jump to navigation

Back to the roots: computer vision ftw July 9, 2010

Posted by Cesar in working me.
Tags: , , ,
1 comment so far

crazy image montage that makes me think of computer vision

Here at Other Ocean, I recently shifted from an iPad/iPhone4 game to another project. While it was great to work with shaders, glsl and what not on the previous project, the interesting thing about this new one is that it will give me an opportunity to work with computer vision again, after a long time.

And I confess I missed it. A lot.

Without getting into details, I’ve been researching image processing and analysis like crazy again and the whole experience has been a lot of fun. Looking for algorithms, going through the IEEE Xplore digital library, searching the ACM Portal, it all brings me back memories from my time at Unicamp, working on my MSc.

The system I’m developing uses what I already know of computer vision, but also incorporates a bunch of new stuff, so there’s a lot of learning involved. I know it means I’m a super geek but right now I have a big pile of printed papers on my desk and I love it.

I’ll write more about it when the project is released.

See you space cowboys…

Advertisements

Project Natal update: looking good! March 6, 2010

Posted by Cesar in gaming me.
Tags: , , , , ,
add a comment

Peter Molyneux, the famous Lionhead designer behind the Fable series, recently talked about improvement in Natal technology. In a previous post, I mentioned Project Natal was dropping the internal chip and the camera had disturbing lags that were preventing optimal gameplay.

It seems Microsoft is taking care of that and the new version of the device is a lot better. I also noticed it is not mentioned anywhere that the hardware reached it’s final state. That means Microsoft is still working to make it even better, which is great.

So one of the concerns, the lag, is apparently going to be resolved by the end of the year, when Project Natal hits the market.

I would love to see the internal chip go back into the product, but that’s not going to happen. However, while it is true that the lack of the internal processor will limit updates to already existing games, the absence of camera lag ensures Project Natal will be as revolutionary as it can be for the new games. I can’t wait to play with it.

See you space cowboys…

Project Natal: changes and concerns January 10, 2010

Posted by Cesar in thinking me.
Tags: , , , , , ,
add a comment

Gamesindustry.biz recently announced Microsoft is dropping the internal chip from Project Natal. Even though there’s some talk about the move being related to making updates easier, it is clearly a move to drop costs. It is understandable, as the device must ship at a reasonable price to succeed.

However, from my previous experiences with computer vision solutions, this might represent a big drawback. Natal is not supposed to be a simple toy, targeting unique, exquisite experiences. It is supposed to act as a reliable replacement for controllers in many ways. Computer vision algorithms, like the ones necessary to process both texture and range data the sensors provide, require significant processing, specially if reliability and response time are big issues (as it is clearly the case).

That means not only old games will not get updates (because there won’t be a processor budget to spare for the vision algorithms), but also that new games using the technology will have to reduce processing somewhere else in order to make the system as responsive and reliable as it has to be.

Not all is lost though and I still have hope. I don’t think Microsoft would make this decision without some confidence most of the appeal will still be there. What I can say is that the presence of the range sensors (as opposed to a simple camera) means a lot of the algorithms can be much simpler than in texture only solutions. Tasks like background subtraction, for example, are almost free when range data is available. And pose detection, be that of the head, hands or the whole body, is also simplified, since the range information makes things less ambiguous.

But there’s also no question a dedicated chip would make the impact bigger and increase the usability of the device. Developers will now have to decide between keeping Natal functionality or improving AI and other gameplay areas.

In a related note, the article also mentions Microsoft is struggling with a 100 ms delay in the system. That’s a very common issue with imaging devices: just turn on your webcam and notice the lag. In a way that’s an even more serious problem than the drop of the chip, since a hardware delay cannot be fixed by software optimization alone, meaning no matter how simple the game is, the delay will still be there. But, again, there’s hope. The fact that it is a well known problem that MS is working to solve means it will probably not be there when Natal is released by the end of the year. Or at least that’s what I hope.

See you space cowboys…

Less buttons, more immersion November 23, 2009

Posted by Cesar in gaming me, thinking me.
Tags: , , , , , , , , , , , , , , , ,
4 comments

Immersion is one of the greatest goals every game tries to achieve. It is more obvious in what I call character based games. When you play Gears of War, you feel you are in Marcus Fenix’s shoes. When you play MGS4 you feel the drama of the dying hero as if it were your own. In Modern Warfare 2, every drop of blood on the screen makes you worry and get cover.

