The non-standard game interfaces discussion thread (move, voice, vitality, etc.)

[upnorthsox]

Hmm, not really accurate though. Should be more like "The Move wand is the controller". Not that catchy though.

It´s quite easy to come up with some witty slogan including the word Move, but I find it a bit hard to find one that speaks equally well to both the casual and the core crowd. You don´t want a slogan that could possibly alienate either one.

Microsofts slogan is very well crafted. It does not scream "gesture gaming", it makes it sound like some sophisticated future tech, bypassing that old limited physical controller thingy.

No doubt Kinect is the easier PR sell, you can even wheel out Steven Spielberg to tell you how it changes gaming forever.

Still, you can do an ad campaign based more around emotions and games. Example:

A father and son walk in to the living room ready to play a game (COD, Madden, whatever kind you expect). There is little Samantha on the floor in front of the TV with the Move.

Dad: Ok pumpkin, we want to play a real game now.
Little Samantha looks back and up: Ok daddy, one more minute.

She then proceeds to take the Move, draws a house, colors it in (with the ball changing colors), reaches over and picks up her virtual EyePet and places him into the house. He curls up in the doorway and lays down, she smiles at him, he smiles back. She then gets up and goes to her dad.

Samantha innocently handing dad the Move: Ok, it's your Move now.

Dad and son look on stunned.

Cut to boilerplate product stuff.

Or anyhow, something like that.

 

[Silent_Buddha]

Then how about facial recognition? Is the resolution high enough for that? I think at E3 they advertised it as one of the features, but if it can't read fingers, can it read features of your face, as they can be even smaller in size

Would facial recognition use only the depth based camera? It's more likely that it would use both camera's. Depth camera to easily isolate the head so that the image processing knows which part of the visual image to process from the full color camera. Perhaps also to identify key locations (eyes, nose, etc) to enhance the accuracy of the image processing of the visual camera. Then the visual camera could be used to pick out finer details.

Regards,
SB

 

[Shifty Geezer]

Depth isn't needed for face recognition. Face detection is pretty robust nowadays. Even cheapish cameras can do multiface detection and smile recognition.

 

[Silent_Buddha]

In which case the low resolution of the depth camera is even less of a factor.

Regards,
SB

 

[Squilliam]

So I take it the performance of the camera was limited because of processing reasons. They couldn't process what they had fast enough to be of any good to games, so therefore they limited the resolution of the camera because any more would be pointless given the consoles performance limitations.

Perhaps they will have to slow down the next generation transition in order for the processing capabilities to catch up. If they need essentially all of the performance of the current Xbox 360 to process a 1280 by 720 video at 60FPS then they could double the performance of the Xbox next over the Xbox 360 and just stand still in terms of image quality.

So instead of needing say 4X performance for the next generation console they would need 5 or 6X to account for the camera overhead on top of anything else they feel needs to be done in the background.

 

[Crossbar]

So I take it the performance of the camera was limited because of processing reasons.

It could also be limitations of the USB connections. Is there any information about the number of bits of the depth information? Quick and robust frame transfers may have been a priority.

I expect the bandwidth of the USB was the reaon why Kinect RGB camera only is 640x480 just like the Playstation Eye. I was really expecting MS to opt for a higher resolution to differentiate Kinect as a HD device, maybe not 720p but at least higher than the VGA resolution. But I guess that will wait until the next generation of Kinect.

 

[manux]

It could also be limitations of the USB connections. Is there any information about the number of bits of the depth information? Quick and robust frame transfers may have been a priority.

I expect the bandwidth of the USB was the reaon why Kinect RGB camera only is 640x480 just like the Playstation Eye. I was really expecting MS to opt for a higher resolution to differentiate Kinect as a HD device, maybe not 720p but at least higher than the VGA resolution. But I guess that will wait until the next generation of Kinect.

Playstation eye is 60fps whereas kinect is 30fps 640x480. Hence there should be enough bandwidth for kinect to push both rgb and depth buffer at 640x480x30fps(same bandwidth as for plain rgb 640x480x60). But isn't kinect using lower resolution for the depth buffer which would imply there is something else going on than pure usb bandwidth.

If bandwidth is the problem instead of computing power it should be fairly trivial to use compression to the data stream. Ofcourse it wouldn't be best possible compression as they wouldn't want to introduce lag but still even some fairly simple compression should make bandwidth requirement a lot more manageable.

 

[Squilliam]

It could also be limitations of the USB connections. Is there any information about the number of bits of the depth information? Quick and robust frame transfers may have been a priority.

I expect the bandwidth of the USB was the reaon why Kinect RGB camera only is 640x480 just like the Playstation Eye. I was really expecting MS to opt for a higher resolution to differentiate Kinect as a HD device, maybe not 720p but at least higher than the VGA resolution. But I guess that will wait until the next generation of Kinect.

I don't think even HD video would saturate a USB 2.0 link, even raw.

According to this: http://web.forret.com/tools/video_fps.asp?width=640&height=480&fps=30&space=raw&depth=8

640 by 480 @ 8bits raw would be 73Mb/S vs 480Mb/S theoretical USB bandwidth.

 

[Shifty Geezer]

Indeed. USB 2.0 is plenty for SD res video. 300k pixels are 5 bytes per pixel (3 bytes RGB and 2 depth) is 1.5MB, x 60 fps is 90 MBs. By comparison PSEye has minimal compression to keep noise down.

