Board index » America's Army: Proving Grounds » AA:PG General Discussion




Forum locked This topic is locked, you cannot edit posts or make further replies.  [ 214 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6 ... 22  Next
Author Message
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 12:42 pm 
Offline

Joined: Fri Apr 11, 2014 4:05 pm
Posts: 431
I watched the video frame by frame:

The first shot was behind the enemy's head and above his backpack (this was where your green aimpoint was before recoil kicked it to the right). A miss.

The second shot was just behind his head to the right but you aimpoint dropped to his upper should/neck area. Another miss.

The third shot was between his left elbow and left shoulder in front of his face but a little to high to make a hit. Another miss.

Consider these factors:

A bullet leaves the barrel at an upward-right angle due to its spin. Your second shot was close enough where it still should have been a kill shot but only according to the paintball decal.

The reddot aimpoint (green for you) has a 4 x 4 pixel resolution. I assume the bullet is only 1 pixel in size. The paintball decal is at least 8 times larger than the aimpoint. If a bullet is 4 pixels square then you have a better argument on the second shot. I'm using your screenshot of 1920x1080 resolution with 72 pixels per inch blown up 3200% under PS.

If the game had absolute zero lag, a jury might award you a kill on the second shot. I am still uncertain and feel your frustration.

One thing is for sure... Jody ain't got your Cadillac and is Guilty!!

_________________
"To know your Enemy, you must become your Enemy.”
― Sun Tzu


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 1:05 pm 
Offline

Joined: Sun Jun 16, 2013 8:41 am
Posts: 1512
io_i_oi wrote:
Consider these factors:

A bullet leaves the barrel at an upward-right angle due to its spin. Your second shot was close enough where it still should have been a kill shot but only according to the paintball decal.

The reddot aimpoint (green for you) has a 4 x 4 pixel resolution. I assume the bullet is only 1 pixel in size. The paintball decal is at least 8 times larger than the aimpoint. If a bullet is 4 pixels square then you have a better argument on the second shot. I'm using your screenshot of 1920x1080 resolution with 72 pixels per inch blown up 3200% under PS.


The game doesn't model bullet spin. Bullets leave the barrel (well, your pawn's camera) traveling straight and are affected by gravity, but practical the effect is *extremely* minor.

As far as I know, the bullet basically has no width; its path is modeled as a line and the game asks "Does this line intersect this mesh?" A pixel isn't really a good unit for discussing in-game dimensions since a pixel doesn't have fixed size and the bullet doesn't really have a dimension anyway.


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 6:11 pm 
Offline

Joined: Fri Apr 11, 2014 4:05 pm
Posts: 431
Hey Bull, I understand the game mechanics.

Yes, the bullet is just a line traveling in a predetermined arc using a mathematical formula. The soldier's head is a rendering for graphics. It is roundish. The hit box (which is a square box) is placed around the head and attached to lower body hit boxes.

Since lag may cause the enemy to actually be 12 inches to the right or left of where it appears to the shooter then how can anyone tell if hit boxes are not registering correctly? You would have to host a 1v1 game on the same computer and test with no graphic rendering to see where the bullet path intersects.

We have a tendency to believe what we see in a game and not realize the mechanics behind the graphics. My post was pun because I get frustrated with shots too, but this was my point:

Quote:
If the game had absolute zero lag, a jury might award you a kill on the second shot. I am still uncertain and feel your frustration.


I am surprised that "bullet spin and climb" is not calculated in. It's a simple formula to program in.

_________________
"To know your Enemy, you must become your Enemy.”
― Sun Tzu


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 6:21 pm 
Offline
Volunteer Community Manager
Volunteer Community Manager

Joined: Mon Aug 02, 2004 11:10 am
Posts: 3695
io_i_oi wrote:
Hey Bull, I understand the game mechanics.