But it is also true for impersonal games; that is, the ones whose focus is not on characters. It is easier to notice it when I replace the word immersion for involvement. When playing Lumines or Tower Bloxx for example, you feel involved in the game: frustrated after a mistake, excited when a new level is reached, defiant when the score of a friend is beaten. Note impersonal and casual games are not the same thing at all. Chess is impersonal as are, heroes aside, most RTSs: replace one Zealot for another, there’s no difference at all. All sports games are like that too, with the exception of modes like EA’s Be A Pro.

Now, there’s a multitude of factors that contribute to how immersive a game is. I recently mentioned one talking about Demon’s Souls: challenge. A challenged player is an involved player. There’s also the connection between player and game at the fundamental design level: some people like puzzles, some like shooters. I will not get in the merits of each one, but if you like a type of game, you will be more into it. Style also plays a big role: when graphics and sound suit the game and your mood, they also improve immersion. In fact, I would say immersion is the reason behind most graphical updates in the games industry. From polygon count and texture size to shaders, in the search for immersion graphics chips are always evolving to provide better and more artistic or realistic visuals.

From here we get to the other facet of immersion I want to talk about: controls. No matter what game you are playing, connecting to it requires controlling it without trouble. You only get into Tetris when you learn how to move and rotate the pieces the way you want, you only appreciate Geometry Wars after getting used to moving the ship with the analog stick, you only feel like Sam Fisher when pulling all his moves gets easy, you only enjoy Fifa 10 when passing, dribbling and shooting becomes second nature.

Over the years, games became more and more complex. And with game complexity came complex controls. The Atari had 1 button. The NES had 2. The Genesis had 3. In the current generation, both PS3 and 360 have 4 face buttons, 2 shoulder buttons, 2 shoulder triggers and 2 clickable analog sticks. And I am not counting dpads, select and start buttons.

Most gamers are used to it. Hell, controllers could have more buttons and that wouldn’t be a problem, not to me. But with the last generation of consoles, we saw a big move in the opposite direction coming from Nintendo. The Wii has less buttons and makes up for it trying to detect something everybody knows how to do: move and point. When I first heard about it, I was very interested, both as a gamer and as a robotic perception researcher. To me, that meant games would become even more immersive, shooters would feel even more realistic.

It is curious that while it is true the Wii controllers increased immersion, that change did not affect most hardcore games. Nintendo correctly (from a business perspective) focused on using the more approachable controls to bring a new crowd to the video games era. And it worked very well for them.

Nintendo’s approach was so right that others have been following it ever since. After the success of the Wii, many game platforms started exploring new and more natural input methods. Touch screens and accelerometers became very popular.

But we will soon reach a new apex. Something I personally have been waiting for since I started studying computer vision. And Microsoft is the one about to pull it off: no controls. No buttons at all. If you haven’t heard of Project Natal before, go check it out. It is awesome.

The idea of a vision system in games is not new, the PS2 had the EyeToy. But there were many technical limitations: from sensor capability (one still eye won’t give proper perception of depth for example) to processing power, as robust computer vision algorithms require a whole lot of processing. Project Natal solves these problems in a very interesting way: a single camera is used for “texture” detection. And instead of stereo vision, they achieve 3D perception with a depth sensor. As for the processing power, Project Natal’s device features a custom processor, which is certainly there to reduce the load on the 360 hardware.

Like Nintendo did, Project Natal’s first efforts will probably aim the casual market and bring more gamers to the table. But that does not change the fact that immersion in video games will take a big leap. Imagine playing Lumines by grabbing the blocks and rotating them with your hands. Or simply using your empty hands to select your playlist. Wouldn’t it be cool? Heck, in Minority Report Tom Cruise needed cool glowing gloves to do what we are about to get with our bare hands. The future is here, my friends.

Anyway, as the number of buttons get close to the limit, after all we only have 10 fingers, new input methods are here to stay. I don’t have a Wii. But I will need a bigger living room when Project Natal becomes available.

See you space cowboys…

%d bloggers like this: