Yesterday GameSpot set off a minibomb with an anecdote about Kinect not recognising two dark-skinned users while lighter-skinned gamers had no problems. Consumer Reports has weighed in, saying this sounds like a claim about laptops from last year.
In a post yesterday evening, since updated, GameSpot mentioned that of three employees of colour, Kinect's facial recognition features had trouble with two of them, and none with white employees. This doesn't affect gameplay, where skin colour has no bearing on skeletal recognition. Facial recognition is used in features like automatically logging in and drop-in multiplayer with Xbox Live avatars.
Consumer Reports said it smelled like a rumour about "racist" HP laptops whose facial recognition features also had trouble with darker-skinned individuals. Consumer Reports debunked this claim then, and put Kinect to a similar test now. The magazine says it is "related to low-level lighting and not directly to players' skin colour".
"Like the HP webcam, the Kinect camera needs enough light and contrast to determine features in a person's face before it can perform software recognition and log someone into the game console automatically," Consumer Reports writes. "Essentially, the Kinect recognised both players at light levels typically used in living rooms at night and failed to recognise both players when the lights were turned down lower. So far, we did not experience any instance where one player was recognised and the other wasn't under the same lighting conditions."
It would seem obvious that lighting is the issue here - not skin colour. Earlier this week I tried using a PlayStation Eye to capture my face for use in EA Sports MMA. I was in a poorly lit living room and the result was ghastly. The fact I ended up in blackface was my fault, not the game's. But headlines like "Kinect has problems recognising dark-skinned users?" don't exactly fix attention on the more likely cause of trouble.
Consumer Reports Debunks The 'Racist' Kinect [Consumer Reports]