product (Divinity being a very good example of that). It's much easier to make the essential of the level-design stand out in a game like that than in, say, a TPS or an FPS - the top-down view helps your brain to constantly map what you're seeing, so if the architecture is minimally realistic, you should have instinctive reactions to what's possible or not in the place you're in. So, even if I'm worried, I'm sure our environment artist will do a great job at keeping the level readable.

In your case, it's all gonna depend on the type of game you're working on, the type of experience you want for your player, the limitations of your platform and what you're able to do with your current skills. Even an insanely pretty game like Paragon lacks finition everywhere in the map - barely any environments in the exterior of the map, rough and unpolished wall/floor connections. This is a MOBA, so you'll want to focus a lot more on the mechanical aspects of the gameplay than on the level of detail of your environment. If you design a first person RPG like the Deus Ex series, then you should definitely make the level of detail solid and impressive, since the world you're building is basically what's keeping your player around. Pick a peculiar visual style like Borderlands or Necropolis, and you'll have fond the answer to your questions inside that same visual style.

So really, find what's your game's all about. After that, take a stand on the matter, and trust your stance through the development of your game.

Share this comment


Link to comment

Interesting thoughts, I had somewhat similar ideas for a game. It was a fantasy setting, open world etc but with large scale battles in first person as part of the gameplay. I looked in to how you could have 10-20k characters fighting at the same time, including large beasts and war structures, and I realised that most of those characters would fit in to a few categories- 1) out of sight 2) far away 3) blocked by others 4) in need of detailed rendering Using these ideas I was able to have a few thousand characters on screen without really pushing the detail as low as I could have. I even changed combat mechanics based on distance too, to just basic dice rolls when out of sight. Basically i just wanted to take LOD to another level.

Oh, god.... I am dealing with fairly simple objects like planets and stars (big balls, essentially). When I have to eal with large-scale character scenes... Holy balls. And yes, it's very much about defining your needs for LoD, and your categories are a good example of it. But the problem is not to divide things into LoD categories. The big deal is making the game do it for you. I can still use distance as an indicator of detail, but the obstructive view, activity levels and more that you mention will have to take it to a whole new level. If you have some good tricks/experience to keep in mind, I would love to know it!! Do you by any chance have a blog that details you exploits in that project?

Share this comment


Link to comment

Silver_Blaustein: (the quoting function failed me, so it's manual this time...)

Whenever I think too long about these things, I start having flashbacks to my (very short and unimpressed) days of playing WoW. I had a thousand ideas for how they should reduce the time spent moving from A to B, because it was a drag having to hold down a button to keep walking with nothing happening. You seem to be at the other end of that, in that it's an experience to just walk through the labyrinthian city, but that experience is not meant to be the game itself. Again, it's a level of distraction, something that appears impressive enough to seem important. And that's when the darkess creeps into you.

You seem to have yours pretty much locked down. I can understand your worries, but to me, it seems like you mainly have to worry about "how much?", like whether you should reduce the distraction content in favor of the detail content. Your thoughts on the symbolic counterweight to distraction interests me, though, and it's something I have long wanted to find some good material on. Mirror's Edge (ME) is a really excellent example, too! The sleek style and simple, intuitive game mechanic of the colors (is it even a mechanic, or are we dealing with practical art at this point?) makes the game pretty and intuitive at the same time, but yes, one seems to inevitably outshine the other; either you enjoy the city and neglect to keep your eyes on the color guides (and you're likely the sort that really wants the open-world version supposed to be in the sequel), or you blindly focus on the colors (that was a weird sentence), to the point of not enjoying the city. ME does have its very, very fast pace play a role, too, though, since you really can't stop and smell the roses in a game built like that.

I think my greatest hurdle is the fact that I want a very realistic, very open experience. The symbolism annoys me. It always took me out of the game, killing the immersion for me. Having something that points out gameplay hotspots just turns me right off. I'm the sort who will gladly hide the HUD of a game, if I can, to get a more "true" experience. Yes, I even play racing games from the driver PoV. I don't play them well, but I like the feel. That means either no symbolic game guides, or strictly ME style guides. But ME style guides conflict with the open world structure. As stated, I honestly have no idea where to place my resources on this, so for the time being, it's non-symbolic open world all the way.

Do you have a development log somewhere that I could browse through? You got me interested in seeing what you do...

Share this comment


Link to comment

Sadly I don't really have anything yet, no, we're still very early in the development and the majority of the team is new to the industry. Because of that, and with the fact that we're all from different countries, coordination is tougher than it should be, and we probably won't have anything like dev logs before a while. I'm thinking about doing my own, although I'd have to find a decent place where to publish it. On GD.net, maybe?

