Week 22 – 21/05/11 – 27/05/11
The gameplay screen I showed previously was the SCV screen. Like I said, balancing and tweaking is still one of my jobs!
For balancing purposes we are outputting SCV on the SCIV environment.
For those people who missed the Famitsu stream, I confirmed one thing. Nightmare will return in SCV.
Thnx to all who watched the Famitsu stream. I was talking while watching everyone’s comments. Harada-san was picking on me the whole time..
The most relevant part of that stream is this few seconds of blurry gameplay.
This is only a few months in, so it makes perfect sense that they’re still screwing around with an engine that’s mostly Soulcalibur 4. Even Soulcalibur 4 was built on the remains of Soulcalibur 3.
Eventually we will see the full extent of Soulcalibur 5’s look, but like the jump from Soulcalibur 2 to Soulcalibur 3 it’s important to keep in mind that these games are on the same console, and though they can refine and tweak many things, they don’t have the luxury that other console games have with how they push their bullshots, because Soulcalibur 5 still has to run at a solid 60fps, while most other big titles are lucky to be pushing 30fps at the best of times. Seriously: next time you’re watching gameplay of a new console game notice how much it chugs along. Most people can tell the difference between 30 and 60fps, and chances are that console game is running at 30, and stuttering below that when the effects start to fly.
For anyone interested, John Carmack recently gave a very good interview about not only iD’s new game, Rage, but also his thoughts on various parts of the industry and what it was like to design a game for current generation consoles. While little of it is relevant to Soulcalibur 5 in particular, it’s always nice to hear Carmack talk about technology and the video game industry.
The right time to show them will come 😉 @DivineDevil999 do u have pictures of the new characters XD
And we have to wonder when that is. With e3 over (at the time of this writing, not at the time of Daishi’s post), the next events for Soulcalibur news are Evolution at the end of July, then the Tokyo Game Show in September. It’ll probably show up elsewhere in the meantime, but those look like the most solid dates right now.
Say what you want about Soulcalibur 3, but at least it continued the series trend of having neat intros. Which is more than Soulcalibur 4 ever managed. You have to wonder how much of a focus the guest characters will get in the Soulcalibur 5 intro.
There’s that. I’ve about run out of things to say about cosplay. I wonder which one of those 2 will make it into Soulcalibur 5. (No I don’t.)
Let me talk about CPU. SCBD has the best CPU in the series because a lot of the AI is manually inputted by the game balancers.
I couldn’t check all characters due to time constraints so some are better than others. Like Setsuka; too strong after a certain difficulty.
The CPU’s moves are inputted by the game balancers taking into account the distance, radius, properties etc.
How the CPU reacts after a hit, a guard, a miss, a counter, in a downed state, in the air are all inputted manually.
We then use probability to add just the right amount of random elements.
So the CPU reacts more like a human, loses control, does horizontal attack against side steps, misses guard impacts etc.
The more variation and patterns we add, the more deep the CPU becomes. So the more time we can spend on each character, the better.
If we don’t have time, the AI will have inhuman reflexes or become one dimensional which makes CPU battles boring.
As I’ve said before, there is a fundamental flaw with AI design in fighting games that doesn’t really have anything to do with the AI itself.
Most people spend most of their time against the AI in a fighting game trying to get something else besides good matches. Almost always they’re trying to unlock something or advance through a mode that is barely related to normal versus play. And while trying to power through AI matches as quickly and efficiently as possible the last thing they want is to have to think about every round. Not when they’re going through a mode that will involve hundreds of matches, or one that they want to quickly get through in order to unlock something.
This makes the idea of “good” AI seem less attractive the more you have to play against it. It’s tangentially related to the way most JRPGs (and action games) have been structured — the main feature being attrition. No single encounter is supposed to be so difficult that a player cannot proceed, but by the time they’re reaching the end of an enclosed dungeon their resources should be taxed enough to make the boss a challenge. In the same way, many single player fighting game modes are filled with dozens of filler battles that lead up to a climactic encounter of some sort, or at least can be punctuated by regular boss fights, only the resources they’re taxing aren’t necessarily on the screen, but in the player’s head: having to fight dozens of AI matches in a row is exhausting, and if each of them was putting up a fight that required equal amounts of thinking and effort it would be exponentially more so.
There are special cases, like the Quest mode in Virtua Fighter games, but I imagine most people would still like to get through those as fast as possible.
Soulcalibur has had the right idea for a while with the special matches in Weapon Master mode on up. Make the AI better at responding to (and utilizing) the special conditions of those gimmick matches. That would make them more engaging and interesting to play against than days of work poured into “good” AI that will still lose to 22B spam.
Also, ghost AIs formed by players, like Soulcalibur 2 and 3 had in the arcade.