Elijah Newton said:
Whoa - the two of you worked on this game? Cool! Congratulations on having a finished product. Also, thank you ThriKreen for the breakdown on what's pushing the processing envelope. That was really well presented. Plus I dig your profile names.
I don't work at Maxis (Have tried though! I like working on environments) but I've been in the industry around 3D engines long enough to (mostly) understand how they work. Having three (point five) published games also helps.
Like people saying the 1.8gb (beta)/3.8gb (full) game install size is "small" by today's standards. Uhm, no, it's about right for a game like this. Of course, I could be showing my age here when I used to remember thinking 9 5.25" floppy disks for Kings Quest IV was "big".
But seriously, a large majority of data for a game nowadays is from high poly meshes and textures, animations, facial animations, voice clips, level data with little being reused between levels. You're not talking about one texture a model, but one for the diffuse, specular, normal, and if an environment which is static, a shadowmap per object in the level (not model, as it can be used multiple times in different lighting situations).
SimCity most likely made use of material instances and reusing textures to reduce resource and memory usage, making things procedural. So it would do things like not bother with normal maps (due to scale), and since all the lighting is dynamic, there's no point in having baked shadowmaps either. So you've just reduced the texture usage in half. Then, using shaders they can apply colour tints to the materials, so instead of one unique texture, one mesh for each skyscraper variant, you reuse the one mesh and and tint the material for it 10 times.
No cinematics, no long winded speeches, just "SimTalk", the data for all the maps is probably smaller than the first level in Mass Effect 1. Since, you know, it's a bare terrain mesh until you start slapping your city down. So yeah, makes sense it's under 4gb in size. Everything else is handled in real-time based on what the player does.
Elijah Newton said:
I understand why there's the smaller size due to those limitations, but could either of you explain the reasoning behind prioritizing the graphics? Because (and I really only have a general background, so pardon if I'm off) I think the simulation part of it probably isn't impacting the performance that much relative to the demands of rendering it as you chose to do. That display of things, though - much as I like graphic artists and think they deserve getting a paycheck - doesn't really impact the game that much. I mean, what are players really getting in terms of an experience from rendered 3D models w/lighting and shadows, etc etc, that they wouldn't get from sprites? In terms of presenting information (which is what graphics do whether they're an excel spreadsheet or or hi-poly model), it seems that there's a very limited gain, if any.
Well, I wouldn't necessarily said they prioritized it. They did say in various interviews that they wanted the scale of the game to be midway between SimCity 4 and The Sims, and doing things like not abstracting the individual Sims at all, and allowing you to zoom all the way out and all the way in. There are concessions with that level of detail, so you have to compromise for how much data to manage and have a target hardware specs for it. People did notice that SimCity is not a console game, right? The RAM capacity on the Xbox and PS3 won't cut it for this game.
Using 2D sprites for that level of detail starts resulting in more data usage than merely using a 3D model, since you have to render out all the possible camera angles and zoom levels to reduce blocky pixels. How do you light it, if you wanted a working day/night cycle? After a certain point, it because easier and performs better to make it 3D instead.
And of course, the pretty factor - do you really want to play a game that is a spreadsheet a la EVE Online, or be able to zoom in and out and pan around the city skyline in full 3D? How do you attract new players to the franchise, and not make it merely a recycle of SC4?
Elijah Newton said:
So my question to you is: am I completely off base? Did the decision to make shiny graphics not impact the complexity of the simulation / game at all? If it did, who makes the final call about the direction a game will take when there's this kind of fork in the road? Is it all set in the design documents at the start (players will zoom in / out, models will need this level of fidelity) or does it get worked out as development progresses?
Well, you can toggle the graphics to low [http://www.thrikreen.com/temp/simcity/Graphics_Low.jpg] and not see the individual vehicles or lights and such. For comparison, everything to max is here [http://www.thrikreen.com/temp/simcity/Graphics_High.jpg]. I'd imagine they'd be pretty aggressive on the LODs so zoomed out that much isn't a good representation of things. I'll have to do another one up close.
Obviously the sim data is still there, just not represented graphically. I think the main impact is in just general computer power, since there's such a wide range of hardware specs out there.
The game does seem to run better now than it did in the betas, but that makes sense as the beta was probably a build from fall of last year, while the one we have now went through several months of lockdown, optimization, polish, and certification. Of course, now that we can spend more time with it, I can take the time to analyze it more, and could be that my previous observations will have to be revised.
Elijah Newton said:
I'm not looking to criticize, though in the spirit of full disclosure my bias is to say that this SC represents a departure from earlier versions... but that's ok. It's not my project / job / etc. I'm really just curious.
It's still the same city management game as before, with some new twists (trade, collaboration with other cities, world impact). The thing with game dev though is, that you get complaints if all you do is recycle from the previous iteration, and complaints if you change things up - it's an all too common "Damned if you do, damned if you don't" situation. Which audience should they listen to? At some point they have to stop and go "OK, we made these decisions, the game has these features - here it is and we think it is fun for you." And people are left to make their own opinions on it, and forgo playing if it doesn't appeal to them. One does not complain that WoW is not WarCraft 4, or StarCraft is not Warcraft, all three games have different design directions, even between the two RTS games.
Elijah Newton said:
* I know there are only so many hours in the day, but IMHO this really ought to be played purely from a professional development standpoint. Someone's doing something really well here despite staggering limitations. Seems like a learning opportunity.
Perhaps, but I've also never played any Elder Scrolls game or most MMOs either, but I can gleam enough from people talking about them to have a basic understanding of the mechanics.
And I'm busy working on my own games.
Elijah Newton said:
** But oh god, as a player, particularly a new player, you spend so much time battling a non-intuitive UI. I'm not even talking about the ASCII presentation. The menus alone would give whoever does this for Maxis night-terrors.
See, reason why aesthetics matter.
That and UI design is HARD.
Elijah Newton said:
*** And I see signs ( subjective, of course, I could be completely off base ) of attempts to compensate for a less complex game by inserting goals for the player to reach.
The goals actually help give some direction for newbie players, as a vet can ignore them. One of the common problems from sandbox games like this is a common game design issue. Give a player no choice, they'll complain of the lack of it. Give too many is just as bad, as the player runs into the confusion over what to do next.
It's one of the reasons I tend to avoid sandbox games like Minecraft, DF and TES, I can't decide what I want to do next. At least SimCity has a general goal of "Make a big, happy city".
And the flip side is, while the simolean bonus for completing the goals can help, you have to decide if spending $500,000 on a landmark for the $10,000 bonus is worth it.
Especially if you don't have the infrastructure to support the influx of tourists as a result of doing it.
There's quite a lot of complexity from what I've seen, but I haven't delved too deep into it.