 

Regarding the WoW thing, I actually agree with you. I was a huge fan of the world building of the old WoW, which showed the world scale in a way that very few games ever could. Walking through the Westmarch or the Barrens was long, but it added a lot to the immersion. It was annoying though, long and boring, and often it was like walking miles and miles through a minefield, with you praying that an enemy won't magically spot you from far away and force you to begin your travel all over again.

 

Travelling through an environment should never be a chore (like, random example, the reboot of DmC and other games like Shadow of the Damned or NeverDead). Architectural principles for city building, even for fantastic worlds, are paramount for that. The environment should be considered a living being that has its own reason for existing. This city, where and when did it began? How's the water supply handled, how did it change through time? What culture do the citizens live in, what variations of the region's culture (say, mideast england) are alive in this place and why? And most important of all, how does the city keeps surviving and evolving (or stagnating) among all that?

 

The environment is definitely not enough for a game, and it's basically useless without decent incentives to keep playing, gameplay-wise. But travelling through places that are sizeable in video games doesn't means boring walking time though, navigating through the cities of the Assassin's Creed series was rarely a chore, mainly because of the fun/buggy climbing system, but also because of the high amount of things to do in it.

- Even if I highly disagree with the increasingly popular principle of giving hundreds of collectibles to the player to find, just for the sake of making them play longer, it certainly works. -

I'm personally gonna make sure that environments are akin to those of the Dark Souls series. You must be attentive, since each unsuspected part of it can cause your very tragic death if you're not careful, and since many secrets (read: more information about the world and actually useful loots) hide everywhere.

Share this comment


Link to comment

The reason I have been a passive consumer, is because I have only negative things to say; although I think what you are doing is great.

Currently tessellation has very few uses. Already it's possible to use LODs to make grass with hairs, the problem is the same as in tessellation; you need to get very near the grass for there to be resources to render the grass hairs.

The process for making 3D objects that use tessellation is the same as for LODs.

If you made a flat polygon to be a wall, then modeled high poly bricks, baked the bricks to a height map, then used it with tessellation you can get high poly bricks when near.

However to get the height map you need a high poly mesh, if you have a high poly you can get a low poly from it.

Tessellation isn't precise it will split your model into many triangles even on flat surfaces, so the low poly model made by a artist will have a lower triangle count and look the same as the tessellated one. Meaning that you could load in the brick wall LOD sooner than the tessellated one, allowing you to have detailed walls from a further distance than tessellation.

Tessellation is used in games for only organic, terrain and cloth. Because these surfaces are always uneven and profit the most from tessellation. Some times the detail LOD of a object can be tessellated when the camera is near, however this is done by using a inaccurate normal map input; meaning that it's only done because it can be done and doesn't make the model look any better.

The problem game engines have isn't how many small details can be drawn when near a object, it's how many small details can be drawn when you are still far from the object.

 

I like what you are doing here, although a universe can be made using LODs, and have been, clearly tessellation works better.

You are exploring relatively new territory, it's interesting and you could discover many things. Thanks for the post.

@Silver_Blaustein

For making cities, check what kind of engine your are using. For a batch engine like Unity you need to use zones to build large cities.

For a instance engine like Unreal, you need Hierarchical instancing for cities.

Share this comment


Link to comment
@Silver_Blaustein

For making cities, check what kind of engine your are using. For a batch engine like Unity you need to use zones to build large cities.

For a instance engine like Unreal, you need Hierarchical instancing for cities.

Thanks, we're using Unreal, and I should definitely check that out.

Share this comment


Link to comment

Funny thing about the sense of world scale in WoW, it's actually an *extremely* compressed world, spatially speaking.  It has just enough slog to *feel* big when holding that "w" key, but it's nowhere as large as a real world would be.  Ditto for the art, usually.  There's just enough detail to let players intimate that there's more there than there actually is, but not enough to get in the way.

This discussion actually reminds me a great deal of the Uncanny Valley effect in visual art design.  As in, too little detail and it doesn't look right (especially with animation), too much hyperfocused attention to detail and it looks *very* wrong.  There's a sweet spot at the other end of the hyperreal curve, but it takes a *lot* of work to get there, and the perch is most precarious.  Most of the time, you're best served with just enough detail to look like it's complex (hinting that there's more if you just stop and look), and movement that's natural *enough* to feel right.  (This is why WoW and Wildstar look more "right" than LOTRO, for example, especially after the WETA-infused Peter Jackson LOTR movies.)

