Jump to content

Photo

Some thoughts on gun mounts, hit boxes, and rolling


  • Please log in to reply
85 replies to this topic

#1
Drakona

Drakona
  • Advisory Board
  • 322 posts
  • Joined: 11-March, 2015
  • United States
x2
x2
So, I did some dogfighting in the latest build of DU and found it a really unsatisfying experience. I don't mean that I lost, exactly -- I did pretty well. Got a lot of kills. But I found the experience of getting the kills to be flaky and frustrating.

I've been thinking about why.

And the answer I've come up with is, it has to do with the relationship between hit areas and gun mounts.

Let's start by looking at what I'm used to from D1.

d1_damage.png

So, in D1, all targets are spheres. So I've drawn a circle here -- that's the target I'm shooting at. On the next line, I've drawn what my own shots look like in relative size. In the simple case, on the left, D2 gauss -- it's just a dot that shoots straight out of my reticle. Like every other FPS in existence. ;) Quad lasers are more interesting, four shots spread between 50% and 75% of the width of the sphere itself. And fusion -- fusion itself is huge. The shots themselves actually overlap.

So, what I've drawn on the next line is, how much damage do I do based on where my reticle is?

In the case of gauss, it's really simple -- did I hit the ship or not? If my reticle was inside the shield's outline, I do full damage. If not, I miss.

Lasers, now. Lasers get a little interesting. Depending on where I'm aimed, either one, two, three, or four bolts land. My reticle can be about a third of a shiplength off of the target, and I still graze it with a shot. Or it can be dead center and all four land.

What I like about this is that it makes lasers a weapon that really reward precision, and that behave differently in different circumstances. In a live dogfight against a target that's aware of me, it's hard to get any of the shots to hit at all, but if I get a good, accurate prediction in, I get a solid hit. Messier lead, less damage. Cool. And if I catch someone holding still? I can line up for a VERY deadly shot.

One thing to notice is that the "maximum damage zone" for lasers is pretty small, but it's practical to hit in some circumstances. I didn't carefully, carefully check that this drawing is to scale -- it's from memory -- but it's close enough to make the point.

Fusion has the same thing going on as lasers, only more so. It's a huge target, but to get full damage, I have to hit my target more accurately. It's okay for fusion to be huge because it's a one shot weapon; I have to hit my opponent both in space and time. But if I do it accurately, I can get a huge amount of damage out of it! And if I get close, partial damage.

That's a nice smooth slope, rewarding improving skill right there. I like that.

So what happens in a typhoon? Simple case, let's imagine a typhoon shooting at a D1 pyro.

pyro_damage_from_typhoon.png

So, I didn't measure this either exactly -- it's from memory -- but the thing I really remember about flying in a typhoon is that those gun mounts are really wide relative to the overall width of the ship. This drawing is kind of a guess at where they are.

So, the first thing that happens is my double damage zone takes on a much more complicated shape. It's no longer a vertical oval, right? It's kinda propeller shaped. That's a lot harder to keep track of and handle in combat. Which you see in the game, right? Getting two of your typhoon guns to hit is pretty fiddly, and kind of requires a target that isn't doing fancy dodging. If D1 quads are easy to miss with entirely in a dogfight, getting a shot inside of that propellar is going to be pretty tough to do on purpose. Hitting the target in space is tough enough without adding the angle into it.

But look what the wide gun mounts do to the triple damage zone! It's teeny! Hitting that is going to involve way more luck than skill, unless someone sits still and lets you line up on them.

But that's theoretical. Typhoons don't have anything spherical to shoot at.

What happens when a typhoon shoots at a typhoon?

typhoon_from_typhoon.png

Well . . . crap. There's the problem.

Where I have to have my reticle, just to hit the thing has virtually no relation to its hull. Between the wide gun mounts and the skinny target, there's no way I'm going to be able to hit that in flight. Forget triple damage. That requires him to be holding still AND me to be holding still and it's so precise it takes multiple tries. But even hitting the target at ALL! That's such a complicated target area, it feels pretty hopeless.

And what's extra interesting here -- it's super sensitive to how I'm rolled relative to my target. If I roll the other ship just 30 degrees or so -- all of my aim spots change radically.

It's worth mentioning that a ship might roll that far while my shots are in the air.

It might also be worth mentioning that a typhoon like to roll really fast.

That's why the dogfights were so unsatisfying. I'd put a lot of shots in the vincinity of my opponents, but whether they hit or not had nothing to do with how well I'd aimed them, and everything to do with how lucky I got. Spray and pray. Bleah.

But this isn't totally accurate and up to date, right? I'm not sure exactly what they did last update, but they made the hit boxes bigger somehow. And the above drawing is kind of a from-memory guess at what the model looks like.

Let's put some shields on it.

typhoon_shields.png

Now I'm really making stuff up. ;)

But if I were to put shields on the ship that were thick enough to preserve its shape as a target, this is about how big I'd make them. What's that like to shoot at with laser mounts that are near the edge of the wings? It's like this.

The triple damage is looking kind of plausible on a still target, but single damage in a dogfight is still a total crap shoot. And it's still extremely sensitive to my roll relative to the target's. But at least if I shoot near the center of the ship, I'll probably hit something. Maybe. It's still a crap shoot all told though.

What if I bring the gun mounts in? Here's the same target (well, I've drawn it as bigger shots rather than shields on the typhoon -- mathematically they're the same thing). But the main thing I've done is move the gun mounts much closer to the center of the ship. They're now 50% of the wing length rather than most of it.

typhoon_plasma.png

That's . . . much better. The target area is now contiguous, so if I aim at where the ship is in general, I'll get some damage. The double and triple damage are still funny shapes, though, and still extremely sensitive to roll.

I'm not crazy about that. Speaking as an experienced dogfighter, getting damage on a target is pretty dang difficult when that target is a sphere. Doing it with a time constraint (fusion) is crazy difficult. 3D space is just . . . humans aren't good at it. Adding in a roll constraint, too? That smells to me like "too difficult to be fun" and "hard enough to get right that I'll just take my chances". Spray and pray again.

But still. The lesson here is that bringing the gun mounts in closer to the center of the ship gives us a dramatically more sensible target area.

So let's step back a second. How do gun mount widths interact with target shape and rolls in general? Here's a study.

Shooting at a sphere with two gun mounts spaced at 50%, 75%, and 100% width. (I'm treating these shots as having no width of their own to eliminate that variable).

laser_study_100.png

The thing to notice here is how dramatically the increased width reduces the double damage area. At 50% width, I *might* be able to hit that in flight if the other pilot isn't very good or isn't dodging me specifically. At 75%, I'd need a still target and some time. At 100%, you'd have to be a computer to do it.

Of these, I really like the 50% the best. I think the fights are most satisfying if the double damage areas are big enough to be skill shots. Nobody likes spray and pray, and having a tiny double damage area is like having a tiny vulnerable point on your ship that sometimes just happens to get hit. Lame.

What if we start varying the shape of the ship? Let's squish that sphere to 75% of its height, and see how sensitive it is to rolls.

laser_study_75.png

What's interesting to notice here is that the 50% case already sees a dramatic difference in the size of the double damage area, just from a sphere that's 75% height on one dimension. How you are rolled relative to your target is literally going to change your damage output by -- I'm eyeballing here -- like a third or a half. But it's continuous and easy to understand and forgiving -- if you get the roll slightly off, the target area doesn't change too much.

This would make me nervous. I'd have to fly with it. But if you want to start introducing roll sensitivities to dogfights, this looks to me like a reasonable place to start. This might be fun. :)

On the 75% laser width line, we've lost all contact with reasonableness. The lasers can straddle the ship along one dimension, the target areas are significantly roll sensitive, and the double damage area is way too small to hit on purpose.

Not even gonna talk about the 100% width line. Don't go there.

One more study -- how about a sphere that's 50% tall in one dimension?

laser_study_50.png

Eeeeew. All of it. That does not even look slightly fun.

------------------

So what's the upshot of all of this analysis?

Well, some recommendations.

I would keep the frontal cross sectional area of the ships as simple as you can in shape. Even just those ovals got out of hand when they got too flat. You could maybe do a rounded triangle. I wouldn't do more.

I would encourage keeping the frontal shapes similar widths as the ship rolls, to keep the roll sensitivity of aiming under control. Like, even if you want to do a propellar typhoon shaped ship? Give it a big center section so you can at least shoot for that.

Bring the gun mounts in so they're comfortably inside the smallest shape you might ever be shooting at with them. 75% as wide. No wider. Like, if a wasp is going to stay super flat? The widest gun mounts in the game should be 75% of that distance apart. Otherwise that ship is going to be spray and pray to dogfight with.

As a final thought, if you're going to ignore all of that and stick with the complex target shapes . . . consider giving me a more helpful reticle. At least tell me what direction and distance my guns are, so I can try to line up!

----------------

But Drak! Other FPSes use complex models and don't have this problem? What are we doing differently?

Well . . . we have two things other FPSes don't: multiple gun mounts and the ability to roll.

In GS:GO, your target area looks like this:

csgo_damage.png

It looks like that every time. You can't roll. You don't have gun mounts to align. Click the dark blue areas for extra damage, that's it.

In Descent, that target would look like this:

csgo_in_descent.png

. . . ;)

Which just goes to show ya. When you copy ideas from groundpounders into a 6DoF, weird stuff happens!

But yeah. In D1, at least? The combination of the stable, simple target profiles and good separation on the gun mounts -- not too little, not too much! -- make a center mass shot into what a headshot is in other games. Weird, I know. But it works in 3D. You get a reasonable, skill-accessible target profile from any direction. :)
  • 14

#2
LotharBot

LotharBot
  • Advisory Board
  • 1,407 posts
  • Joined: 11-March, 2015
  • United States

When I was proofreading this, it reminded me a lot of the chaotic systems I studied in grad school -- Julia sets and oddly shaped basins of attraction.

 

With the relatively simple geometry and moderately spaced gun mounts, "aim for the center" is a very strong strategy.  Complex (especially non-convex) geometry plus gun points wider than the narrowest target area make aiming for the center into much more of a crapshoot, and they create an odd dynamic wherein your best target points are off center and at odd angles.  Which definitely makes the game feel more random and more sensitive to "lucky shots".

 

I think this post made me smarter.


  • 0
Come play original Descent! Full version setup instructions or Shareware unzip-and-play
Voice chat on Descent Rangers Mumble server address mumble.descentrangers.com port 31393

#3
Demosthenes

Demosthenes
  • Backer
  • 39 posts
  • Joined: 04-November, 2014

So after reading... the upshot is to go for giant hitboxes?

 

I mean it seems like a lot of thought and effort was put into the composition of the topic especially with all of the diagrams and such, but Wingman has hinted at implementing weapon convergence so this seems really premature with regards to the guns, and Lothar already made a thread about making the hitboxes really big. We're in super early alpha and they still don't have all of the basic things implemented yet, like weapon convergence. Your graphs are going to go out the window when that gets implemented because they assume a constant trajectory relative to your other shots. With weapon convergence is implemented we'll end up with X- shaped paths for the weapons on 2-bangers and some shape I can't even wrap my head around with the Typhoon. Once that's in your whole complaint about shots not going where you've aimed will be moot.

 

I also disagree with the idea that weapons should be spaced such that roll is immaterial, if I'm in a wasp and being shot at by someone it's not fair to remove from me the tactic of reducing my profile to my enemy to favor your preferred tactic. My tactic is as old as naval combat.


  • 1

#4
LotharBot

LotharBot
  • Advisory Board
  • 1,407 posts
  • Joined: 11-March, 2015
  • United States

So after reading... the upshot is to go for giant hitboxes?

Nope.  This is entirely new analysis; don't treat it as a mere rehash of what I said earlier.

 

The point is that hitboxes that are convex and well-proportioned make for the most consistent experience.  The more complex or distorted they are, the more "randomness" you get -- the more the game feels like spray-and-pray because of the discontinuous, chaotic nature of the aim-here-for-strongest-hit regions.  There's certainly room for hitboxes that are different based on ship orientation.  But "keep it under control".

 

The "narrow gun mount" image (5th from top, under "50% of the wing length") gives a similar effect to convergence, yet still shows odd sensitivity to shot placement and relative roll.  This type of sensitivity creates an effect that is essentially random given that you're shooting a moving target with slowish weapons through internet lag.  


  • 0
Come play original Descent! Full version setup instructions or Shareware unzip-and-play
Voice chat on Descent Rangers Mumble server address mumble.descentrangers.com port 31393

#5
sdfgeoff

sdfgeoff
  • Backer
  • 54 posts
  • Joined: 17-March, 2015

We can also analyse aiming challenge in terms of the number of dimensions of complexity to land a shot. Assuming that you can place a shot exactly where you want to:

  • Hit-scan FPS's are 1-dimensional: you have to your enemies angle from you (another dimension, polar co-ordinates)
  • Non-hit-scan FPS's are 4-dimensional: weapon travel time, enemies angle from you, enemies distance, enemies velocity relative to you
  • Traditional Descent is 7-dimensional: weapon travel time, 2 lots of enemy angle, a distance (spherical co-ordinates relative to you), plus 3 lots of enemy velocity
  • Introduce shaped hit-boxes ... 13D: weapon travel-time, 2 enemy angle, a distance, 3 enemy rotations, 3 enemy linear velocities, 3 enemy angular velocities
  • Introduce convergence ... 14D. (although it technically is a factor of distance or time, it's another level of complexity, which is what we are looking at here)

 

To prove the validity of this assumption that ease of hitting someone is dependant on the number of dimensions:

  • Using a hitscan weapon is easier than a time-of-flight one. It eliminates the time componenet as well as all velocity vectors. In D3, the mass-driver is a 2D weapon: two angles. All other weapons are 7D. That's why it's so much easier to hit with. Same with Vulcan and Gauss. They're simpler weapons to use.
  • Hitting a stationary target is elementary. It strips away a whole heap of dimensions to do with motion.

