KickStarter Phoenix Point - the new game from X-COM creator Julian Gollop

[Grotesque]

Current version of the game does not have accuracy stat but finished game should have it.

Then it is weird that not even once was mentioned that the circle area is dependant on the accuracy rating of the soldier. I would've thought that if you aim to make a video on the differences of the aiming mechanics compared to xcom, that would be on the list.



And now for some math time and how "honest" the game pretends to be.

For the shots to have a 50% probabilistic chance to "land" in the yellow circle, the yellow circle area should have 50% of the bigger blue circle.
After some measurements and basic calculations, on the right is how big the yellow circle should really be.

If they can't get even this easy stuff right, what confidence should I have they get the complex systems right behind true RNG or that what they advertise is indeed true at the math level?

 

[ArchAngel]

Current version of the game does not have accuracy stat but finished game should have it.

Then it is weird that not even once was mentioned that the circle area is dependant on the accuracy rating of the soldier. I would've thought that if you aim to make a video on the differences of the aiming mechanics compared to xcom, that would be on the list.



And now for some math time and how "honest" the game pretends to be.

For the shots to have a 50% probabilistic chance to "land" in the yellow circle, the yellow circle area should have 50% of the bigger blue circle.
After some measurements and basic calculations, on the right is how big the yellow circle should really be.

If they can't get even this easy stuff right, what confidence should I have they get the complex systems right behind true RNG or that what they advertise is indeed true at the math level?

It does not have to work like that. That depends on weapon. Maybe the weapon has a real problem shooting straight, in that case 50% will not be perfect 50% but more wide. Or it is very precise weapon and then 50% will be smaller circle.

 

[passerby]

If they can't get even this easy stuff right, what confidence should I have they get the complex systems right behind true RNG or that what they advertise is indeed true at the math level?

Circles represent intersection of bullet spread cones and the screen surface.

They use an unrealistic model where angles at which bullets are shot are spread evenly within the cone, hence diameter for 100% is exactly 2x bigger then for 50%.
Truly realistic model would have a nonlinear distribution taken from empiric data ( Gaussian ? ), where most shots are concentarted closer to the center but there are some outliers around ( random muscle shakes, distractions, bullets with out of tollerance parameters, etc. ).
So realistic model would have even smaller 50% circle not bigger. Then each subsequent shot in the burst mode should have the center of the probability cone randomly kicked around, but the devs don't aim at being a military sim anyway.

What you posted on the other hand is just bullshit with no basis in physics, or math, because diameters are proportional to bullet spread cone angles, not surface areas.
Your picture just represents a failure at understanding and abstracting a high school level physics problem into mathemathics.

This model would shine the most in a realistic 3d environment, with free movement and unit stances.
In super abstract environment alligned to a cube grid, with chest high cover boxes, it's a bit of an overkill, but it's an easy and accurate way of implementing aiming at and hitting individual body parts I guess and they wanted this feature to stand out from other xcom clones.

Random note: The way they first play shooting animations and bullet hits and separately getting hit and damage numbers later is so awkward. They should calculate everything under the hood first and then display it simultaneously.

It does not have to work like that. That depends on weapon. Maybe the weapon has a real problem shooting straight, in that case 50% will not be perfect 50% but more wide. Or it is very precise weapon and then 50% will be smaller circle.

Rather single shots should be more concentrated, while burst and auto modes should be less concentrated to abstract gun being kicked around. Gun accuracy should just scale the circles.

My guess is they will just have an even distribution for all weapons and fire modes, while weapon and eventually unit stats will just scale the circle.
It's a casual game, with no ambition of being a military sim and the only reason they have this at all, is because they wanted individual body parts hitting to stand out from other nuXCOM clones.
No matter the motivation and whether it really adds much in this particullar game, it's still more accurate then any other XCOM clone.

 

[luinthoron]

Hey, you started it. I was just questioning stuff in the video, I didn't know it would upset at least two of you that much.

You took the video's description of XCOM's hit calculation and somehow decided it applies to PP instead. If that doesn't deserve the 'retadred' rating, I don't know what does...

 

[Grotesque]

bullets are shot are spread evenly within the cone

what part of RANDOM you don't fuckin understand?
they stated numerous times that shots are generated RANDOMLY within the cone of fire .
The blue circle is the base of the cone and its area is (should be) dependent on the distance to the target.



Now how they generate the random trajectories within weapon's cone they NEVER gave details about it.

hence diameter for 100% is exactly 2x bigger then for 50%.

I don't even...

because surface area of the circles doesn't represent angles of weapon shots spread.