The animation is where it comes to a head, though, in my experience.  We instinctively know what humans move like, and the slightest hint of artificiality shatters the illusion, even with perfect textures, shaders and lighting.  And yet, take Mr. Incredible and his family, or even the Big Hero 6 crew, and in so many scenes they *feel* right, mostly because of the *exaggeration* instead of the slavish adherence to what seems to be objectively true of motion and emotion.  This is where animators really hone their craft, learning to exaggerate the parts of a scene that make the motion feel right, even though it often isn't, technically speaking.  It's *right enough* with no glaring issues like sideways gravity, but it's so often more about the gesture than the hard-nosed reality.

I'd suggest that there is a game design equivalent of the Uncanny Valley in the player's experience with the game world.  Some might call it "flow", or perhaps "zen gaming", or some other linguistically meaty allusion, but for me, I tend to just think of it in terms that I'm familiar with, those of animation and art.  Exaggeration, intimation, presentation and allusion go a long way if you can get players on board.

As in, if a player expects detail in a wall that they are supposed to be paying attention to, it is good to have that detail.  If they are expected to just breeze on by, as in the Mirror's Edge example cited above, it is a detriment to put detail where it doesn't buy you anything but distraction.  It's not just a waste of resources, it's a drag on the player experience.  It all has to do with what you as a designer want the player to notice, and what the players are likely to do.  

A video game isn't a movie, it's interactive, but it isn't a purely Pavlovian experience (QTE abuse aside) either.  It's in an interesting space where players want control, but are generally OK with squishing off the walls of a padded tunnel experience as long as the collisions aren't too jarring and the light at the end of the tunnel is interesting enough.

As with the Uncanny Valley, it seems to me that designers need to learn to leverage player expectations of detail/gameplay and just hint that your work is going in that direction, and their imagination will make up the rest.  Don't do shoddy work and hope they do the rest, but meet them partway with competence and confidence, hopefully with some stylization to deflect the rational analysis part of the subconscious, and let their imagination and your sneaky hints do the heavy lifting.

I'd cite Mario Kart on this one, along with the DIRT series.  Those are very different games, but they *feel* right because of how the premise leverages expectations.  It cuts both ways, though, too.  Someone who cut their teeth on one of the more hardcore Need For Speed simulation-heavy games, for instance, will experience some dissonance (perhaps even subconsciously) when playing DIRT, but won't reflexively balk at Mario Kart in the same way.  Similarly, one of the Motorstorm games won't trigger the same instinctive "this isn't real" reactions, because it doesn't *try* to be real.  It tries to be pretty and mechanically consistent for the desired effect, but not true-to-life *real*.  Plausible and entertaining, not practical and precise.

Tangentially, I worked on Tiger Woods PGA Tour games, specifically trying to make water, reflections and shadows work well on the PS2.  Reflections are especially easy to fake a lot of the time, with very, very low detail and only the broadest strokes of the same shape as the object they are supposed to be perfectly replicating.  People notice if they aren't there, or if they are grossly incorrect (especially in motion), but there's a surprisingly low threshold of acceptance for most players.  I wound up thinking that if you could really polish the core experience and let the presentation sort of fade away on the margins, you could get away with a lot of... I'll just call it "gameplay vignetting".  As in, the stuff that's happening on the margins sort of Gaussian Blur out of relevance, and it's OK that they do so.

This isn't just a budgetary saving, both in production and in processing, but it's also helpful for the gamers.  As long as you make that transition smooth enough where people can accept that they should just pay attention to the core of the experience and ignore the periphery, they will have a better experience, too, since they aren't getting lost in the weeds, paralyzed by detail or gameplay overload.

Of course, a Minecraft sort of game will be all about getting lost in the weeds (though even there, detail only goes so fine), so again, it's back to how that Uncanny Valley curve is shaped in your design.  Where do you want players, how fast are they moving from there to somewhere else, and what should they focus on?  Give them enough to stay near your preferred experience, don't make the walls too hard, and trust their imagination.

...I know, lots of blather, but I think there's something to the Uncanny Valley extended metaphor and leveraging player expectations instead of trying to Do It All and make everything available and perfect, which all too often just gets players bogged down in the details, where they complain about things all the more vocally.

Share this comment


Link to comment

Travelling through an environment should never be a chore

Amen to that!! I think that sums up a lot of my worries about the level of detail. At some point, the detail becomes a burden, if it is not thought through right and properly connected to the concept of the game (see my previous Super Mario example as an, uhm, example). WoW especially gave me the feeling that it was trying to more show off a lot of world building detail, rather than provide me with game content. I have not played, and probably will never play, No Man's Sky, but I get the feeling that this is also the big problem for a lot of people playing that; those gazllions of worlds feel more like a distraction from the fact that the game cannot follow up on them. They become, to use your word, a chore.

