Tweet
Hello Madden faithful - it's Monday, and it’s time for another blog! This week we'll be following up a little bit on the last PRO-TAK blog [link]. We covered gang tackling in detail last time, so for this blog we thought we’d dig a little deeper on our pass blocking, and also mention how PRO-TAK’s dynamic steering technology will help take our blocking a little bit farther into that next level that we know we need to hit of realism. At the top of nearly every community wish list for the past 3+ years has been OL/DL interaction…so this year we’re proud to say we’ve spent a really good amount of time improving this area, and we’ve also built a strong foundation that we can continue to add on to in the future.
A look back
Before we dig too far into the details of our blocking improvements, let’s take a look at previous blocking systems within football games. For this purpose, we’ll look at both the Madden NFL and the 2K series.
Madden NFL
Madden’s blocking system has typically always been based around quick looping animations to enhance user control. A quick looping/hand-fighting animation plays quickly (sometimes less than a second), and during that time the game checks to see if a special move should be tried, or if either player should move forward or backward, or if either player should disengage for any reason, or possibly if the animation should just continue to loop. The bonus of a system like this is that it can be typically very responsive. If you hit a button, it responds rather quickly and a new animation can start. If a ball carrier bounces his run outside after a block animation has started, the defender can break off of his blocker quickly and not get stuck in an animation for too long. Example video from Madden NFL 09:
The main drawback though is that you get very little movement in the animations. You can’t consistently form a pocket when players are playing a multitude of animations back-to-back-to-back. So, when a win/loss decision (that contains randomness) is being made 50 times per play, your level of predictability and consistency in the system is very sub-par. In real life, we all know that even an average defensive lineman has a chance to get past an all-pro offensive lineman (albeit small), but if we put that kind of percentage into our logic, the animation selection can end up being inconsistent since it’s checking to see who’s winning so often.
2K
We all definitely agree that the 2K series did a good job focusing on blocking in their iterations. They had true pocket forming, good motion capture, and good AI. 2K was actually a great visual target we were looking to reach when designing our system for this year. For typical blocking situations, they implemented much longer animations (instead of short looping ones), and this helped their visual quality quite a bit. I don't have a full video of their good stuff...but believe me we think they did a really good job.
Their system also had a drawback though, (which is typically always the negative with using long animation clips) since the user and CPU, in essence, lost control during animation playback. When an animation started playing, the AI really couldn’t update or change it, so you’d see situations where defenders engaged with blockers play really long animations the end up taking themselves well out of the play because they can’t recognize that someone ran past him:
After looking at the best aspects of both games extensively, we set out designing some of the things we wanted out of our new system. We scrapped a lot of our logic (and nearly all of our animations for players getting engaged) for both run blocking and pass blocking. For run blocking, we wanted players flowing the play much better, especially on outside runs, and for pass plays, we wanted the real pocket forming. But we knew that we needed to combine the best of both worlds, taking the positives of each system and limiting the ‘negatives’. We wanted you the gamer to still feel like you were in control, but also wanted much better visual quality and realism. We knew that implementing much longer animations was key, but we also knew that the players in those animations had to be ‘informed’ to what was going on around them. We came up with two new solutions to solve these issues:
1. Branch Windows
A quick explanation of a branch window: a specific duration of time during the playback of an animation when another animation can be triggered. In Madden NFL 09, we added branch windows to all of our tackles - as something that has been referred to the “Breakout Engine†(by our marketing partners) to our tackling. These branch windows allowed you to dynamically break out of tackle animations after they had already started. IMO this has been one of those subtle forever-game-changing features, mainly because I find myself trying it when I play other games, and I miss it if it’s not there. To me that’s usually the signal of a gameplay mechanic being a real success...something like the hit stick is also in that same category…you don’t know how you lived without it once you play with it. Anyways, we brought this same technology over to our blocking animations for Madden NFL 10...so now for every single engaged blocking animation, you can use the right stick to pick a direction to try and break off of the block while it’s in progress. With us adding long animations for visual quality while still retaining the goal of keeping the animations ‘informed’, we had to make sure that they could be interrupted, whether it’s to put your hands up to knock down a pass or to spin inside off of a lineman after you’ve driven him outside.
Something different about blocks though that tackles don't have is that there are actually two branch windows. There is one branch window at the beginning before you get engaged…this was key to make sure that you could keep your hands out and get away from a guy before getting ‘engaged’. This is typically where the elite defensive linemen can cause havoc and put immediate pressure on the quarterback or running back by getting in really quickly. The second branch window is much longer (it lasts the duration of the main section of the block animation) and this is where you can perform your standard special moves. Using special moves in this branch window creates those moments you see every Sunday - examples like a defensive end looping around to form the pocket and then dynamically "branching" off of his block to get to the quarterback (so you don't have those situations where the defender doesn't seem aware of his surroundings). Here are some perfect examples of the branch windows in action - showcasing a defensive lineman being able to break out of his animation early to make a play:
2. PRO-TAK Steering and Manipulation
The next piece in our puzzle was a big one, and that was applying the PRO-TAK technology to the blocking animations. Again looking back at those 2K examples, we didn’t want the case where defenders would go right past the ball without recognizing it. You can’t necessarily capture just a ton more animations for each situation either, because the movement of the QB or running back is totally dynamic, and obviously can change drastically after the blocking animation has already started. So the steering technology used in the PRO-TAK gang tackles was re-purposed for these animations…allowing us to procedurally manipulate these animations and move them to the target based on the ratings of the players involved. In layman’s terms, this means that a strong defender can push a blocker right into the face of the ball carrier, or a blocker can steer a defender wide of his target, and the exact same animation can be used. I showed this video last week but I’ll show it again as it showcases how the exact same animation can be dynamically manipulated based on the circumstances of the game:
We are looking forward to showing more blocking improvements (especially in run blocking) in the coming weeks as it gets more finely tuned. Hopefully this shows a good start and the positive direction that our blocking is moving in. Creating real life blocking AI in a football game is obviously an extremely complex matter, but we think the system we have implemented is a great foundation for us to build on for the future. Feel free to post in the comments section below, or put your ideas or suggestions in the forums.
A look back
Before we dig too far into the details of our blocking improvements, let’s take a look at previous blocking systems within football games. For this purpose, we’ll look at both the Madden NFL and the 2K series.
Madden NFL
Madden’s blocking system has typically always been based around quick looping animations to enhance user control. A quick looping/hand-fighting animation plays quickly (sometimes less than a second), and during that time the game checks to see if a special move should be tried, or if either player should move forward or backward, or if either player should disengage for any reason, or possibly if the animation should just continue to loop. The bonus of a system like this is that it can be typically very responsive. If you hit a button, it responds rather quickly and a new animation can start. If a ball carrier bounces his run outside after a block animation has started, the defender can break off of his blocker quickly and not get stuck in an animation for too long. Example video from Madden NFL 09:
Madden 09 blocking from easports on Vimeo.
The main drawback though is that you get very little movement in the animations. You can’t consistently form a pocket when players are playing a multitude of animations back-to-back-to-back. So, when a win/loss decision (that contains randomness) is being made 50 times per play, your level of predictability and consistency in the system is very sub-par. In real life, we all know that even an average defensive lineman has a chance to get past an all-pro offensive lineman (albeit small), but if we put that kind of percentage into our logic, the animation selection can end up being inconsistent since it’s checking to see who’s winning so often.
2K
We all definitely agree that the 2K series did a good job focusing on blocking in their iterations. They had true pocket forming, good motion capture, and good AI. 2K was actually a great visual target we were looking to reach when designing our system for this year. For typical blocking situations, they implemented much longer animations (instead of short looping ones), and this helped their visual quality quite a bit. I don't have a full video of their good stuff...but believe me we think they did a really good job.
