By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Microsoft - Kinect Question: what happened to scanning items in?

greenmedic88 on 22 July 2010
thx1139 said:
Reasonable said:
thx1139 said:


What gave anyone the idea you could scan anything and the game/software would react to any old object.  I always took that hey I can scan my own skateboard in a skateboard game and the skateboard in the game will look like mine and nothing else.  Using the scanner or a 3d barcode like the one in this thread shows that scanning can be done.  All this talk about Kinect cant do X that it may have showed in concept videos is BS if thier isnt software now that does it.  One the thing hasnt launched yet and two even at launch you arent going to see full capabilities.  At launch of a piece of hardware when have we ever seen full capabilities.


I'm assuming from the OP he's referring to some early marketing videos MS showed which featured scanning of object and immediate use of that scanned object in a game environment.

I always took this as marketing hype as clearly, unless you hook up Kinect to something like LBP engine you're going to be restricted to pre-configured content for games and while some scanning will be supported it's unlikely to be anything like that shown in the videos.

I assume that as well.  I watched the video. The kid scans a skateboard and then a skateboard that looks like his is useable in game.  My assumption is that it was a skateboarding game and the software simply skinned the in game skateboard to look like his.  So maybe sometime in the future a skateboard game will do just this.  Honestly something simple like a flat board on a skateboard should be perfectly doable.

That should technically be pretty easy to code since it's already been done in LBP with a single lens camera to create in game "sticker" textures. But if it's just textures or creating flat plane (2D) images, there isn't any need for or benefit of 3D object scanning even if Kinect can physically do this on a practical, working level.

I'm more or less willing to write the whole "skateboarding" video segment off as potential concept application rather than something that was actively being incorporated into a game.

For starters, when you think about it, why would you need to "scan in" your skateboard, if the game already had the related resources (animation cycles, rigs, etc.) to make the on-screen character animate like a skateboarder? Presumably it would be for a skateboarding game, meaning the game already has 3D model/resources for a variety of skateboards and wouldn't need to "scan in" your personal skateboard in order to play.



Around the Network
greenmedic88 on 22 July 2010
Porcupine_I said:
greenmedic88 said:
Wikkid666 said:

It's definitely happening... Kinectimals does it.  But you have to buy the stuffed animals.  Buy a new one... scan it into the game then you have a virtual pet.


It's not the same at all.

Kinectimals is the equivalent of buying "Eye of Judgement" playing cards (only one at a time and with each card being sold with a stuffed animal) and having the camera scan in the data code on the card to activate the in-game character on screen.

The stuffed animals are not only unnecessary (you just need the card) they have nothing to do with the way Kinect "sees" what it's supposed to activate in game.

I wouldn't be completely surprised if you could just print out a copy of the card itself and use that instead of one bought with a toy, assuming the cards don't use any special inks (IR based or otherwise).

It's not anything like holding a stuffed animal "Cookie Monster" in front of Kinect and seeing it magically appear in Kinectimals.

i believe the stuffed animals need to be bought per se, as i understand it they are part of a advertising campaign by Microsoft together with burger King. 

I'm pretty sure this is the case as well (stuffed animals for Kinectimals are actually pretty spot on as a promotional item), but it's a really bad example to show that 3D object scanning works with Kinect because it's not scanning anything but a 2D code on a card.



ArnoldRimmer on 22 July 2010

Scanning real 3D objects in is something we're most probably not going to see, or only in some very simplified way.

First of all, you would have to scan the object from all sides to get a proper 3D representation. Doing that precisely in a simple, consumer-friendly way would be rather tricky.

But that's notthe main problem. Because the much bigger problem is that the accuracy of Kinect's distance sensor is in the range of several centimeters. That's perfectly fine for example to identify a human body standing in front of a wall or in the middle of a room. Kinect should also be capable of creating a rather accurate 3D representation of a person standing in front of it, because a human is a rather big object, and since human bodies are rather similar in shape, a small number of parameters (height, length of legs, approximate abdominal girth etc.) are enough to rather accurately describe a human body. So if the system knows that it's supposed to scan a human body, it only has to measure a few parameters.

But if Kinect would try to scan a 3D object which it doesn't know anything about and which may even be rather small, being only a few centimeters accurate is a major problem for creating an accurate 3D representation.

But as others have already pointed out, scanning a 2D image (for example the design of a real skateboard, to use it as a texture on an in-game skateboard, or scanning a picture you painted, or scanning a 2D barcode) will of course work fine. But that's simply not very amazing, that's possible with every webcam and has already been used with the Playstation Eye a long time ago, so Microsoft will not advertise this as some kind of interesting feature.