Now, far be it from me to nag about a game having obstacles. I start my games on the hardest level I can possibly handle, and always aim to move to harder levels ASAP. But when overly detailed worlds simply become something you have to slog through in order to get where you're going, or when (dear god) you have to search a landmass the size of Lower Manhattan for a single stinking flower, it's not a challenge, it's a chore. It's a distraction, a tagged on extension to the game only there to add playtime, without any actual fun.

Part of me suspects that we simply have not developed the notion of content fast enough to match the vastly increased amount of detail that modern hardware and software can handle. To use a quote from a movie that has been burdened by too much distraction, "we were so focused on what [we] could do, we never stopped to think about what [we] should do". As stated, with some tweeks and adjustments, I can make the SDNM engine truly infinite. Endless seas of galaxies with an endless flood of planets (which will hopefully soon become truly varied in content, still working on that), all of it detailed down to the last pebble and hair. But does it make sense to do that? Or will it just become,,at best, unimportant backgrund noise, and at worst, a burdensome distraction for a player trying to find something useful ("is it this rock? No. Is it this one? No. Is it....")?

I wish there were extensive studies on this to inspire me, but we seem, as an industry, to still be totally confused by the fact that it is even becoming possible to do!

Share this comment


Link to comment

The reason I have been a passive consumer, is because I have only negative things to say; although I think what you are doing is great.

Currently tessellation has very few uses. Already it's possible to use LODs to make grass with hairs, the problem is the same as in tessellation; you need to get very near the grass for there to be resources to render the grass hairs.

The process for making 3D objects that use tessellation is the same as for LODs.

(and the rest of the post)

Firstly, I see nothing negative in anything you wrote. Don't know if I misunderstood your leading statement, but I very much appreciate your comment :)

And yes, tesselation has limited uses, at this moment. Personally, I think that it's because some concepts just don't work well entirely on their own. Even cell phones would never have had any use, if it had not been for advances in the completely seperate industry of rechargable batteries, for example. The fastest car is no better than the slowest, if there are no long roads or tracks to drive fast on, and so forth.

I did another post on procedural generation. Being a severe math-addict, I have a great love for PG, and a great frustration at the lackluster way it gets used today (yes, Spelunky is adorable, but the PG is still minimalistic). But PG has another problem, in that it tends towards bland, copy-pastish "random head, random legs, random color" creations. Someone on Youtube pointed out that no, No Man's Sky does not have gazillions of differnt planets, it has about six, and they are mostly just color swaps of one another. This is the Achilles Heel of PG.

My work is based on having these two concepts compensate each other. "Synergy", as the suits like to call it. Tesselation, as we mostly know it now, requires detailed instructions on how to increase the LoD. That puts pressure on artists, who suddenly need to define insane details on even low-poly models, so that they can tesselate properly. This is not an advantage, it's a chore. It's a distraction. And that is why, as you say, it has no (well, few) uses. But if those details are PG math, things look very different. You mention brick walls. Although it might be very deep into the (seemingly) unimportant side of things, what if a brick wall was just that: A brick wall. All the artist has to do is label it "brick wall". Then, the PG engine generates a grid of bricks in whatever color it is allowed to, for a distant texture LoD. Up close, it uses some fairly simple math to turn that texture into actual brick models. It's mostly just a rough rectangular box, after all. Stack them with some mortar space between, and you have a wall. Even closer, the bricks can be given actual modeled roughness via PG, including chips and streaks, miscolorings, flaked mortar, and more, all depending on a few stats (old building? Plenty of chipping and flaking!). My still very theoretical math has parts that do this with architecture. From a distance, a window might be just a colored rectangle. Closer, it might have a curvy top, maybe multi-paneled glass. Closer, its frame has a shape (funny thing: Draw two points on a piece of paper, and a squiggly line between them, which never moves flat-out backwards, and you will almost always end up with a usable profile for the shape of an artistic window frame's edges). Up even closer, and you notice chips, flakes, crooked glass panes, and so on. The artist only needs to place the window and add a few labels for control (maybe enter a preset art style, or a simple integer to denote level of complexity). And once you get that working, it's not that far removed to have the PG engine make the entire building, combining windows (which can keep the artistic style across all windows, maybe with the occassional replaced window standing out randomly) with brick walls, and adding some interesting geometry to the building (see this for some cool ideas, and note his remark at 3:16).

