Stunts: How Could They be Changed?

[Shogi]

He misspoke. What he should have said is:

i feel like making third person actually work well would impede the whole project.

Better?

[Vino]

Then maybe I just don't get the logic of how it could actually do that.

[Shogi]

Time spent working on a functioning camera rather than the rest of the game.

[TheOtherChris]

Wait, I haven't played Uncharted 2 but doesn't that do a lot of things similarily to what we've talked about?

[bryt]

i should probably outline reasons for my views on the third person feature. please mind that I actually really love the idea of having a third-person camera system in the game, especially for the stunting... but my arguments mostly center around the amount of work in would entail, or the technical issues.

note that i'm not really against the idea of having third/first-person mixed in the game because i think by having them both you can get away with fudging the accuracy of the visual feedback and people will forgive you more in such case. but by choosing strictly one or the other, i think the camera work and animations need to be tweaked much more in order to satisfy their overall purpose.

Warning!!! Wall of text:

CAMERA DYNAMICS:
every game i've ever played that had a third-person camera system can be grouped neatly into two categories; crappy, and passable. note that there really isn't any such thing as a "great" third-person camera, since they all have their faults and many of those faults are view subjectively by different gamers. in the case of bad camera systems, most of the time their main weakness is being too simple combined with too stupid. Having a camera that is always focused on your character dead center, with no control smoothing/dampening, and no collision testing or surface shading features, etc.

The passable camera systems i've seen have always managed one thing really well, the player focus. However managing player focus always seems to work against player-self awareness. So you might see your objective, but you cant see where you are standing and whether you are about to fall off a ledge. The best examples of games that did this right might be Uncharted, Vanquish, Dead Space, Batman: Arkham... , etc. These are all recent games for a reason, since I don't think developers really understood third-person mechanics very well before this gen. It's worth noting that each of these games have camera systems that could mostly be interchangeable. They all need aiming support (which is why I excluding other good games that don't), but they all also offer the ability to see yourself from any angle and get an idea of how you fit in your environment.

In these examples, the camera has two distinct modes that the player can use freely at any time. I'll call them exploration, and aiming modes. In each example the aiming mode is tight like a firstperson game, the camera moves up to at least as close as the upper torso... and aiming sensitivity goes lower to allow more precise shooting. In exploration mode each game has its own variety of implementations but the general rule seems to be easy/quick looking with low precision, and user contextual anticipation (Vanquish does this really well). The dynamics of the camera in exploration modes are too complicated to really summarize here.

I have no problem with the camera in the aiming mode, thats easy enough to emulate. But the exploration mode sounds like a nightmare *to create* for a stunt based game. And in order for the third-person camera to feel functionally worth it I think you really need both camera modes.

ANIMATIONS:
with a FPS camera the player's arms/weapon can be made to do whatever we wish in the v-model while performing exotic stunts. this means that if we want to enable people to shoot each other while doing cartwheels, then there really isn't any reason why we cannot support this visually. just have the v-model show the player's shots going off, and have the full body animation do a detached cartwheel and spam animate his weapon going off and pointing in the general direction of his aim. the arm need not show accurate aiming in this case, it just needs to give other players enough feedback to know he's shooting in their direction.

however if we rely only on third-person animations, somehow we have to make the cartwheel look functional when the player shoots his weapon. this would probably mean all sorts of crazy blending and maths that calculate the blend weights along the cartwheel or some other voodoo magic. or, it will look like ass and thats all we get... ass.

if we use a mixture of third/first-person modes then nobody really cares if the third-person animation lacks accuracy. if you care, go first-person and everything will look tight. if you like the third-person enough to look past the minor lack of polish, then fine. everybody wins. which includes the animators since they only have to worry about making the v-model anims accurate, while having a little bit of creative license with the full body stuffs. the cartwheel is just a hypothetical, or course.... but this argument can be applied just about every action the game may support.


If we plan on putting a "good" third person camera in the game, then I would advise to treat it like a major feature that should be prototyped early. The technical end alone is enough to scare me off; and then we have to figure how to support it with animations. The way I see it, its either going to be crappy, or its going to be wishful thinking that I cant be bothered with.

[TheOtherChris]

Wait, I haven't played Uncharted 2 but doesn't that do a lot of things similarily to what we've talked about?

Disregard that, it looks like a clusterfuck (http://www.youtube.com/watch?v=FZG1ICEX ... re=related).

[Vino]

Okay. See this is why I wanted your feedback on this particular issue.

CAMERA DYNAMICS:

If we do an aim-in mechanism, isn't this solved? During aim-in there's a tighter camera over the shoulder and during stunting there's a looser camera that shows the whole body.

Not that I'm really worried about the player needing to see his feet since there's not really a lot of exploration or jumping puzzles in this kind of game. All that matters is the ability to keep the reticle on the target and have a vague understanding of where all of the environmental obstacles are in relation to the player.