Yes, the bullet is just a line traveling in a predetermined arc using a mathematical formula. The soldier's head is a rendering for graphics. It is roundish. The hit box (which is a square box) is placed around the head and attached to lower body hit boxes.

Since lag may cause the enemy to actually be 12 inches to the right or left of where it appears to the shooter then how can anyone tell if hit boxes are not registering correctly? You would have to host a 1v1 game on the same computer and test with no graphic rendering to see where the bullet path intersects.

We have a tendency to believe what we see in a game and not realize the mechanics behind the graphics. My post was pun because I get frustrated with shots too, but this was my point:

Quote:
If the game had absolute zero lag, a jury might award you a kill on the second shot. I am still uncertain and feel your frustration.


I am surprised that "bullet spin and climb" is not calculated in. It's a simple formula to program in.


That's not correct. The game doesn't use hitboxes. It uses the mesh intercept that was previously described.

_________________
Image

Image


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 6:24 pm 
Offline

Joined: Sun Jun 16, 2013 8:41 am
Posts: 1512
io_i_oi wrote:
Since lag may cause the enemy to actually be 12 inches to the right or left of where it appears to the shooter then how can anyone tell if hit boxes are not registering correctly?


Client-side hit detection was designed around removing lag from consideration for the shooter. The mesh used for collision by the client is in the same place as the mesh where the enemy soldier appears to be.


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 6:27 pm 
Offline

Joined: Sun Jun 16, 2013 8:41 am
Posts: 1512
m_hermann wrote:
That's not correct. The game doesn't use hitboxes. It uses the mesh intercept that was previously described.


Well, there's the player model, and then there's a much simpler set of underlying meshes that are used for hit detection. The simpler model saves on CPU effort for calculating whether line A crosses plane B etc. and could fairly be referred to as "hit boxes" (or, more accurately, "hit spheres, pills, cylinders, and boxes" :P ). The important point is that the head really is treated as a spherical shape that's quite well matched to the player model, not a box.

That's undoubtedly what you were referring to, just thought I'd clarify a bit :mrgreen:

JaK posted images before of those meshes (along with a few other things that are normally hidden), but I think those are out of date now.


Last edited by BCPull on Wed Oct 22, 2014 6:33 pm, edited 1 time in total.

Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 6:32 pm 
Offline
Developer
Developer

Joined: Tue Sep 11, 2012 3:28 pm
Posts: 413
io_i_oi wrote:
Hey Bull, I understand the game mechanics.

Yes, the bullet is just a line traveling in a predetermined arc using a mathematical formula. The soldier's head is a rendering for graphics. It is roundish. The hit box (which is a square box) is placed around the head and attached to lower body hit boxes.

Since lag may cause the enemy to actually be 12 inches to the right or left of where it appears to the shooter then how can anyone tell if hit boxes are not registering correctly? You would have to host a 1v1 game on the same computer and test with no graphic rendering to see where the bullet path intersects.

We have a tendency to believe what we see in a game and not realize the mechanics behind the graphics. My post was pun because I get frustrated with shots too, but this was my point:

Quote:
If the game had absolute zero lag, a jury might award you a kill on the second shot. I am still uncertain and feel your frustration.


I am surprised that "bullet spin and climb" is not calculated in. It's a simple formula to program in.


A hit box is not a box nor is it a square, it is pill shaped if anything and matches up extremely close to the character mesh in all areas. It is within 1/16 of an inch in all areas.

Under a client side hit detection target is where your client draws it. Thus the "client" part of client side hit detection. It is under the server side model where the target may be ahead or behind where it appears to be, depending on latency.


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 6:47 pm 
Offline

Joined: Sun Jul 06, 2008 1:20 pm
Posts: 1010
[Dev]CaterpillaR wrote:
Under a client side hit detection target is where your client draws it. Thus the "client" part of client side hit detection. It is under the server side model where the target may be ahead or behind where it appears to be, depending on latency


You see this is the part i dont get. You and the rest of the devs appear to be so convinced that server side is where a hit box may be behind or ahead. As far as i see its pretty consistently where it should be.

