[Kickstarter] Project Eternity by Obsidian (Baldur's Gate, Fallout, Planescape, etc.)

I thought my explanation earlier explained all the difficulties of going full 3D. What was wrong with it? What kind of proof do you want?

It is pretty much impossible for real-time 3D to match the fidelity of pre-rendered background. Or, like I said, you could literally have an isometric game that renders everything in real-time (matching whatever you did with your 3D effort), but just fix the camera (which should free up some resources).

3D visuals can certainly be very impressive, but it isn't easy. That wall that looks fine in isometric view suddenly looks like a N64-level texture when you zoom the camera in. Lot of work needed to get high-quality visuals from all perspectives (instead of just one, fixed perspective from high above).

Look at what one guy did in ~2 weeks, just as a small distraction:
<iframe width="560" height="315" src="http://www.youtube.com/embed/bRylpn1sqxk" frameborder="0" allowfullscreen></iframe>
I believe he might have spent some of that 2 weeks thinking about story, gameplay ideas, etc. He just did this on his own, due to all the interest in games like Wasteland, Fallout, Last of Us, etc.

Another demo of his I believe:
<iframe width="560" height="315" src="http://www.youtube.com/embed/dwuWeRdha3I" frameborder="0" allowfullscreen></iframe>

His actual game, which has a bit more polish/animated effects (again, done by only him I believe...well...outside of maybe the the voice over stuff):
<iframe width="560" height="315" src="http://www.youtube.com/embed/61ILFgiYaRU" frameborder="0" allowfullscreen></iframe>
(this game is called Stasis, in case anyone is interested. Looks very interesting IMO, assuming you're good with isometric view)

Meanwhile, we have 3D games that take years to complete, and sometimes have up to 600 people working on them. Not the best example, but I believe PD stated it takes them ~30 days just to complete a car model for GT5. 3D usually requires expensive 3D tools (Google tells me CryEngine 3 costs $1.2M to license, and I seem to recall UE3 being around that too; throw in systems for physics, 3D AI, animation, etc). And you can still get games that look pretty ugly (Bethesda/Bioware/etc.).

If you want me to pull a quote from Feargus Urquhart saying that Project Eternity would cost 10x more if they ditched isometric, then I probably won't be able to do that (although I wouldn't be surprised if he said that, but I don't care to hunt down a quote when it seems like common sense).

Where was this mixed reaction? Everything I've seen has generally been very positive. Maybe some minor complaints about overall color (i.e., complaints about art direction I guess), but that's about it.

Just checked out Eurogamer's article:
http://www.eurogamer.net/articles/2...ct-eternity-screen-looks-shockingly-realistic
Eurogamer describes it as "jaw-droppingly realistic," and pretty much all the comments say similar things. Far from mixed IMO.

(unless you're talking about people who are complaining about the isometric view, and not necessarily the visuals, but I don't particularly care about those...my point was they produced something that most/all consider easy on the eyes...unlike say some of their 3D games)

 

New update includes talk of a documentary. Should be good. Almost at $3.1M (think it is there if you include Paypal).

Continued discussion with Dei, which I'll spoiler tag since I doubt most of you care to read through it (I did post a nice tech demo demonstrating dynamic lighting with pre-rendered backgrounds though, which might be nice to check out):

Spoiler

That quote was describing an approach I hadn't really thought about, since it seems kind of silly. Why render things in real-time, but then fix the camera? It should be just as cheap as if you didn't do the camera, but surely there would be better ways.

Notice more explanation/specifics prior to that. This includes a misunderstanding on your part (I believe) about what's actually being done here. PE isn't using 2D hand-drawn sprites, so you don't really have increased costs by going higher-res (that's the only way your claim made any sense to me). Characters, enemies, items, etc., will be rendered in 3D in real-time, so you can animate/model/implement physics much like you would in a "normal" 3D game.