I think what really bothers me is that you're right, this seems like new territory. I lack sources of real inspiration for the deeper stuff, but I guess that's what discussing ideas is for :)

Share this comment


Link to comment

Funny thing about the sense of world scale in WoW, it's actually an *extremely* compressed world, spatially speaking.  It has just enough slog to *feel* big when holding that "w" key, but it's nowhere as large as a real world would be.  Ditto for the art, usually.  There's just enough detail to let players intimate that there's more there than there actually is, but not enough to get in the way.

This discussion actually reminds me a great deal of the Uncanny Valley effect in visual art design.

[......]

On the slogging: Yup. That's pretty much the case in a nutshell. Press "w" and get the sense that there is so much more than there is, and never really do anything with it. The ultimate expression of "distraction".

On the comparison to the Uncanny Valley... I never thought of that. It still has to make its way around my brain, but yeah, there are a lot of similarities. I did some animation work a long time ago, and the work with humans brought me quickly into UV territory. I can see there is a connection to this, but it's something I need to really think about.

I guess (and I'm thinking this as I write, stream-of-consciousness style) that maybe the details can overwhelm part of the brain that is experiencing it. Human brains are designed to reject the grand in favor of the narrow, to focus our mental resources (does my hobby study of neurology and psychology shine through?). Eyes, even, are built to only see detail in the middle, and blurry outlines outside that (read text on you cell phone screen, then slowly move the phone outside the center of your field of view, without letting your eyes follow it. You'll see the text suddenly just blur up completely). Maybe the deeper thought processes connected to that makes the detail/distraction issue pop up, just like the UV effect. The sudden proximity to real images (or levels of detail good enough to seem like real environments) makes the brain focus its full resources on it, and then, the material just starts to fall apart under scrutiny.

For me, the big issue is when a big part of the entire idea of the game is detail, as in my case. I am tired of games doing the "oh, no, you can't interact with this. No, no, you can't cross that invisible wall". It completely breaks immersion. But there seems to be no dial to adjust between that and complete detail. The best I have seen is honestly Minecraft, and again, it might be because it makes a massively detailed world, but with a far less detailed game, to compensate (whether they planned it that way or not). But what happens when you take the full-on gameplay of, say, [sniiiip] Okay, I just realized I have no other game to compare that thought to. I literally cannot think of a game that provides nearly unrestricted use of existing environments. Even games like Witcher, Horizon: Zero Dawn, GTAIV and so on have very set restrictions on what can be used and what cannot, up to and including the aforementioned symbolic game guides. I can't pick up any rock in Witcher or use every device found in GTA, a lot of it is just there for show. And a lot of that becomes distractions. Oh god, it's come full circle! [brain goes pop]

Anyway, yeah. There is a strong feel of Uncanny Valley somewhere in this. I can't put my finger entirely on it, but it gives me pause to think. And thinking is half the battle. Or something. Gotta go pick up brain pieces now....

Also, clearly, I have no problem with "blather". Blather on. I got a lot out if it. Blather on!

Share this comment


Link to comment

@Embassy of Time, you seemed to have forgotten that a 3D model is vector math. 3D software is nothing more than a visual math editor.

The math for making 3D objects is very complected, that is why we made visual editors to get an idea of what we are doing.

If flakes and scratches could have been added with nothing but some simple math then tools like that would already have found it's way into a 3D editor. 3D sculpting is the current best solution.

You could do it all on a height map, then do the tessellation, however you hit the inaccuracy problem. I recommend you do make a wall so that you can see how it works.

You could also use math to make the brick wall, however a 3D model would be faster.

If you kept the models as simple as in Pixel city or as simple as you described your wall, then you wont have as much of a problem, it will still be there just small.

You could build a library of 3D models to use as a base, so say a 3D Brick wall that a modeler makes, uploads to a library and anyone can use it. However like I was saying the process for doing this is the same as making regular game assets, so why wouldn't they just make a regular game asset that would perform better anyway.

The only advantage I can think of is that a modeler would only have to model a few bricks for your engine, that you then use to make the walls. The modular approach isn't new to games, the down side would be the large generation times.

You could just then apply some of that easy math, the the already made models. :)

Share this comment


Link to comment

@Embassy of Time, you seemed to have forgotten that a 3D model is vector math. 3D software is nothing more than a visual math editor.

[....]

The only advantage I can think of is that a modeler would only have to model a few bricks for your engine, that you then use to make the walls. The modular approach isn't new to games, the down side would be the large generation times.