Circle area (the base of the cone) represents the area where the bullet trajectory intersects the plane at a certain distance that contains the base of the cone. The position of the intersection point is RANDOMLY chosen within said area.

doesn't represent angles of weapon shots spread

What does angles have to do with anything? Angles in respect to what? Do you even know what an angle is?

 

[Alienman]

Hey, you started it. I was just questioning stuff in the video, I didn't know it would upset at least two of you that much.

You took the video's description of XCOM's hit calculation and somehow decided it applies to PP instead. If that doesn't deserve the 'retadred' rating, I don't know what does...

I didn't though, retard.

 

[passerby]

Spoiler

bullets are shot are spread evenly within the cone

what part of RANDOM you don't fuckin understand?
they stated numerous times that shots are generated RANDOMLY within the cone of fire.
The blue circle is the base of the cone and its area is (should be) dependent on the distance to the target.

Now how they generate the random trajectories within weapon's cone they NEVER gave details about it.

hence diameter for 100% is exactly 2x bigger then for 50%.

I don't even...

because surface area of the circles doesn't represent angles of weapon shots spread.

Circle area (the base of the cone) represents the area where the bullet trajectory intersects the plane at a certain distance that contains the base of the cone. The position of the intersection point is RANDOMLY chosen within said area.

doesn't represent angles of weapon shots spread

What does angles have to do with anything? Angles in respect to what? Do you even know what an angle is?

Simplifications in their model are:

1) Bullets travel at stright line trajectories.
( They simplify by skipping drop due to gravity and deviation due to wind and bullet induced turbulence in the air ).

2) Bullets trajectory direction deviates at random angles from the direction shooter intended to point a gun at within ( 0 - maximum deviation angle ) range and each deviation angle value within range is equally probable.
( While empiric statistical data would provide some nonlinear probability distribution, close to Gaussian probably ).

Now please explain me what real physical aspect of shooting is modeled by spreading shots evenly accross arbitrary circle surface area ?
It doesn't model shit and would result in a model where hitting around the target is more probable than hitting at target.
Their model is at least simplification by omission of some aspects of the accurate one and in practice it stray away less from the accurate one too.

Go back to school.

 

[Alienman]

Yeah, random dice roll for each individual shot that will either hit or miss, and you can't affect it in any way. Not sure how that is better than % in any single way. But I got my answer to my question. Only way to improve the circle is by perks or equipment, and shots are left to random chance otherwise.

 

[passerby]

Hey, you started it. I was just questioning stuff in the video, I didn't know it would upset at least two of you that much.

You took the video's description of XCOM's hit calculation and somehow decided it applies to PP instead. If that doesn't deserve the 'retadred' rating, I don't know what does...

I didn't though, retard.

It's how I understood it though. Now, whether it's your failure at writing in english, or my failure at reading english, I'm not sure, but as you can see I'm not the only one

 

[Alienman]

I would bet on you two since where do I mention Xcom hit calculation in my post? I was just wondering how shooting would be calculated for each individual shots since % would not be used.

 

[passerby]

Yeah, random dice roll for each individual shot that will either hit or miss, and you can't affect it in any way. Not sure how that is better than % in any single way. But I got my answer to my question. Only way to improve the circle is by perks or equipment, and shots are left to random chance otherwise.

They could run a two variable integral function over the circle area that would calculate in the background with decent accuracy probability of hitting each body part and refresh it live as you move around aim cursor, as well as the sum of each body parts probabilities as probability of hitting the target at all. Just showing it graphically like they do now is perfectly enough to make an informed decission where to aim, but showing percentages over body parts wouldn't hurt either and would be cool to look at.

Edit:
With their simplified model, the integral function is just a constant. So it'd be not intense to calculate at all. Very close estimation of probability would be just a proportion of some radial grid points within body part area, to all radial grid points within the circle.
It always amaze me how programmers of a game with multimillion budged can't never figure out simple things players would enjoy.

 

[Grotesque]

Spoiler

bullets are shot are spread evenly within the cone

what part of RANDOM you don't fuckin understand?
they stated numerous times that shots are generated RANDOMLY within the cone of fire.
The blue circle is the base of the cone and its area is (should be) dependent on the distance to the target.

Now how they generate the random trajectories within weapon's cone they NEVER gave details about it.

hence diameter for 100% is exactly 2x bigger then for 50%.

I don't even...

because surface area of the circles doesn't represent angles of weapon shots spread.

Circle area (the base of the cone) represents the area where the bullet trajectory intersects the plane at a certain distance that contains the base of the cone. The position of the intersection point is RANDOMLY chosen within said area.

doesn't represent angles of weapon shots spread

What does angles have to do with anything? Angles in respect to what? Do you even know what an angle is?