 

And this is just for aiming, not taking into acount your own positioning and dodging. People have enough trouble jumping from 4D aiming to 7D aiming. Here we are sitting at 13D contemplating making it easier by adding a 14th.

 

By this analysis: we have a system where aiming is nearly twice as complex as Descent was, and hitting in Descent was hard enough

 

In my (limited) game development experience, the games that people enjoy the most are the simple ones conceptually. You have to be able to get your head around the game in a matter of hours. Even hard-core games have to be simple enough to explain to people. You can quite happily talk about leading shot in a linear sense, but I have never heard anyone ever talk about leading shots in an angular sense.

 

 

 

 

 

Before I leave, I just think I'd quickly point out the limitations of my analysis. Aiming is evern more complex then I've described here. Each velocity componenet also has an additional acceleration component, which brings in a couple of other time variables as well. Then there's your own ship aiming-rotational-velocity limitations, which I can't think how to dimensionalize.

 

Next time you're playing descent, and someone asks what you are doing, reply something like:

"Playing a 74 dimension game. I have to compute weapon travel times, ship locations, ship velocities, the trajectories of all incoming projectiles, the predicted course of homing missiles, balance the shield and energy on my ship, and somehow not hit into walls either."

Yeah, if you can play Descent, you're boss.


  • 2

#6
Void Stalker

Void Stalker
  • ModeratorMod
  • 1,700 posts
  • Joined: 05-November, 2014
  • United States
x2

I had to upvote that post for the great analysis there, it's just spectacularly in-depth.  The weapon / mount / reticle / hitbox diagrams were spot on with my own observations while playing. It is a big reason I have stopped using the Typhoon most of the time, since my shots usually just go right around the other drones, with the exception of the Panzer. I tend to use the Panzer because of it's shield, or the Wasp because it's guns are closer.

 

It's obviously  a big issue, because of the buzz about it, so here are a few notes from things posted in previous threads and discussions that might contribute;

 

1.  A drone with a single center-mounted weapon might be a great test-bed item in proving grounds to further refine the combat experience.  It might even make a great drone to include in the game itself.

 

2. The inclusion of precise aim-points on the reticles, like those in D1, made for each drone and it's gun layout, will help with shot placement. 

 

3. Convergence is a possible solution that appears to be in the works.  I believe they are looking at even making it adjustable too, so you can tailor your "cone of fire" more easily. (I do like this approach, as I think it keeps the ships unique with their firing patterns while allowing customized play to style.)

 

4. Moving the gun mounts closer to center is a fairly elegant solution, however, does that mean a full rework and remodel of all of the drones.  That could be a big problem and be too late in the pipe to do that. 

 

5. The "big bubble" shield is another idea floating out there that makes a lot of sense.  Some say that the hitbox then just becomes a D1 sphere all over again.  Others suggest making the shield one big hitbox, but the hull a smaller one, to increase the skill cap.


  • 0

-Tracy,  Founder, The Tunnel Rats

"Just remember this - it is amazing what can be accomplished when no one cares who gets the credit." -Wingman

Note: Everything I post is as a fan of Descent: Underground except for official statements - which will be written in Moderator Blue.


#7
Wingman

Wingman
  • Game DeveloperDev
  • 1,929 posts
  • Joined: 04-November, 2014
  • United States


This is exactly why we are looking at adjustable convergence.

 

Let the user tailor their firing to the distance they like, and have the reticle be a lot more useful.

 

Nice write up.

 

 

WM


  • 1

#8
Whiteshark

Whiteshark
  • Backer
  • 527 posts
  • Joined: 01-April, 2015
  • Finland
x2

Easy fix: introduce emd cannon.


  • 0

Wouldn't it be easier to read if this was on Comic Sans?

 


#9
Wingman

Wingman
  • Game DeveloperDev
  • 1,929 posts
  • Joined: 04-November, 2014
  • United States

Easy fix: introduce emd cannon.

 

I want an Ack Ack cannon that is picked up and has limited shots.....ROB...I WANT AN ACK ACK CANNON !!!!

;)

WM


  • 0

#10
Rooster61

Rooster61
  • Backer
  • 156 posts
  • Joined: 07-June, 2015
  • United States

Drakona's post is well thought out, and really details why combat differs greatly in D:U than from classic Descent, but I do think it leaves out one important wrinkle that is unique to D:U.

 

Because the hitboxes were spherical in the classic Descent, it did not really matter what attitude a ship took relative to your own. You would hit them the same from the front, the top, the side, the bottom, the back, and any combination in-between. As she so clearly pointed out, having weapon's spaced so that they overlap in certain zones made for very precise, very predictable combat.

 

In D:U, the opposite is true. The attitude of your target matters. Most of her analysis deals with a target that is facing you head on, and in this situation, yeah it is very hard to hit with traditional Descent weapons. However, what is left out is when a ship is NOT in a head on attitude.

 

Because classic Descent hitboxes can be hit from any angle, there is less of an incentive to strafe around your target. Sure, it is usually ideal not to be in front of your target's LOS because that means you are likely to be hit by their weapons, but in some cases, there is a trade-off since strafing around your target so that you are hitting them from a non frontal angle may not yield as good of a chance for YOU to hit them (due to walls, exposing yourself to another pilot in the room, hazards like lava, ect...). This is a good thing, because it makes you have to decide if it is more advantageous to take the sure shot head on, or try to strafe around and avoid fire at the cost of probability that you will hit the opponent.

 

In D:U, the incentive to hit from the broadside is increased drastically. In the same situation mentioned above, a D:U pilot is going to be more inclined to go ahead and take the side angle because there is so much more exposed hull to hit than from a frontal or rear angle. Compound this with the fact that you don't want to get hit by your target's weapons, and you end up with dogfights where strafing to an odd angle is the most beneficial tactic.

 

This also gives extra depth to D:U in that some ships don't have this cross-section advantage. It lends more individuality to each drone, and I think reducing the advantage some drones have over others to dodge will take away from what can potentially set D:U apart from classic Descent.

 

Now, as a classic Descent pilot myself, does it feel natural to prioritize attacking from odd angles in this way? No, it feels very strange, as I am used to facing off with someone head on and only going from a different angle if it actually makes my shot more likely to connect while reducing oncoming fire. It's frustrating to find myself tailing someone in D:U, as in classic Descent, it was usually a death sentence for your target, but here I find that often my target can skirt away due to the tiny cross-section that is exposed if they don't turn and expose their hull.

 

All this said, this game is not classic Descent, and I'm not so sure that this new dynamic is a bad thing. Realistic hitboxes are making me rethink how I need to fly to take someone down, and while Drakona is correct that there is more of a spray and pray tilt to head on D:U dogfights, I feel like the trade-off is that it requires more skill to get around the side of the opponent in D:U than in classic Descent.


  • 5

#11
Steel01

Steel01
  • Backer
  • 41 posts
  • Joined: 19-March, 2015
  • United States
Has anyone suggested having spherical shields in classic mode? Or as a server configurable option in multiplayer?
  • 0

#12
DeserveVictory

DeserveVictory
  • Underground
  • 93 posts
  • Joined: 14-April, 2015
  • Canada
x12

Is the ship hurtbox in D1 truly perfectly spherical? I would've sworn that I can rotate 90 degrees to my opponents orientation and reduce their chance to hit me with both shots. Weird.


  • 0

"Sometimes I wonder. I wonder where it began. Plants burst forth from soil and animals from their parents. Rocks are born from stars. But what about ideas? Where do ideas come from? I wonder about how opinions are seen as more valuable than knowledge. I wonder about the beginning and the end."


#13
PyroJockey

PyroJockey
  • Backer
  • 927 posts
  • Joined: 10-March, 2015
  • United States
x2
Is it possible to have a spherical hit box for shields and have a second hit box for hull damage that conforms to the fuselage of the ship?

This is on the pretense that the shield is a spherical force field projecting from the center of the ship. Once it has been eliminated all that remains is the hull armor which presents a smaller but less resilient target.

I don't know if it's possible in UE4 but thought I should put it out there if it isn't already.
  • 2

There are times you strive for greatness, to reach lofty goals and accomplish

noble deeds. Then there’s the rest of the times you just want to blow stuff up.
Posted Image

[font=arial]Note: Everything I post is as a fan of Descent: Underground except for off-color statements - which will be written in PyroJockey Plum[/font][font=arial].[/font]

#14
deftflux

deftflux
  • Backer
  • 147 posts
  • Joined: 11-May, 2015
  • United States

I don't think Drakona was saying that we should make every drone a sphere again.  Rather, the hitbox shapes could be simplified for a more predictable experience that doesn't seem as random (spray and pray).  Even an oval changes the game drastically, as shown in the diagrams.  Even if the Wasp were an oval, it would still have the advantage of taking less damage at different bank angles, but it would be more often one laser hitting instead of two, rather than them both missing completely.

 

I'm not a fan of the weapon convergence idea.  I agree with the point that it makes the game even more complicated when it's already too complicated.  Moving gun mounts closer makes more sense.  That way, the shots are closer together through the whole travel of the shot, not just at certain depths.  We want to reduce the cost of entry into this game, not make it higher.

 

I have been struggling with keeping my interest in D:U.  To be honest, I haven't even played the latest build yet.  It doesn't seem like anything I do has much of an effect on the outcome.  If I dodge randomly, it turns out just as well as if I really focus on dodging smartly.  If I aim randomly, I seem to get just as many hits as if I really try.  So I'm not motivated to try, and there's no sense of urgency when I fight.  It's too random.  Part of this is exacerbated by the fact that I can't tell when I'm successful or not.  The "bump" noise when you hit someone blends in to everything else.  The instant "poof" of when they explode is very unsatisfying.  I still don't notice the indicator of when I'm taking damage.  I have to watch my shield indicator, which takes my eyes off of aiming or dodging.  So I play a few matches, get bored, and stop.

 

I really think simplification around hitboxes and getting hits is sorely needed.  No, we don't necessarily want to go back to the D1 sphere, but I don't think we should swing to the other extreme either.  It isn't working.


  • 1

#15
defcon_x

defcon_x
  • Underground
  • 1,450 posts
  • Joined: 20-March, 2015
  • United States
x12
x4

Is it possible to have a spherical hit box for shields and have a second hit box for hull damage that conforms to the fuselage of the ship?

This is on the pretense that the shield is a spherical force field projecting from the center of the ship. Once it has been eliminated all that remains is the hull armor which presents a smaller but less resilient target.

I don't know if it's possible in UE4 but thought I should put it out there if it isn't already.

 

I think this is a brilliant and immersive idea. Just have the hull HP get a far lower maximum than the shield HP (presumably about 200 like D1-D3?) so that fights are not dragged out too long in the 'no shield' zone.

 

It also makes sense to bring basic laser firing just a liiiitle closer to the center of the ships, and have "quads" be what stretches the range away from the center sight, to catch dodges.


  • 1

LINUX OS: MINT 17.X • DESKTOP: CINNAMON 2.X • CPU: i3 • RAM: 8GB • GPU: GTX960 • Thrustmaster T16000M • HELP DE-BUG D:U!


#16
DeserveVictory

DeserveVictory
  • Underground
  • 93 posts
  • Joined: 14-April, 2015
  • Canada
x12

I don't think Drakona was saying that we should make every drone a sphere again.  Rather, the hitbox shapes could be simplified for a more predictable experience that doesn't seem as random (spray and pray).  Even an oval changes the game drastically, as shown in the diagrams.  Even if the Wasp were an oval, it would still have the advantage of taking less damage at different bank angles, but it would be more often one laser hitting instead of two, rather than them both missing completely.

 

I'm not a fan of the weapon convergence idea.  I agree with the point that it makes the game even more complicated when it's already too complicated.  Moving gun mounts closer makes more sense.  That way, the shots are closer together through the whole travel of the shot, not just at certain depths.  We want to reduce the cost of entry into this game, not make it higher.

 

I have been struggling with keeping my interest in D:U.  To be honest, I haven't even played the latest build yet.  It doesn't seem like anything I do has much of an effect on the outcome.  If I dodge randomly, it turns out just as well as if I really focus on dodging smartly.  If I aim randomly, I seem to get just as many hits as if I really try.  So I'm not motivated to try, and there's no sense of urgency when I fight.  It's too random.  Part of this is exacerbated by the fact that I can't tell when I'm successful or not.  The "bump" noise when you hit someone blends in to everything else.  The instant "poof" of when they explode is very unsatisfying.  I still don't notice the indicator of when I'm taking damage.  I have to watch my shield indicator, which takes my eyes off of aiming or dodging.  So I play a few matches, get bored, and stop.

 

I really think simplification around hitboxes and getting hits is sorely needed.  No, we don't necessarily want to go back to the D1 sphere, but I don't think we should swing to the other extreme either.  It isn't working.

 

I wonder then what you would think if you were to try D1 for a bit and compare it to D:U.


  • 0

"Sometimes I wonder. I wonder where it began. Plants burst forth from soil and animals from their parents. Rocks are born from stars. But what about ideas? Where do ideas come from? I wonder about how opinions are seen as more valuable than knowledge. I wonder about the beginning and the end."