You could just then apply some of that easy math, the the already made models. :)

Sorry. I failed to explain my point well, I think (I blather a lot :) ). I meant it as an example of PG LoD, that could be transfered, in concept, to other work. The artist could, for example, make a spaceship. But the game could use that pristine ship to make aged, broken, faulty or other versions, without the artist needing to put in the extra work. Think of it as having someone to color in your drawings for you, you just tell them the colors. And then you can tell them to make alternates with different color schemes, as in the classic 80s/90s games and TV shows (think Sub Zero/Scorpion/Reptile/etc. from Mortal Kombat. Or Power Rangers). The PG finishes the work according to the artist's requests. That means the stored spaceship model can be fairly simple, as the PG engine simply adds the different details on the fly. Thus, tiny details (as in physical details, little scratches or color gradients) need not burden the artist much, or take up much space on the harddrive. And the PG engine would presumably be able to use the same schemes on other spaceships, too.

I think the problem is that I see PG as something else than most others. I don't see the "mix and match" version so much as an evolutionary modelling assistant to the artist, which can work on the fly during play. I should really find a way to formalize my thoughts on the matter. Maybe in future blog entries.....?

EDIT: I forgot to add that those functions to add scratches etc. to models do exist in some 3D suites. I am an old Blenderhead, so I mostly know it from there, and there seems to be both generic functions and well-established techniques (some macroed, some not) for doing this kind of thing. I remember last I was active on the forums that someone was making a distortion tool that used cloth-like physics to take a model and partially crush it, like a car that has hit a pole, and some add-on that could "break glass", useful for things shattering glass on impact or creating buildings with broken windows and such. I honestly do not know the state of these things anymore, as I fell out of 3D animation a while back (I decided to become a school teacher. Yeah, my life is weird...).

Share this comment


Link to comment

 

Travelling through an environment should never be a chore

Amen to that!! I think that sums up a lot of my worries about the level of detail. At some point, the detail becomes a burden, if it is not thought through right and properly connected to the concept of the game (see my previous Super Mario example as an, uhm, example). WoW especially gave me the feeling that it was trying to more show off a lot of world building detail, rather than provide me with game content. I have not played, and probably will never play, No Man's Sky, but I get the feeling that this is also the big problem for a lot of people playing that; those gazllions of worlds feel more like a distraction from the fact that the game cannot follow up on them. They become, to use your word, a chore.

 

 

As you say, NMS feels uselessly enormous, but it's not so much the size of it that's the problem than the lack of actual diversity in the procedural generation. It's a gimmick that I never really liked, and it's way too popular these days. You may make your world randomly assembled for a fresh new experience each time (on paper), but the quantity of individual components that will be put together by the algorithm is always limited, and depending on the complexity of said algorithm, you're probably gonna run into the same variations of the same components over and over again, which the player will inevitably notice when he/she hit 10-20 hours in a game like Minecraft, NMS, or any other procedurally generated universe. For some it's perfectly fine, for others like me it's a deal breaker.

A game doesn't have to be actually huge to feel huge. The Dark Souls series is and should be a lesson for every game developer, in scale management as well as in level-design.

The DS games, and particularly the first, are masters at giving you open yet restricted levels. While you can feel sometimes that you're in a tunnel-shaped level-design, forced to go from point A to point B, the truth is that every path is connected to another, and that everything you see is basically explorable. That huge castle in the back? You'll get there eventually. That forest at the bottom of the undead settlement? You'll be there in a few hours, and you'll learn to hate it. By proposing a design that's actually very restricted but with a logically connected environment that's architecturally consistant with itself, you feel like the world you're in is wayyyy too big for you alone to explore.

Still in the DS subject, there's this article ( http://kotaku.com/5874599/what-dark-souls-is-really-all-about ) that I invite you to read when you have some free time. It speaks of the main reason the environments in the game aren't a chore to explore. Namely, it's the constant possibility of danger and the pattern that you end up noticeing in the game that everything is not as it seems (fake walls, hidden ledges and secret zones). By making the player paranoid/inquisitive by design, the developers made the world seem filled with possibilities and adventure (more like gory survival, but hey, adventure sounds nicer). I could tell you very exactly how to get from the starting point in DS1 to the end boss, and how to avoid the most part of the traps and dangers, and I'm not even that big a player. Plus, I haven't played the game in over two years. That's how aware of your surroundings the game makes you. This is a result I aim at reproducing, while knowing that I probably won't even get anywhere near it.