I assumed you misunderstood that point, and why you thought producing PE would be more costly than I was claiming (if they did go that route, I'd agree 100% with what you're saying). I've interchangeably used 2D to refer to the method PE is using (and fixed camera, isometric perspective in general), and 3D to use the method you're requesting (something like NWN2/Dragon Age/etc. which has a controllable camera). Those probably aren't the best terms to describe this though, given that PE's assets will begin as 3D renders and the game will in fact include tons of real-time 3D rendered "stuff" on-screen (characters, enemies, items, etc.). I don't know if that's led to any confusion.

Make the game like NWN2 (I'm assuming that's the approach you'd prefer). Update visuals for 2014 standards. Zoom out to isometric view. Now remove (or never implement) the camera system (cut QA time quite a bit since you only need to QA one perspective), and don't bother working on hair/cloth physics, don't bother modeling small things like individual fingers, no need to create 500 different weapon models (while noticeable up close, they wouldn't be from isometric...just need the basic shapes), no real need for LOD algorithms, etc.

And again, I think that's a silly way to do it. I'd rather they prerender backgrounds. Let one guy do it in his spare time for 6 months on his home PC instead of paying a small team of artists/programmers to build 3D environments that would need to be rendered in real-time, and only possible on expensive CPUs/GPUs. Tim Cain actually spoke briefly about the direction they're going and the advantages it brings ("exquisitely detailed environments without the polygon cost" and ability to do lots of animations).

You're right, you can avoid animating (or even completely modeling/texturing/etc.) certain objects that won't be seen. But when you have a user-controlled camera, a LOT more things can be seen. That's the point. And why things can get so costly. Imagine in the "1000 years later" tech demo I showed, if you had to work on the interiors of those trains (you should clearly be able to see inside them if you zoomed in). There's more stuff you can see in 3D (due to direct camera control), and because of that, you have to work on more things.

As for animation and physics, since the characters/enemies/items/etc will be 3D (only the background won't), they should be able to use those techniques. They won't have to hand animate everything (though they can if they decide that's better for some reason). I'm assuming you meant animation during combat, items dropping, traps going off, etc. Although again, you can get by with "good enough" from isometric view (seeing the same attack animation from close up probably won't fly, but would be acceptable from isometric). Also don't have to worry about hair/cloth physics and all those things that don't matter unless they look terrible.

The only problems would be with interacting with the objects in the 2D background. For example, if a character dies and falls to the ground, there won't be an actual "object" to react to their fall, so they'd fall the same on multiple types of terrain. Obsidian could model certain things in to fake this (e.g., actually create a 3D log for characters to fall on, instead of relying on prerendered log), although not sure if it is really worth the extra polygons/CPU cycles. If done right, this likely won't be noticeable.

You can have problems with lack of depth and perhaps the background being too "static" (though I think both can be addressed with the right style/design). But the actual visuals themselves? There's not really any comparison between the two. One image has to be produced in a fraction of a second. The other can take days/weeks/whatever to produce, using expensive rendering equipment (keep in mind real-time visuals may look good on $500+ GPUs, but they'll look like a jagged, blurry mess on integrated graphics). On top of the rendering, you can have artists paint in details to increase the amount of detail with a pre-rendered image. If you're comparing the details/visuals in a real-time screen vs pre-rendered screen, I don't really see how anyone could prefer the former (hell, that's kind of the point of putting out "bullshots").

Now, you may prefer real-time because you can manipulate the environment, change perspectives, etc., but when it comes to the actual detail in each approach, there's no comparison to pre-rendered backgrounds.

You can get some pretty nice dynamic lighting with pre-rendered backgrounds:
<iframe width="560" height="315" src="http://www.youtube.com/embed/-Q6ISVaM5Ww" frameborder="0" allowfullscreen></iframe>
And, of course, you could actually render everything in 3D and use the same lighting you would otherwise. That said, I'd prefer pre-rendered backgrounds with lighting techniques shown in the video (although likely much more advanced). Should look as good, if not better, and frees up resources to do other things (or allows game to be played with integrated graphics).