ANIMATIONS

You're saying that in the absence of first person animations, all third person animations need to be extremely precise. I would agree that they need to be more precise than they would otherwise but I don't know about extremely. I would be okay with a 30 degree error in precision of where the arms are pointing so long as the feedback from the bullet impacts is string enough. After all I feel like if we lock the camera to third person then the player's eyes will be on his target most of the time, since this is such a gun-heavy game. (If it were melee heavy that'd obviously be different.)

Also, while we may do a bit more work on the animating side for the player models, in my opinion it's worth it for two reasons. For one we're not using Source engine we're using Unreal (probably) which has a much better animation system, and for two for the extra work we add in animating the player model we cut out a lot of work in creating the first person v models and animating them. I think animating a first person model requires much more work and precision than animating a player model, since the action is always up close to the camera and we're always on the right side of the uncanny valley as far as realism in character movement is concerned.

One of the main reasons I'm thinking this is to learn from lessons in CF. We could have done that game completely third-person, in my opinion, and cut out a lot of work. Think about all of that careful blending you had to do for the first person view, we wouldn't have to have things be nearly as spot on if we don't have a first person view.

[Wish]

Take the sure thing.

[bryt]

One of the main reasons I'm thinking this is to learn from lessons in CF. We could have done that game completely third-person, in my opinion, and cut out a lot of work. Think about all of that careful blending you had to do for the first person view, we wouldn't have to have things be nearly as spot on if we don't have a first person view.

You and I are thinking in the same general direction then, we're just coming at it two different ways.

edit: This won't be very comparable to CF in that we would use actual v-models this time (locked to the camera). CF was a nightmare because the v-model WAS the third-person model. We hacked it a bit for better flexibility in this regard, but it could have been much easier to make better quality v-models if we went a more traditional route.

I disagree about the v-model issue. I don't mean to suggest the v-model anims will be easy but I don't think that it will be too hard to make them not only good, but even awesome. In fact i'm fairly confident that the v-model anims could be so awesome that they really overshadow any other animations we could put into the game. The v-models are always in your screen, and they are arguably the most important visual cue for feeling "cool" will playing a game with guns. By using v-models we streamline our gateway to that feel good nerve in the player and tighten our scope for polishing the effect. Third-person anims that would support a first-person game can be spammed for the most part since when you see your opponents or teammates they are usually on the move and you don't have a lot of time to really focus on the animation qualities. This factor gives us a huge amount of leeway in setting our quality standards as a whole.

Plus having v-models opens up our design to allow for things that wouldn't really make sense if you tried to make them work cohesively in a strictly third-person game. Simply put, you can allow people to shoot each other while performing stunts that wouldn't normally allow gunplay, or not. With v-models the decision is a simle yes or no, and you get very little art content consequence for changing your mind.

[Vino]

Wish, you are not contributing to the discussion.

In fact i'm fairly confident that the v-model anims could be so awesome that they really overshadow any other animations we could put into the game. The v-models are always in your screen, and they are arguably the most important visual cue for feeling "cool" will playing a game with guns. By using v-models we streamline our gateway to that feel good nerve in the player and tighten our scope for polishing the effect.

This is the obvious benefit of first person v-models. I think though that it comes at a cost of a lot of work. Here is a graph of development difficulty in my opinion in illustrated form:

Code: Select all

1st and 3rd:
V Models:   [=========]    (9)
3rd Models: [===]          (3)
Total:      [============] (9 + 3 = 12)

3rd only:
V Models:   []       (0)
3rd Models: [======] (6)
Total:      [======] (6 + 0 = 6)

This is not including the extra programming testing and design work for the first person view. So yes, the third person models are much easier to produce if you can rely on view models, but the additional work to make player models of acceptable quality is much less than the total sum of work to include two view modes. It also makes for simpler implementation and design. So the way I see it the total difficulty is substantially less without first person.

Whether or not this is worth the work saved is an important question, but not the one I'm asking.

Simply put, you can allow people to shoot each other while performing stunts that wouldn't normally allow gunplay, or not. With v-models the decision is a simle yes or no, and you get very little art content consequence for changing your mind.

What? I don't get the connection here. How is there any functional difference between 1st and third? I thought it's purely a visual an aesthetic difference. There's no reason you can't have shooting during stunts with third. I don't really see an impact on art here.

[bryt]

i estimate your third-person anims demand is low by at least 1/3rd of what would be needed for the game to not look like ass (remember this character model would be right in front of the player all the times).

i'm not going to debate this forever... but perhaps you should study more how other games use v-models these days, and contrast that with a study of good 3rd-person gameplay. v-models are largely used as a means of cheating player perception, not displaying accurate mechanics. most v-models aren't even lined up with the shots fired, instead they are angled to show things off the best way possible. the v-model, animations, and everything else about the player view is tightly controlled and designed for maximum impact. very few FPS games, or even action games in general, bother to interrupt player control over gunplay in favor of realism or any other justification that you would think should stop gunplay. FPS games have capitalized on that by just ignoring such factors, because they can and nobody seems to notice enough to care. yet traditional FPS games still do interrupt gunplay for critical design reasons when necessary, its just never forced upon the developers... its a deliberate choice.

Third person games however, force you to acknowledge gunplay interruptions by the fact that its kinematically impossible to dive-roll diagonally backwards while firing a gun accurately forwards. try to make sense of that visually and see where that takes you. in the end you have to make a design decision... go for it? or just make it so "stunts" completely interrupt gunplay as a rule to avoid the technical headaches and extra animation content demands. want the freedom to change your mind about it? too bad, you're stuck. good luck telling your animators and coders to rework their pipeline without destroying previous progress. you and I know exactly how dangerous that can be when people are volunteering their services.

With a FPS game, you can pick and choose how much detail or work you want to put into the 3rd-person anims. you are never forced to go beyond the bare minimum set, its simply an option to do so. The way I see it, its less work by a long shot, and much less damaging if you change how you want the 3rd-person anims to work later on. the way I see it, you are risking the difference between a non-profit mod, and a full commercial game by the workload demands being debated here. while I would love for us to be successful at making a good 3rd-person action game... i have a hard time believing in it. perhaps someone could pitch an idea that would change how I feel about it, but so far the pie has only gotten higher.

tbh Vino, i'm very surprised you aren't running like hell from the prospect of doing this the less conventional way. Are you trying to start another 5 year project? or are you just playing devil's advocate for shits and dribbles?

[Shogi]

I believe the expression is "shits and giggles"

Anyway I see where both of you are coming from. First person would most likely be slightly easier, at least from an animation perspective. Vino is underestimating the third person animations. We'd have to spend a lot of time making them look realistic and working out a system to actually blend the animations (and the camera too). This is the kind of thing professional studios would go to a motion capture studio for. Bryitis brought up the point of mechanics in terms of shooting while stunting, and I'm guessing he's talking about 1 or 2 things: Guns pointing in the right direction smoothly between animations and actually having them make sense or how much harder aiming in 3rd person is (personally I was pretty shit at shooting people in 3rd). It's not impossible, but it is more difficult than a "6". My personal instinct/desire is to go for third partly out of ambition and a desire to set ourselves apart, and also because I feel 3rd would be more visually pleasing. It may end up being more difficult, but if we want this to actually be successful we can't shuck ambition for safety at every turn. Yes we have to set realistic expectations, but if we risk nothing we gain nothing.

[bryt]

and I'm guessing he's talking about 1 or 2 things:

i was referring to all of the above... the animation technicalities i could get behind if I was convinced it was worth it. but from a design perspective we may find that in version 1.5 or 2.0 we made the wrong call and thus need to change it. to be certain we need to understand the gameplay design first and foremost. and to my detriment, i know how Vino likes to work. nothing will be set in stone upfront and "iterative" will become our religion for a while, or at least until the "fun" has been established. from an artistic perspective this scares me a little bit because I would like to know rather than guess that my personal pool of blood/sweat/tears is actually being put to good use. but more than that I want to be confident that we will be able to find the "fun" mixture of gamplay without being over-encumbered by the bias and fear of regret that we invested too much into the wrong idea.

edit1: more than anything i'm just scarred of another 5year project. if i even get the slightest wiff that this is going that way i'll probably just lurk and do nothing. and a good 3rd-person game is not beyond our abilities, sure.... but it may be beyond our resolve.

edit2: my last suggestion would be to plan on a FPS gameplay model, but do so with 3rd-person as a possible expansion later. it will be easier to understand what we're up against for this feature after we have a working FPS style game. essentially the gameplay should be agnostic is this regard as I see it. If we find that our FPS doesnt fit well into a 3rd-person mold then it was probably a bad idea to start with. if it all works out then it will make a great feature addition to the game, and we'll have plenty of reference content to base our work off of. THIS is my vote for the safest and most ambitious approach simultaneously, with the added benefit of player choice.

[Antagonist]

I'm firmly convinced that an FPS would be more original and cool for the kind of game we're making. These days, every (action movie-like) action game that's churned out by triple A studios is a 3rd person. I think we can make it work and I loved seeing the way the first-person-view pictures Chris took turned out.

The way those pictures were received, seriously makes me feel like taking a bunch of pictures of stunts and suitable animations for stuff, like in the aesthetics thread. I'm really into this whole Double Action/FAG/Last Action Game thing now and even though I can't model, map or code for shit, I want to lend a hand.

I just don't want you guys to think I'm fruity because of it

[Vino]

Alright. I don't think we need third person animations so high quality that they take 3x longer than I indicated, and I don't see a problem with guns not always pointing exactly where the shots are going. I especially don't get what you mean when you suggest I'm trying to do this unconventionally, I don't see that at all. So I still disagree and I suppose that's where the discussion rests.