I stilll think the low resolution is a sensible compromise. The depth detection seems to run at a quarter resolution of the depth CCD going by the PrimeSense articles. If higher resolution won't gain many benefits in game, there's no point using the higher resolution CCD. Thus they could double up on the order of the same CCD for both colour and depth cameras.

We'll need someone to confirm that the CCDs are identical devices.

 

[Crossbar]

Playstation eye is 60fps whereas kinect is 30fps 640x480. Hence there should be enough bandwidth for kinect to push both rgb and depth buffer at 640x480x30fps(same bandwidth as for plain rgb 640x480x60). But isn't kinect using lower resolution for the depth buffer which would imply there is something else going on than pure usb bandwidth.

If bandwidth is the problem instead of computing power it should be fairly trivial to use compression to the data stream. Ofcourse it wouldn't be best possible compression as they wouldn't want to introduce lag but still even some fairly simple compression should make bandwidth requirement a lot more manageable.

Yes Kinect is using a lower resolution buffer for the depth buffer 320 x 240, if I remember right. So yes that will put less stress on the communication link, there may be some other design considerations behind this.

What I don´t know is if the colour depth of Kinect and PSEye differ. Does anyone have information about this. Is it 24 or 32 bits, higher, lower?

Unless they are using lossless compression, compression will introduce artifacts that will make image processing harder. Lossless compression is fairly complicated and give pretty bad compression rates so I don´t find it likely they use compression if they intend to do some image analysis to complement the depth information.

I don't think even HD video would saturate a USB 2.0 link, even raw.

According to this: http://web.forret.com/tools/video_fp...ce=raw&depth=8

640 by 480 @ 8bits raw would be 73Mb/S vs 480Mb/S theoretical USB bandwidth.

Well theoretical bandwidth is usually just a theoretical bandwidth and Sony and MS have probably chosen some wide safety margin to allow the stream to go via low quality USB hubs sharing other devices etc.

 

[Rolf N]

What I don´t know is if the colour depth of Kinect and PSEye differ. Does anyone have information about this. Is it 24 or 32 bits, higher, lower?

The PSEye gives me 16bit YUV on the PC, which is actually pretty typical for USB cameras. No idea if there's some other special colorspace that only the PS3 can use.

 

[Crossbar]

The PSEye gives me 16bit YUV on the PC, which is actually pretty typical for USB cameras. No idea if there's some other special colorspace that only the PS3 can use.

I seriously doubt the CCD of the PSEye is matching the colour space offered by 16 bit YUV, so I would expect it to use some other format for the actual data transfer, when used together with the PS3.

 

[Arwin]

I was showing off some PS Eye Augmented Reality stuff to a guy interested in doing something with it professionally, and it showed painfully clearly that the PS Eye's webcam's strength is the framerate, not the picture quality. Even with my lights all the way up, the picture looks grainy and a big TV and the contrast with the sharp graphics generated by the PS3 don't help, to be honest.

On the other hand, EyePet looked and played better than I expected. Look forward to trying it with the Move (and hopefully 3D at some point).

 

[Rolf N]

I seriously doubt the CCD of the PSEye is matching the colour space offered by 16 bit YUV, so I would expect it to use some other format for the actual data transfer, when used together with the PS3.

Eh, not sure what the problem would be, but just to clarify: 16 bits per pixel. That's not better (or worse) than RGB888, just easier to shove over the wire.

I was showing off some PS Eye Augmented Reality stuff to a guy interested in doing something with it professionally, and it showed painfully clearly that the PS Eye's webcam's strength is the framerate, not the picture quality. Even with my lights all the way up, the picture looks grainy and a big TV and the contrast with the sharp graphics generated by the PS3 don't help, to be honest.

Depends on framerate. It should be very competitive to other cameras in the price/size range at 30fps. Capturing at higher framerates means less exposure per image, which leads to more noise and needs more light. The 120fps mode only gives a quarter of the resolution, too.

 

[patsu]

The PSEye has 2 modes. When you dial it clockwise (blue LED ?), which mode is it ?

 

[Arwin]

The PSEye has 2 modes. When you dial it clockwise (blue LED ?), which mode is it ?

Zoomed out max (vs Zoomed in max on red)

 

[Crossbar]

Eh, not sure what the problem would be, but just to clarify: 16 bits per pixel. That's not better (or worse) than RGB888, just easier to shove over the wire.

You are absolutely right. I mess-read a sloppy description of YUV where they used a 16 bit coefficient when calculating each channel.

clearly that the PS Eye's webcam's strength is the framerate, not the picture quality.

I also believe a very high dynamic range is one of the PSEyes strengths.

 

[Arwin]

I also believe a very high dynamic range is one of the PSEyes strengths.

Could be. It's decent for a webcam, but it's very bad for a camera, imho.

In other news, Sony is going for a tried and tested advertising strategy in Japan ...

PS Move girls

http://www.youtube.com/watch?v=1S2dbJIrIl4

:lol

 

[patsu]

Could be. It's decent for a webcam, but it's very bad for a camera, imho.

In other news, Sony is going for a tried and tested advertising strategy in Japan ...

in Taiwan ! says the poster's name of the video.

[size=-2]Can they teach me how to play ping pong ?[/size]

EDIT: King Fu Live ad:
http://kotaku.com/5611048/put-yourself-in-a-ps3-game-then-kick-someones-ass

 

[Arwin]

I'm so far not really impressed by this game at all, but it's starting to shape up much better now. I definitely like the way it puts you into the comic book story.

I thought by the way that this game would also come to Kinect, but now I can't find any information on that anymore?

http://kotaku.com/5611048/put-yourself-in-a-ps3-game-then-kick-someones-ass