NBA 2K15 Gameplay Video - Miami Heat vs. New York Knicks (IpodKingCarter)

[Da_Czar]

Quote: Originally Posted by eko718

After watching that vid again, it does seem like actions should be sped up significantly. Players need to go around picks and execute movements with a sense of urgency. The computer just jogs around in this vid... none of these actions seem to actually work and get them open, which results in a stagnated iso affair.

One note on that. In short distances players don't always sprint in game because if the user lead passes it will mess up the rhythm of the play. If you watch calderon he set's up the cut first and then he has, what.. 10 feet between were he is and where he will receive the ball. Sprint cuts in the game wouldn't allow him to stop in time.

All the movement this year is coordinated. If you recall last year the play would break with complex movement because guys were sprinting and bumping.

Also as you see more of the game you will see we have a variety of movement. If you always move at the same speed your easier to guard was the philosophy. So with offscreen plays like floppy's or 12 quick's you will not see all out turbo you will see him build speed and sometime hit a burst when it counts to get open.

I think this gradual change of speed and the smoothness of the animations is giving the impression of a lack of urgency. Maybe they need to be even MORE urgent which is what I hear you saying.

We have counters for cheating defenders so it is not all about beating a guy to a spot this year. Also the D has penalties for turbo usage so offscreen has a tad more cat and mouse this year.

What I THINK you will see as you see more is that the game is running at a 48 min pace. It is hard to get a feel for that just watching a play here or a play there. Also you should notice teams like houston have much less complex plays that therefore run faster that theoretically should enable them to have more possessions than a team like the knicks.

NBA players don't always sprint everywhere. I do agree that in general the game doesn't LOOK explosive be it changing speeds on offense or running off a screen off ball.

It is def a different pace than what you were used to in 2k14. Once you get hands on if you choose to pick it up or demo it. And you can see the benefits of the movement we have on offense then I want to know how you feel about speeding that up and negatives that are associated with it.

We matched the offense closely to what is the most the motion system can handle. I changed philosophy in the design of the plays 3 times this cycle according to how the motion system was being developed.

Considering all pro's and con's I felt really good about where the system was at.

But as I said my ears are def open. Just sharing that it was not a trivial decision and a lot of testing and thought went into it my part. That by no means, means LOL it can't be better.

[AceDawg5]

Quote: Originally Posted by Da_Czar

Thanks for your comments. Fist let me say yes the AI will not look fantastic on every possession. Just an FYI the AI was evaluating about 3 different options that all could have separate branches within the triangle. While waiting there It decided to bypass them and you can see Melo then come off a back-screen for the post up. Then he decided to attempt a score instead of going to Melo. For movement through the lane I didn't like it last year when bigs were sprinting to go short distances It looked unnatural especially with all the bumping etc. So we made sure to have the bigs use more controlled movements. I will be eagerly watching for feedback from those who play in depth to see if that turns out to be a benefit or if we should speed them up. So the AI wasn't stuck he was just evaluating options. Maybe we need to speed that up.. We will see. I know it looks familiar to last year when he wasn't really processing anything The difference is last year he really only had one option and was waiting for it or stuck. This year it's a bit different. Time and the feedback of fine gents on here will let us know how far away we are. I believe we will be able to talk about some of the improvements here soon.

My only gripe about this video is why r the players sprinting off of the inbound pass. I don't think that makes sense but they do it on every inbound. Is that the CPU overusing turbo or the actual inbound animation? In other words, do I have control over how often that happens?

[luda06]

Marc Price (of Good Game Bro) has been critical of the A.I. in NBA 2K (and their competition) for some time, with some excellent analysis, comparing real NBA offenses with their 2K counterparts. The footage from 2K14 looks largely similar to what we see in the OP. The A.I. executes, but it is generally slow and illogical. This is an area that should be in consideration for an overhaul for the next iteration.

The A.I.'s logic should be much more simple in nature. Currently it appears to calculate inefficient variables (location, ratings, defensive position, play-calling, etc.) in an inefficient equation. I think an analogy would be the A.I. attempting to use trigonometry to triangulate the distance between my couch and the TV, instead of just grabbing a ruler.

The A.I. does some things well, such as attacking a defender that is out of position, and that may be a huge part of the problem. It's too reliant on expecting an opposing defense to slip-up, and the plan b is to just cheese the opponent with precision passes, precisely timed jump-shots, or a bevy of isomotion/post moves.

Every play should be able to bend a defense based on the personnel at hand. The Knicks' A.I. using Amare should have taken into account the time available (9 seconds remaining). From there, the Knicks should have either cleared out for Amare to isolate and attack McRoberts, or hurried up for Melo to catch the ball deeper into the low post.

The opening seconds of the OP's video shows no real action going on. Again, this is unacceptable. Calderon is not the personnel at hand to have the ball in his hands with 9 seconds on the clock, if he's not initiating any sort of two-man action.

The A.I. should immediately break a play-set once it determines there's not enough time for it to develop, and shouldn't settle for a weird Calderon/D-Wade isolation that eats up the shot clock if something breaks down or goes awry. Isomotion isn't working? Okay, quick post. Another pick and roll? Maybe, there should be dribble hand-offs in the next game as well. I don't all of these options to occur in one possession, but systematically develop based on the variables I've listed a bit later.

So, time and personnel should be constants within this equation the A.I. calculates each possession offensively. As a constant, these numbers never change. From there, the variables should be questions such as:

What did the defense do the previous trip down the floor? Did we have open shooters? Did we get a uncontested shot, or draw a foul? Were we able to set up our offense?

Basketball is no different from a game of chess or football in regards to strategizing what to do. The A.I. should be calculating moves way ahead of the opponent similar to how humans do. The A.I. should be attempting to set up the knock-out punch instead of randomly calling plays from a play-book.

Playbooks themselves should be another variable the A.I. calculates within this equation. Coaches should be able to pick out and run the most time efficient sets based on their personnel. From there, contengencies should be very simple, yet carry the same purpose that fits the skillet of the players on the floor.

Floppy plays for example? They're terrible in 2K. Human controlled opponents will cheat the defender around the screen and attempt to intercept the pass at best. These plays should probably be removed from the game if guards and forwards using the screens can't flare out or back-door to the hoop if the defender cheats, or if another defender helps.

Why? Because the point guard will stand for at least 8 seconds, waiting for the play to develop. A simple floppy, into a low-post entry pass provides two opportunities for a wing player to have an open look in the same span of time. Penetration from a pick and roll, can find an open shooter just the same.

The A.I. should calculate what play to run based on the personnel. San Antonio? Floppy post. Los Angeles Clippers? Pick and roll action. The initial desired result is the same (an open shooter), but the contingencies are baked right into the play design itself, making the reads much more simple in nature for a player to make. Either a teammate is open, or the player with the ball is in prime position to score the ball.

[Vni]

Great post man.

[The 24th Letter]

I definitely understand the gripe with some of 2k's animations...but as these conversations carry on, there tends be a bit of an exaggeration of how detrimental they are to game. We saw a game last year that suffered from a lack of sufficient animations and it wasn't pretty. We've also seen this in past 2k games. For me personally, I
could have the most crisp controls ever created...if contact can't be generated a reliable rate because of the lack of sufficient animations...THATS not basketball either. People like to throw "physics" out there as the end all solution to all of this...but those physics would be plugged into animations as well.

Animations are necessary. There just needs to be a balance.

[Boilerbuzz]

Quote: Originally Posted by turty11

as bad as it is, you can get a more accurate assessment on game play from CPU vs CPU rather than watching someone who is really bad vs the CPU or someone who is far too good for the difficult setting these early game play releases are on. there is no hiding from it, what we have seen game play wise is HORRENDOUS, early build or not.

No you can't. The game isn't designed to be played between 2 AI players. It's a user game and if it wasn't, then why is it a game? You guys over react to the smallest things, frankly. I don't like the video because the decision making and actions seemed basic or not as fluid and creative as a human. I PLAY the game. Not watch it. THAT is why you get the game.

[JBulls]

Quote: Originally Posted by The 24th Letter

People like to throw "physics" out there as the end all solution to all of this...but those physics would be plugged into animations as well. Animations are necessary. There just needs to be a balance.

Physics would be the end all be all if you had a supercomputer as your console lol but they have to work with the hardware they have.

It's just that gameplay + animation has yet to make that visible "next-gen" jump like the graphics have imo.

[Boilerbuzz]

Quote: Originally Posted by vtcrb

No Ball Movement. The CPU looks LOST on the floor, standing around. Nothing Fluid at all about this. CPU vs CPU is BEST way to test out a game and I have to say i did expect more with all the hype the game has been getting. Now USER vs USER should be fine for guys, as it pretty much has been for years, but I Dont see where the CPU AI has improved. Still time for them to make tweaks.

Compared to previous versions, this is a HUGE improvement. In the previous versions of the game, I would have noticed it was AI v AI immediately. Not so much this time. But if you guys want to convince yourselves of worst, have at it. Doesn't change anything. Read DaCzar's post about the depth of the play system. Explains quite a bit actually.