The fundamental issue im having with understanding the logic behind client side which is "so much better" because the box is calculated on the client and not the server. But what happens when a calculated hit on 1 persons client is a miss on another? How does the system decide whos correct? Dev's completely ignore the flip side to this battle between server-side and client-side.

Dont just feed me why your way is right. Tell me why server side is really so much worse. Because there are logical flaws in saying that hit-box miscalculations are exclusive to server-side.

_________________
eKComm


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 6:59 pm 
Offline

Joined: Sun Jun 02, 2013 8:49 pm
Posts: 4049
Location: Behind you, Boo!
EKCommander wrote:
[Dev]CaterpillaR wrote:
Under a client side hit detection target is where your client draws it. Thus the "client" part of client side hit detection. It is under the server side model where the target may be ahead or behind where it appears to be, depending on latency


You see this is the part i dont get. You and the rest of the devs appear to be so convinced that server side is where a hit box may be behind or ahead. As far as i see its pretty consistently where it should be.

The fundamental issue im having with understanding the logic behind client side which is "so much better" because the box is calculated on the client and not the server. But what happens when a calculated hit on 1 persons client is a miss on another? How does the system decide whos correct? Dev's completely ignore the flip side to this battle between server-side and client-side.

Dont just feed me why your way is right. Tell me why server side is really so much worse. Because there are logical flaws in saying that hit-box miscalculations are exclusive to server-side.



They have explained on several occasions the pro's AND con's of each system. They are leaning to the side which levels the playing field for high pings. They have acknowledged within so many words that the CSHD model could also favor those high pings. But they choose this over high pings being isolated off the servers completely. You cannot blame them for that. This is also why they released the SS switch. For those players/servers without the foreign player requirements.

_________________
Image
-ACI- Launcher Mod (Auto-Installer)
Custom Loading Screen Video's
Image
Steam --> ℑąƙ
Stats --> ℑąƙ



Quote:
Only because a mod asked nicely.


Top 
 Profile  
 
 Post subject: Re: Hit detection video
 Post Posted: Wed Oct 22, 2014 7:11 pm 
Offline

Joined: Fri May 02, 2014 3:25 pm
Posts: 216
Location: Philly
JaK- wrote:
EKCommander wrote:
[Dev]CaterpillaR wrote:
Under a client side hit detection target is where your client draws it. Thus the "client" part of client side hit detection. It is under the server side model where the target may be ahead or behind where it appears to be, depending on latency


You see this is the part i dont get. You and the rest of the devs appear to be so convinced that server side is where a hit box may be behind or ahead. As far as i see its pretty consistently where it should be.

The fundamental issue im having with understanding the logic behind client side which is "so much better" because the box is calculated on the client and not the server. But what happens when a calculated hit on 1 persons client is a miss on another? How does the system decide whos correct? Dev's completely ignore the flip side to this battle between server-side and client-side.

Dont just feed me why your way is right. Tell me why server side is really so much worse. Because there are logical flaws in saying that hit-box miscalculations are exclusive to server-side.



They have explained on several occasions the pro's AND con's of each system. They are leaning to the side which levels the playing field for high pings. They have acknowledged within so many words that the CSHD model could also favor those high pings. But they choose this over high pings being isolated off the servers completely. You cannot blame them for that. This is also why they released the SS switch. For those players/servers without the foreign player requirements.



I bet the eur0s were pumped when CSHD was put back in place on the official servers.

_________________
#bp game director & founder


Top 
 Profile  
 
Display posts from previous:  Sort by  
 
Forum locked This topic is locked, you cannot edit posts or make further replies.  [ 214 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6 ... 22  Next

Board index » America's Army: Proving Grounds » AA:PG General Discussion


Who is online

Users browsing this forum: No registered users and 5 guests

 
 

 
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
cron