Loss in quality can certainly be pretty bad, though it is unlikely you'd actually see it in a 3D game (with camera control). If the game allows you to see something from such different perspectives, they'll swap between low-res and high-res versions (and actually, probably multiple versions in-between). So you'll get something that looks fine in isometric (like you said) and doesn't take a lot of resources (so you can render all that stuff in real-time with reasonable performance), and if you zoom-in, you can see the high-res wall (or whatever). Not saying anything new, but with this, it should be clear that devs would have to create multiple versions of said asset. In other words, more work, and possibly going for assets that will rarely be seen by players. Guess some might decide to not worry about it for that very reason, though then you get things like those Dragon Age ground textures. I'd prefer not getting any horrible textures, and that devs don't bother spending time creating a super-high quality version of someone's shoe.

You can create an isometric game that looks amazing with very little resources. Kind of my point from earlier.

Don't see too many NWN2/Dragon Age style games made by teams of one and worked on primarily as a hobby (and even if you did...they likely look not so good...but I did say Dragon Age style, so I guess that's OK).

I gave both CryEngine and UE3 as examples (CryEngine has some more "modern" features that I'd prefer to see in a 2014 game, maybe more comparable to UE4). AFAIK, they both cost about the same (I want to say Epic asks for more royalties, but I might be wrong). In addition to that, Obsidian has already stated that it would cost too much to use their own internally developed engine with PE (and Onyx has already been fine-tuned for a game you're talking about). Engines used for these games are expensive.

Never said low budget games couldn't look nice.

If I said anything along those lines, it would be that a low budget RPG with a ton of content (like PE) would likely not look nice unless they went with isometric/pre-rendered backgrounds.

Journey looks great, but it only has a few hours of content. It took about the same amount of time PE is expected to take to develop, and I'm thinking the team size for both is/will be roughly the same (15-20 people each). So maybe Obsidian could go that route, but hopefully no one is expecting a RPG with only 3 hours of gameplay.

Couldn't really care less what 4chan thinks. Prefer opinions from Obsidian's forums, NeoGAF, RPG Codex, or even this forum. None of these seem to have "mixed" reactions.

I was also speaking only towards the visuals of that screen (i.e., "Look, Obsidian can produce something that looks beautiful."). I get that some people still don't like the decision to go isometric. Although even that seems to be a small minority in the places I mentioned above, but no point in discussing that further IMO.

Just provided a random result from Google that demonstrated the reactions I've seen. Majority (practically all) of the comments I've seen regarding that screenshot (from journalists or fans) mirrors what was written in that article.

There have been multiple points, as this discussion has evolved as more claims have been made.

I initially responded, due to your claim that isometric games are obsolete. I disagreed with that claim and tried to explain how gameplay would be as good (or better) with isometric with this type of game. Gave up on that point since I don't think we can see eye to eye there. That can definitely be subjective, so while I disagree, can't really do much about that.

The last few posts, I've been addressing this (which isn't quite so subjective):

I feel like I've demonstrated enough proof to show the problems with 3D.

How about this?

Mass Effect/Dragon Age/Modern AAA Games: $40M+ (average budget for modern game is $15M-$20M, though I've seen as high as ~$28M)
Fallout/Project Eternity: ~$3M
(While Fallout and PE are direct from the source, more or less, the others are based on studies/estimates as you rarely get that directly from the studios; if you disagree with them for some reason, just look at their staff sizes, multiply salaries by development time, and you'll likely come up with similar numbers)

Not a good example? How about Obsidian's own Dungeon Siege 3? DS3 seems close to what you're wanting, but just with different gameplay/art style. But that's yet another "full 3D" game with a much bigger budget than what PE has (don't have numbers, but Obsidian have said as much).

Can you provide a modern party-based RPG (with towns, quests, complex classes, etc.), built the way you'd like (real-time 3D graphics with camera control and all that), which was built with a budget around $3M-$5M? Every example we've discussed probably has a programming team that is almost as big as what the whole PE staff will be (or the staff for the original Fallout).