Share this comment


Link to comment

Those are issues I struggle with a lot right now. I just got stars, planets and moons working, and I look at the screenshots and think to myself "what is content, and what is fluff?". I.e., what is detail and what is distraction. I do disagree that the size alone can be a problem (the deal breaker, as you call it), but I fully agree that the main problem is the "gimmick" of PG. In nearly all games that actually mention it, it feels a lot like those 3D movies that have one or two moments of something flying at the audience. I may in fact feel about 3D movies as you seem to feel about large-scale PG, i.e. that it is so gimmicky that it turns me off almost on instinct. I don't usually go to 3D movies, prefering the 2D version. The only one that ever did it for me on the 3D was Kung Fu Panda 2 (do not mock me...), because it had really immersive environments that never drew attention to the 3D effect, but utilized it well, nonetheless. I know no equivalent for PG in this, not game that makes PG a big deal, without making a big deal out of the fact that it uses PG. To me, the ultimate use of PG would be if you play the game and it feels like you feel about DS (I never played DS,I'm not big on fighting games anymore), that it's a tight and meaningful experience, and that what you can see in the distance will play a role. But the PG means that there are more tight experiences beyond that, which tie into it. This is perhaps the biggest reason I feel PG is so poorly utilized, in that it is mainly used to either make a simple variation for the game (as in Spelunky's levels), or that it goes overly vast. It seems like someone just decided that it's a trade-off, and that annoys me.

Man, if I actually pull this off, I'm going to split the waters a bit on how to use detail in games.... Well, petty conflicts are good PR these days, or so it seems!

Share this comment


Link to comment

I come from graphic design and digital communication, and create a game during my free time. I think you pointed out a very essential question for videogames, but this question, before games was the same for painting, sculpting, filming... When an art reaches the point to be able to imitate reality in a perfect way, new questions arise : what's the message ? Is everything useful ? Can I abstract, just put colors ? Can I rebel against ultrarealism ? Now we're more like in the middle of Pop-Art revolution for videogames, where indie devs appears all around the world and biggest games are becoming commercial martketed products. Nowadays, average videogame player is 35 years old : it is a new form of Art, and must now question itself through independant creators communities.

Share this comment


Link to comment

I come from graphic design and digital communication, and create a game during my free time. I think you pointed out a very essential question for videogames, but this question, before games was the same for painting, sculpting, filming... When an art reaches the point to be able to imitate reality in a perfect way, new questions arise : what's the message ? Is everything useful ? Can I abstract, just put colors ? Can I rebel against ultrarealism ? Now we're more like in the middle of Pop-Art revolution for videogames, where indie devs appears all around the world and biggest games are becoming commercial martketed products. Nowadays, average videogame player is 35 years old : it is a new form of Art, and must now question itself through independant creators communities.

Is the average age really 35?? Do you have a link to statistics, that sounds insane! But yes, you're right that ultrarealism is causing people to "rebel" and create non-realism games. I think it might be the Uncanny Valley effect, as stated earlier. People get uncomfortable as things become "too real", and they burn out. So they want something lighter and simpler, and old school graphics styles provide that. Plus, mobile gaming is a good platform for 80s and 90s levels of detail....

Share this comment


Link to comment

 

@Embassy of Time, you seemed to have forgotten that a 3D model is vector math.

[....]

The artist could, for example, make a spaceship. But the game could use that pristine ship to make aged, broken, faulty or other versions, without the artist needing to put in the extra work...

The PG finishes the work according to the artist's requests. That means the stored spaceship model can be fairly simple, as the PG engine simply adds the different details on the fly. Thus, tiny details (as in physical details, little scratches or color gradients) need not burden the artist much...

EDIT: I forgot to add that those functions to add scratches etc. to models do exist in some 3D suites. I am an old Blenderhead, so I mostly know it from there, and there seems to be both generic functions and well-established techniques (some macroed, some not) for doing this kind of thing. I remember last I was active on the forums that someone was making a distortion tool that used cloth-like physics to take a model and partially crush it, like a car that has hit a pole, and some add-on that could "break glass", useful for things shattering glass on impact or creating buildings with broken windows and such...

 

You are right about tools for making scratches and wear existing for 3D software, however like I said they produce bad results. If you wonder how bad these results are then just look at indie 3D games vs AAA games. Most indie developers are stuck using lower than 25K models when in truth it shouldn't matter if your model is  25K or 60K as most things up to the batch count is a once off cost.

Small improper made triangles quickly drives up the cost of both the shader and the mesh.

These tools also don't become mainstream as they ruin already well made work. If you want I could upload images showing exactly how bad the results are.

Techniques for creating these details either require a AI with human level of intelligence or circumstantial math that has a branch for every possible condition. Even if it was possible to make this you would need to program in every tool used and would need to execute the math and freeze the game while it's solved.

 

Mostly details like this are done on a material level, because it's where you can do this kind of thing without damaging the model. For example Substance painter and Quixel suite are tools that operate on this level; Both are resource hogs and can bring your PC to a standstill.

Sculpting tools are also mostly used at a material level, with a high poly mesh made and rendered as a normal map.