Their system also had a drawback though, (which is typically always the negative with using long animation clips) since the user and CPU, in essence, lost control during animation playback. When an animation started playing, the AI really couldn’t update or change it, so you’d see situations where defenders engaged with blockers play really long animations the end up taking themselves well out of the play because they can’t recognize that someone ran past him:
Blocking animations - APF2k8 from easports on Vimeo.
After looking at the best aspects of both games extensively, we set out designing some of the things we wanted out of our new system. We scrapped a lot of our logic (and nearly all of our animations for players getting engaged) for both run blocking and pass blocking. For run blocking, we wanted players flowing the play much better, especially on outside runs, and for pass plays, we wanted the real pocket forming. But we knew that we needed to combine the best of both worlds, taking the positives of each system and limiting the ‘negatives’. We wanted you the gamer to still feel like you were in control, but also wanted much better visual quality and realism. We knew that implementing much longer animations was key, but we also knew that the players in those animations had to be ‘informed’ to what was going on around them. We came up with two new solutions to solve these issues:
1. Branch Windows
A quick explanation of a branch window: a specific duration of time during the playback of an animation when another animation can be triggered. In Madden NFL 09, we added branch windows to all of our tackles - as something that has been referred to the “Breakout Engine†(by our marketing partners) to our tackling. These branch windows allowed you to dynamically break out of tackle animations after they had already started. IMO this has been one of those subtle forever-game-changing features, mainly because I find myself trying it when I play other games, and I miss it if it’s not there. To me that’s usually the signal of a gameplay mechanic being a real success...something like the hit stick is also in that same category…you don’t know how you lived without it once you play with it. Anyways, we brought this same technology over to our blocking animations for Madden NFL 10...so now for every single engaged blocking animation, you can use the right stick to pick a direction to try and break off of the block while it’s in progress. With us adding long animations for visual quality while still retaining the goal of keeping the animations ‘informed’, we had to make sure that they could be interrupted, whether it’s to put your hands up to knock down a pass or to spin inside off of a lineman after you’ve driven him outside.
Something different about blocks though that tackles don't have is that there are actually two branch windows. There is one branch window at the beginning before you get engaged…this was key to make sure that you could keep your hands out and get away from a guy before getting ‘engaged’. This is typically where the elite defensive linemen can cause havoc and put immediate pressure on the quarterback or running back by getting in really quickly. The second branch window is much longer (it lasts the duration of the main section of the block animation) and this is where you can perform your standard special moves. Using special moves in this branch window creates those moments you see every Sunday - examples like a defensive end looping around to form the pocket and then dynamically "branching" off of his block to get to the quarterback (so you don't have those situations where the defender doesn't seem aware of his surroundings). Here are some perfect examples of the branch windows in action - showcasing a defensive lineman being able to break out of his animation early to make a play:
2. PRO-TAK Steering and Manipulation
The next piece in our puzzle was a big one, and that was applying the PRO-TAK technology to the blocking animations. Again looking back at those 2K examples, we didn’t want the case where defenders would go right past the ball without recognizing it. You can’t necessarily capture just a ton more animations for each situation either, because the movement of the QB or running back is totally dynamic, and obviously can change drastically after the blocking animation has already started. So the steering technology used in the PRO-TAK gang tackles was re-purposed for these animations…allowing us to procedurally manipulate these animations and move them to the target based on the ratings of the players involved. In layman’s terms, this means that a strong defender can push a blocker right into the face of the ball carrier, or a blocker can steer a defender wide of his target, and the exact same animation can be used. I showed this video last week but I’ll show it again as it showcases how the exact same animation can be dynamically manipulated based on the circumstances of the game:
Blocking steering from easports on Vimeo.
We are looking forward to showing more blocking improvements (especially in run blocking) in the coming weeks as it gets more finely tuned. Hopefully this shows a good start and the positive direction that our blocking is moving in. Creating real life blocking AI in a football game is obviously an extremely complex matter, but we think the system we have implemented is a great foundation for us to build on for the future. Feel free to post in the comments section below, or put your ideas or suggestions in the forums.
Comment