Simplifications in their model are:

1) Bullets travel at stright line trajectories.
( They simplify by skipping drop due to gravity and deviation due to wind and bullet induced turbulence in the air ).

2) Bullets trajectory direction deviates at random angles from the direction shooter intended to point a gun at within ( 0 - maximum deviation angle ) range and each deviation angle value within range is equally probable.
( While empiric statistical data would provide some nonlinear probability distribution, close to Gaussian probably ).

Now please explain me what real physical aspect of shooting is modeled by spreading shots evenly accross arbitrary circle surface area ?
It doesn't model shit and would result in a model where hitting around the target is more probable than hitting at target.
Their model is at least simplification by omission of some aspects of the accurate one and in practice it stray away less from the accurate one too.

Go back to school.

You either are retarded troll or an imbecile.
You're like a schizoid that rambles about random shit without the ability to convey simple information and all he can do is to ruminate words.

I might've caught a glimpse of how you imagine they generate the randomness of trajectories by using variable angles but the rest is utter garbage ( Gaussian distribution, empiric statistical data lol )

Me back to school ok
You back taking your medication

 

[rusty_shackleford]

Current version of the game does not have accuracy stat but finished game should have it.

Then it is weird that not even once was mentioned that the circle area is dependant on the accuracy rating of the soldier. I would've thought that if you aim to make a video on the differences of the aiming mechanics compared to xcom, that would be on the list.



And now for some math time and how "honest" the game pretends to be.

For the shots to have a 50% probabilistic chance to "land" in the yellow circle, the yellow circle area should have 50% of the bigger blue circle.
After some measurements and basic calculations, on the right is how big the yellow circle should really be.

If they can't get even this easy stuff right, what confidence should I have they get the complex systems right behind true RNG or that what they advertise is indeed true at the math level?

because the former looks better from a UX perspective due to the reality being unintuitive

 

[passerby]

You either are retarded troll or an imbecile.
You're like a schizoid that rambles about random shit without the ability to convey simple information and all he can do is to ruminate words.

I might've caught a glimpse of how you imagine they generate the randomness of trajectories by using variable angles but the rest is utter garbage ( Gaussian distribution, empiric statistical data lol )

Me back to school ok
You back taking your medication

I've explained their model in literally two simple sentences, but I've also added two optional sentences *in brackets* explaining how I'd make it, if I wanted to simulate it accurately in my hipothetical dream tactical simulation game, on a forum about RPGs, the horror.

Lets leave my schizoid ass for a while...
Please explain us how distributing shots evenly across the whole circle surface area like you've proposed, is mathematically correct model for shooting genius. I'm waiting with anticipation for your simple explanation without schizoid ramblings.

 

[Grotesque]

explain us how distributing shots evenly across circle surface area like you've proposed is mathematically correct

where the fuck I proposed they do that?
Or at least try to explain the mental gymnastics you undertook to conclude that?

What I said was plain and simple: if a point has the ability to randomly appear on an area, if you split that area in two equally large halves, now the point has 50% chance to appear in one area AND 50% chance to appear in the other area.
Is that shit simple enough for you to comprehend?

 

[Grotesque]

Current version of the game does not have accuracy stat but finished game should have it.

Then it is weird that not even once was mentioned that the circle area is dependant on the accuracy rating of the soldier. I would've thought that if you aim to make a video on the differences of the aiming mechanics compared to xcom, that would be on the list.



And now for some math time and how "honest" the game pretends to be.

For the shots to have a 50% probabilistic chance to "land" in the yellow circle, the yellow circle area should have 50% of the bigger blue circle.
After some measurements and basic calculations, on the right is how big the yellow circle should really be.

If they can't get even this easy stuff right, what confidence should I have they get the complex systems right behind true RNG or that what they advertise is indeed true at the math level?

because the former looks better from a UX perspective due to the reality being unintuitive

or they could've wrote 35% instead of 50% near the yellow circle (35 or whatever the probabilistic chance number is in respect to that area)



Now let's all listen to the soothing voice of Tim Cain

 

[rusty_shackleford]

explain us how distributing shots evenly across circle surface area like you've proposed is mathematically correct

where the fuck I proposed they do that?
Or at least try to explain the mental gymnastics you undertook to conclude that?

What I said was plain and simple: if a point has the ability to randomly appear on an area, if you split that area in two equally large halves, now the point has 50% chance to appear in one area AND 50% chance to appear in the other area.
Is that shit simple enough for you to comprehend?

shot spread is typically modeled using a gaussian distribution, not uniform distribution

 

[passerby]

explain us how distributing shots evenly across circle surface area like you've proposed is mathematically correct

where the fuck I proposed they do that?
Or at least try to explain the mental gymnastics you undertook to conclude that?

What I said was plain and simple: if a point has the ability to randomly appear on an area, if you split that area in two equally large halves, now the point has 50% chance to appear in one area AND 50% chance to appear in the other area.
Is that shit simple enough for you to comprehend?

The circle radiuses will be proportional to %, when you distribute angle deviation of bullet trajectories from aiming direction uniformly within the deviation angle range. This is what they did, hence 50% chance results in half of the radius.

The surface areas for each 50% will be equal like you are proposing, only when you distribute shots uniformly across the circle surfaces.
If you do this and trace trajectories, the angle deviations will concentrate around maximum angle deviation, which is not only even less accurate model, but also would be annoying as fuck to play, since most of the shots would hit around the target.
It's even clearly visible in your own picture, where area representing 50% of hits is a thin outer diameter ring. You are too dumb to understand implications of your own model.

I am talking about uniform distribution of angles within radial coordinates and you are talking about uniform distribution of points on the surface within cartesian coordinates. It seems you don't understand these concepts, I'm done.

 

[Togukawa]

shot spread is typically modeled using a gaussian distribution, not uniform distribution

And if outer disc is 2sigma (95% chance), then inner disc at half width should be 68%. So if anything, the 50% hit circle should be smaller, not larger.
Uniform distribution for the angle deviation makes a lot less sense than using (truncated) gaussian, but I guess snapshot spent more time adding shoulderpads than thinking about their ballistics model.

 

[passerby]

explain us how distributing shots evenly across circle surface area like you've proposed is mathematically correct

where the fuck I proposed they do that?
Or at least try to explain the mental gymnastics you undertook to conclude that?

What I said was plain and simple: if a point has the ability to randomly appear on an area, if you split that area in two equally large halves, now the point has 50% chance to appear in one area AND 50% chance to appear in the other area.
Is that shit simple enough for you to comprehend?

shot spread is typically modeled using a gaussian distribution, not uniform distribution

This is even not what the argument was about. He called me a schizoid for even mentioning gaussian distribution.

He calculated circle radius representing 50% chance with uniform distribution across circle *area* not uniform distribution of distance deviation from the target, like Snapshot did and called it more accurate.

 

[luinthoron]

I would bet on you two since where do I mention Xcom hit calculation in my post? I was just wondering how shooting would be calculated for each individual shots since % would not be used.

You specifically mentioned a roll to see whether it hits or misses with no other outcomes possible. This is exactly what XCOM does. PP, on the other hand, has no roll to hit at all, but rather traces the path of each individual projectile and looks where they end up, be it hitting the target, an obstacle, your own soldier in the line of fire, or any object in the background. Is this explanation simple enough for you to understand?

 

[Grotesque]

explain us how distributing shots evenly across circle surface area like you've proposed is mathematically correct

where the fuck I proposed they do that?
Or at least try to explain the mental gymnastics you undertook to conclude that?

What I said was plain and simple: if a point has the ability to randomly appear on an area, if you split that area in two equally large halves, now the point has 50% chance to appear in one area AND 50% chance to appear in the other area.
Is that shit simple enough for you to comprehend?

shot spread is typically modeled using a gaussian distribution, not uniform distribution

And is it well suited to use the Gaussian probability density function instead the uniform one?
All values are equally likely to "happen" within a certain finite range with uniform distribution whereas with Gaussian distribution, values cluster around the mean (the average) with peripheral values less likely

 

[Grotesque]

He called me a schizoid for even mentioning gaussian distribution

you're a butthurt schizoid with the ability to express himself near the level of a worm and a virulent one on top of that
get laid

 

[Togukawa]

And is it well suited to use the Gaussian probability density function instead the uniform one?
All values are equally likely to "happen" within a certain finite range with uniform distribution whereas with Gaussian distribution, values cluster around the mean (the average) with peripheral values less likely

Yes.
https://en.wikipedia.org/wiki/Central_limit_theorem

 

[holla_cabezas_de_mierda]

I would bet on you two since where do I mention Xcom hit calculation in my post? I was just wondering how shooting would be calculated for each individual shots since % would not be used.

You specifically mentioned a roll to see whether it hits or misses with no other outcomes possible. This is exactly what XCOM does. PP, on the other hand, has no roll to hit at all, but rather traces the path of each individual projectile and looks where they end up, be it hitting the target, an obstacle, your own soldier in the line of fire, or any object in the background. Is this explanation simple enough for you to understand?

I actually don't understand this. How does PP 'decide' if you hit or not by 'tracing'? What determines a hit or miss?