This simple math your talking about is extremely complex if you want it to be useful, or it has to be kept very basic and used with what ever results you get.

If you could create a tool, that allows even basic destruction on a mesh level and it produces usable results, then your engine would be worth millions as a 3D modeling tool already, even before a game is made with it.

 

About breaking or morphing 3D objects: This is easy, braking a object from a other piece is just a matter of using what you have. Changing the form of a object is as simple as adding and subtracting values to the existing ones.

Making a scratch on a 3D model involves removing a part adding a new part from a reference then stitching it in. Mathematically this means removing a part of an equation replacing it with a completely different equation and only ending up with the intended change as a result.

Because of how complex 3D models are, adjusting even a small piece will effect every thing around it.

Share this comment


Link to comment
Do you ever wonder why, if the things you are imagining but have yet to implement are possible, triple A companies with multimillion dollar budgets and teams of hundreds of skilled developers and artists haven't made such a game? I'm sure they would love to as content creation is currently the most expensive part of the development process by far. Do you think it is really just a lack of imagination on their part?

Share this comment


Link to comment
I'm commenting on my phone so being brief. It occurred to me that could come across as terse. That wasn't my intention and I admire your enthusiasm.

Share this comment


Link to comment

@Scouting Ninja: If I ever described the math as "simple", that's my bad. It's not simple, but everything I look at says that it is, at least, possible. There is a lot that can be done with textures, true, but at some point, it creates the wellknown "flat feature" effects that have become a bit of a joke. I recently watched people play both Subnautica and Star Citizen, and I cringed when they got close to a round hatch or the like, and it became painfully clear that it was just a painted on texture. I don't think the problem is that this is something extremely difficult to achieve (difficult, yes, extremely difficult, no), but that the investment in time and math just is not worth it for most professional studios. I do it because of an obsession with math, but they have literally no motive to, because people will enjoy the games with or without it. But that means that the results from pulling it off can never be used to move beyond that and into more advanced territory. Which I guess means it's entirely up to me to prove it feasible :o

Share this comment


Link to comment

Do you ever wonder why, if the things you are imagining but have yet to implement are possible, triple A companies with multimillion dollar budgets and teams of hundreds of skilled developers and artists haven't made such a game? I'm sure they would love to as content creation is currently the most expensive part of the development process by far. Do you think it is really just a lack of imagination on their part?

I wondered that for ages. Then I learned about the industry. AAA games development is based almost entirely around safe bets, and they follow very narrow formulas. Ever notice how a lot of expensive games (to develop. But costly to buy, too) look very alike? Yeah, not a lot of risk taking there. If the "money means they can do better" argument was solid, there would be no indie scene at all. Sadly, what I talk about might be a huge boost in technical terms, if it works, but it has no real market value over just producing a new first person shooter with some quick-to-design characters :( Also, most people in the AAA industry are becoming more and more specialized, so turning the ship in new directions is supposedly a nightmare. They stick to what they know. And to my great frustration, nobody really knows much about the things I try to do, which means I have to work entirely from scratch. But maybe they'll buy everyting if I succeed. One can always hope :)

And thanks for the comment about using the phone, the first comment did in fact seem a bit blunt. I now know you only mean well :)

Share this comment


Link to comment

You are right about the AAA not wanting to take risk, lot's of people starve when they do. You are also right about them sticking to what they know, this is human nature.

What your doing is great, even if it's a bit naive. However you have shown skill and determination to take these problems on so we will be watching; although mostly in silence unless you ask for comments.

I also don't think the things you are trying to do are impossible, I believe they are more than "extremely difficult" as you put it. As hard work is the easiest wall to climb, if that was all that stood in the way then games like this would be the norm.

I have seen a few projects start like this and although many of them haven't reached there goal, all of them achieved some form of greatness. So I really don't wan't to discourage you and will be around if you ever need advice on art.

Share this comment


Link to comment

What your doing is great, even if it's a bit naive. However you have shown skill and determination to take these problems on so we will be watching; although mostly in silence unless you ask for comments.

I need to ask, who are "we"? Are you writing on behalf of a group? ARE YOU THE ILLUMINATI??? Jokes aside, thanks for the comments and confidence, and I'm always interested in your comments. But seriously... who are "we"?? :o

Share this comment


Link to comment

I need to ask, who are "we"? Are you writing on behalf of a group? ARE YOU THE ILLUMINATI??? Jokes aside, thanks for the comments and confidence, and I'm always interested in your comments. But seriously... who are "we"?? :o

 

I am typing on behalf of myself and my colleagues -we are working long night shifts- to fill the time we have been talking about your post and procedural content.

Share this comment


Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Advertisement