#17
Drakona

Drakona
  • Advisory Board
  • 322 posts
  • Joined: 11-March, 2015
  • United States
x2
x2
A couple quick responses --

On convergence:

It's hard to know about anything without playing with it, but I really don't expect convergence to help here. Sure, there will be some range at which your shots get close enough that you get the nice behavior from D1. But there will be other ranges at which you get the crappy (is there a name for these blue things? Imma call them) damage profiles, too. Tight gun mounts and simple targets would guarantee they're all nice. Convergence and complex targets will result in damage profiles that vary with range, and which are all crappy in their own special way.

One of the things to realize is that in a dogfight, you're engaging at a variety of ranges, between half a cube and 6 or 7 cubes. Sometimes longer, but that's what's typical. Which is to say, the ranges at which you're engaging vary by a factor of 10 or more. Convergence is linear, and I've suggested the gun mounts be half as far apart as they are. To get that to happen at the range in which dogfights run, you're going to have to have the weapons cross over around 10 cubes or so. And it's still going to mean crappy damage profiles for anything that isn't a 4-7 cube distance away. Tighter convergence, further profiles get crappy. The whole thing's linear. It just isn't mathematically possible to have a convergence value that will make the damage profiles nice at the full variety of ranges at which dogfights happen without moving the gun mounts.

No, what I would expect convergence to do is impoverish dogfights by limiting the range options, which is going to make it even harder to hit people. And less consistent.

Yeah, we don't know until we play with it. But I don't think that's the answer. Instead of introducting another variable to a situation that's already too complex, I would look to simplify.


On frontal profiles:

So, the thing about the side profiles is, you don't need complex targets to encourage this. Far from it. In D1 you have a massive incentive to get around to someone's side. If they can't see you, they don't know which way to dodge, and become dead very fast. They also can't shoot back. ;) These are both huge, and even new pilots learn that pretty fast. Which is why a target that is aware of you is always more or less looking at you.

In fact, getting an angle on someone is such a win tactic in D1, that every dogfight is more or less a struggle over trying to do just that. There are people who even specialize in it. And it isn't easy. They don't hold still to just let you fly around them, you know. Getting a shot like that involves a lot more than just flight -- it involves trickery and shot placement, too.

Even very bad pilots want to keep sight of you. And it isn't really very hard. Getting around to someone's side in a dogfight isn't the normal way to get damage. It's burning them and embarrassing them. You can do it, but it's not the typical case.

Here's a live fire demonstration. This is a dogfight, me vs. Mark392. You don't have to watch the whole thing, but skip around to a few places? Notice -- how often is he presenting me a frontal profile? How often is he on my screen (=I am presenting him a frontal profile)? Pretty much the whole freakin' time. I get occasional looks at angles further off center than 30 or 45 degrees, but they're brief and rare. If they last long enough for me to take a shot, it's generally a kill.

https://www.youtube....6bglPeCcrw?t=77

This is also a good video to watch with images of those blue profiles dancing in your head. See how hard it is for me to hit the guy? Those are 10m spherical bubbles I'm shooting at. Imagine trying to hit something more complicated on purpose against a pilot who resists to this degree.

But yeah. Point is, I'm spending every minute of that video trying to get around to his side. And I'm trying really hard, you know? I'm getting in close and doing a full trichord and then reversing direction and trying to get him to overreact, all just to get a split second look at the top of his ship. He's doing the same to me. We do accomplish it for brief periods from time to time, but it's uncommon and certainly isn't where the bulk of the damage comes from. Most of the hits are prediction and pushing against a fully aware target presenting a frontal profile.


On shield bubbles:

I think this would be a really good solution.

The funny thing is, the Descent community has been over this ground before. D3 offered us the option to have accurate weapon/ship collisions or to use shields. I didn't play D3 enough to really care about it, but I do know that the community more or less unanimously agreed that accurate collisions sucked. I didn't know why. But now I think I do.

That's not to say shield bubbles have to be spheres, (though I do feel compelled to point out that this whole thing is smack in the center of "make the ship fly right"). But I would work up to complexity, not down from it. Try ellipsoids that differ from each other in size by, let's say, 75% on all axes or (maybe?) 50% on one. That might be too complicated to be fun, but it might be cool. I wouldn't try anything more complex as a starting point.

I do think it's fair to at least let the D1 solution compete. Do what D3 did. Let us play with shield bubbles or without, and see what's more fun.