 

New update:
http://www.kickstarter.com/projects/obsidian/project-eternity/posts/328976
At about $3.3M (excluding Paypal) with about 24 hours to go. They were at 9 levels for their dungeon, but I think since they just got 65K backers, it goes to 10.

Obsidian playing D&D:
http://www.ustream.tv/channel/project-eternity-live-stream

(working on another response to Dei, maybe the last, which I'll also spoiler tag...don't want to crap up this thread while the project is winding down)

 

Response to Dei's post:

Spoiler

Please don't tell me we've been posting without even understanding what we both meant about 2D and 3D.

If this has been about 2D sprites versus 3D renders, I can bail out of this thread. There's absolutely no disagreement from me there. I apologize if I misunderstood your point (although I am curious why you'd bring this up when it has little-to-nothing to do with this thread...PE isn't using that approach, and few isometric games in general follow that approach...for obvious reason).

If you're also arguing that "Dragon Age" style (what I call 3D RPGs with controllable cameras) games don't cost substantially more than the approach Obsidian is going with...then I think you're wrong. I'm not sure how I can convince you how though, so...whatever. I know we're both sick of this.

Don't really time them, nor do I sit down and complete it in one sitting, so no idea how much time total. The last one was by far the most time I spent. IIRC, I didn't read your reply until around ~1PM yesterday (was busy in the morning and then got caught up with the stratos jump). I seem to remember reading and working on a response while watching the Rockets yesterday. Shared time on it and other things during the afternoon/evening, and finally posted at 6PM I guess. Some of that time I spent "researching" as well (e.g., looked up multiple things regarding dynamic lighting in isometric games like PE, although only posted one "find").

 

I agree, but...

And I agree...

Huh?

It is pretty much the same thing as PE (although it actually has some character sprites in the demo, unlike PE). How is that 3D, while PE isn't?

I literally took that video from a thread about isometric games with pre-rendered backgrounds (i.e., PE), and how you could apply dynamic lighting:
http://forums.obsidian.net/topic/60...ning-large-pictures/page__st__80#entry1203609
(don't read too much of that thread though, since several people are posting their assumptions about how this approach is much cheaper than doing everything in 3D)

Between that, comments like "2D is actually harder in high res," talk about sprites (PE doesn't use sprites...at least in the way most refer to it), etc., I got really confused about what you meant. The only way that makes any sense is if you were talking about 2D sprite tech (where you couldn't use 3D models in game for animation/physics/etc., and had to actually hand-draw everything for every interaction, not to mention handle lighting somehow...though seems like there are methods for lighting 2D sprites too). Hence the confusion.

Either I'm confused, you're confused, or maybe both.

I could say the same.

I've provided several games (with budgets) showing "cheap" costs for isometric 2D games and pretty much nothing but big budgets for "3D" games (and when budgets aren't available, I've mentioned staff sizes, which directly correlate with budgets of course). It seems much more reasonable to me that PE (and every other party-based RPG like it, such as Wasteland 2, Shadowrun Returns, etc.) is going "2D" because 3D is in fact a "big hassle." Why would not a single game like this be made the way you're requesting, especially given how obsolete the isometric approach is?

All it would take is a single game to prove my argument null and void. If some team of 15-30 could produce a 3D game with features/content similar to PE (preferably including "jaw-dropping" visuals, but perhaps even "**** mountain" visuals could pass) on a budget not much larger than PE (whatever is less than "terribly expensive" compared to ~$1M), then it would be obvious that you could do 3D very cheaply. In that case, maybe Obsidian is just trying to bring back Infinity Engine nostalgia with PE. But at this point, all I see are a bunch of obsolete, 2D isometric games and expensive, yet-not-so-obsolete 3D games.