But yeah. I know some of you don't like what sounds like a simplistic, low tech solution. But the thing is? Making things more complicated and difficult doesn't always make them more fun. You want games to be difficult, but you want them to be accessibly difficult. Learnable. Skill-rewarding. Interesting. And the thing about 6DoF is that it's already way to difficult for most people to find fun. As FPSes go, it's a very hardcore genre.

You could make it harder in a lot of ways, right? You could force people to solve math problems while flying. ;) As a less silly example, you could remove friction from the game. I've done some frictionless 6DoF flying recently. Well, 4 or 5DoF. Quadcopters. It's intense. Maintaining control over your vehicle is the most you can hope to accomplish for quite a long time.

I wouldn't recommend it as a combat game, though. It's more difficult, just like complex profiles are harder to hit. That doesn't make it a better game.


An addendum:

Lotharbot had a great comment on this last night that he didn't have time to write up. I'm writing it up for him now because he's busy today.

So, you can compensate for lag in a game, but what you can't compensate for is the connection variability packet to packet. Right? In my games that's usually about 5-10 ms. On top of that, if you'r e playing the game . . . I dunno, let's say you're vsynced at 60 FPS . . . there is a delay of up about 15 ms between what you see and what's happening.

Conclusion: Anything that requires you to make a shot in a window of time more precise than 25 ms is luck.

So meaningful regions of the damage profile should be bigger than that. Probably significantly bigger.

For reference, the most dimensionally sensitive gun in D1 (hit in 3 dimensions of space PLUS be accurate in time) -- fusion -- has a double damage hit area that's a little narrower than a pyro. Even aiming at a target doing a full trichord perpendicular to you, that's a window of about 100 ms. Allowing for a graze and looking at a single chording target? You're looking at more like a 250 ms window.

I would like to suggest that the more dimensionally complex you make the targetting, the bigger these windows need to be to compensate for it. So if you do want aiming to depend on angle and range and the phase of the moon . . . you better slow the ships way down and make them physically thicker and larger than they are.
  • 2

#18
Void Stalker

Void Stalker
  • ModeratorMod
  • 1,700 posts
  • Joined: 05-November, 2014
  • United States
x2

How do you design a chess game that also appeals to people that play checkers and tic-tac-toe?


  • 0

-Tracy,  Founder, The Tunnel Rats

"Just remember this - it is amazing what can be accomplished when no one cares who gets the credit." -Wingman

Note: Everything I post is as a fan of Descent: Underground except for official statements - which will be written in Moderator Blue.


#19
Star Machinist

Star Machinist
  • Backer
  • 157 posts
  • Joined: 23-March, 2015
  • United States

Great analysis and discussion. I've also noticed how difficult it can be to land shots in a Typhoon if I'm flying at an angle to my target. In particular if a narrower ship like the Wasp gets close enough, it feels as though none of my shots actually connect. I get that the Typhoon is more of a "suppressing fire" ship, hence the spinning wings and wide spread of fire. I acknowledge that setting convergence could potentially alleviate that issue.

 

However, I'm uncertain about the idea of fiddling with weapon convergence in the middle of a dogfight. I mean, combat in Descent is frenetic. Unless weapon convergence is automatic, I feel like it'd be a bit distracting to attempt to adjust convergence on split-second timing. What would likely happen is that a pilot sets convergence for a particular range and then attempts to engage all targets at that range.

 

I think I'm more in favor of increasing hitbox size, or perhaps even the effective impact area of the weapons themselves. Perhaps the best solution lies somewhere in the middle, though. Perhaps some basic form of weapon convergence combined with larger hitboxes/impact areas?


  • 0

See You Underground

Also unfortunately called: 

- Star Masochist (thanks Wingman)

- Star Misogynist (thanks Plaidhatter)

- Star Machicolation (thanks dunkelza)


#20
Vorxion

Vorxion
  • Backer
  • 438 posts
  • Joined: 01-April, 2015
  • United States
x2

For all that in-depth analysis, my answer to the solution is simple:  evolve your gameplay so that you're not coming at your targets head-on or directly from behind.  Flying a Typhoon?  Come at them from above/below to offer yourself the biggest cross-section possible.  You have to do the same sort of thing in World of Warships, and that's nowhere near as maneuverable.

 

Isn't this considered emergent gameplay...that thing which is so desirable in other circumstances?


  • 0
KS Backer #2314 - Original D1/D2 player. Own D1, D2, D3 from GOG.com.
Windows 7 Pro 64-bit, EVGA GTX-980 Ti Classified, i7 4960X @ 3.6GHz, 64GB RAM (1866 clocked to 1600 for stability), GeForce 364.51




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users