CDPR could possibly prove me wrong if they make a party based RPG (and I'll gladly eat crow, that would be awesome!), although again, Polish salaries are an advantage. Staff size seems large (more comparable to Dragon Age than Fallout or presumably PE).

Needed to clear other stuff up first. I still don't know what to say about #2 given that you think the dynamic lighting video I posted was from a 3D game. If you can rebut any evidence I show with "that's 3D!" then I don't really have much of a shot of winning that argument.

As for #1, I already said there's no point in debating that. I actually more or less agree with it. I just think there are trade-offs, and depending on the game, the advantages aren't really worth it and/or new disadvantages pop up (e.g., more difficult to manage party). That's largely a subjective debate though, and we already had that. Though again, I find it funny that so many experienced developers are putting out isometric RPGs, and the fans are so eager to play them.


On a minor side-note, I just realized Obsidian have said they're playing around with zooming in/out with PE. Didn't know that...interesting. Would like to see that. Wasteland 2 does that, but their approach is a little different.

 

Short Dei response (final one, I'm done, as I'm sure he is):

Spoiler

And for some reason you're assuming PE won't be doing this? No polygons rendered with complexity only limited by sprite resolution. Seems like exactly what these guys would want.

(I wouldn't doubt it ToEE or maybe ID did it actually, which seems to be the inspiration for a lot of the visuals.)

And again, I find it funny that no one in the Obsidian thread bothered to make this point...maybe since they all realize it is perfectly applicable to PE. I'm sure that's just yet another wrong assumption on my part though, and the day/night cycles in PE will take a whole lot of work and probably suck.

I doubt they paid half of the programmers to sit around doing nothing for a couple of years. So yeah, I don't literally know what they were doing, only how many programmers/artists/designers/etc. worked on the projects. And sure, maybe they just threw money away on nothing, but that doesn't seem reasonable.

Again, if I'm assuming wrong, surely someone could have put out a game to prove your point.

NWN2? Yeah, they had almost as many programmers/scripters as Fallout (and likely PE) had people total.

Sure, but if your assumptions are true, surely we'd have something. Isometric games are obsolete (and in fact, until Kickstarter, were like adventure games and others in that publishers wouldn't back them). Yet...nothing.

Well, can't beat that argument.

Huh? I just don't care to discuss that since we'd get no where (pretty clear by this discussion, which is about something much more objective).

 

Last (or near last) update from Obsidian before the end of the Kickstarter:
http://www.kickstarter.com/projects/obsidian/project-eternity/posts/329689
Small part:

No final stretch goal (just going to make everything better). They might include some of the things being requested though.

There is a behind-the-scenes look at the screenshot they provided earlier, which is pretty neat (can't wait to see more later).

They put out a $350 loot bag as a new reward tier, but it already sold out.

At $3.6M (excluding Paypal), and about to get another level for the dungeon (70K backers). 6 hours to go. They'll have a livestream to count down the project at noon PDT (guess that's 2PM for us central people).

 

Initially rendered that way (possibly with some expensive equipment), but they essentially take a "picture" of it, paint over it (at least in this case), then use it as a 2D background. When your play it on your end, it is just using that 2D picture (although in this case, your computer will also be rendering 3D objects on top of the 2D picture, like the enemies, characters, etc.). It isn't rendering those polygons or anything (probably can't, at least not in real-time).

Agreed.

I also love being involved in the development process in general. Just to be able to "tag" along and find out what is going on while Obsidian makes this game will be interesting (and if it is a huge failure, maybe seeing it blow up will be worth the price of admission). The Double Fine documentary has been great, as has the multiple updates from other successful Kickstarter projects. Better than going 2-3 years of dead silence, not even knowing if a game is even still in development (what the hell is going on with Final Fantasy Versus XIII?).

I'm still a little mad because Jane Jensen keeps hinting at a "Mystery Game" which is apparently almost ready to be released, yet the publisher doesn't want to reveal it yet. Such a huge difference in seeing her describe her Kickstarter project and that project.