Google+新添杀手级应用：自动图片编辑

    我们中的大多数人都不可能成为摄影大师安塞尔•亚当斯。不过，谷歌（Google）却打赌其算法能让我们的照片具备大师风范。

    周三，谷歌宣布对社交服务Google+进行升级，增加了包括自动美化在内的多项照片处理功能。谷歌使用了某种机器学习算法，此前，该算法被应用于语音识别和自动翻译等领域。这项新功能非常强大，可以调整照片的光照和对比度，还能柔化皮肤、消除低光照片中的噪点以及去除红眼。

    Google+工程部副总裁戴夫•拜斯布里斯表示，“我们并不想把你的照片变成艺术品。我们只是让它变得更漂亮。”

    很多软件都有类似的图片美化功能，例如大名鼎鼎的苹果（Apple）iPhoto。不过，谷歌表示自家软件能自动处理用户上传至Google+中的所有照片，这倒免去了用户亲自动手。谷歌还声称自己的算法超凡脱俗。“我们所做的东西在这个世界上只有极少数人能做到，我们的专业性无可匹敌。”Google+产品开发部副总裁布拉德利•霍洛维茨如是表示。评测者和用户届时可以验证下谷歌的王婆卖瓜究竟是否属实。

    谷歌的照片服务不光有图片增强，它还能从用户上传的大量照片中自动挑选出最佳作品。所以，你不用再绞尽脑汁地从“亚洲之旅”或“乡村风光”中挑出20大靓照了。该服务会自动去除那些模糊或曝光不佳的照片，然后从其余看似相差无几的照片中挑出它认为最棒的。当然，自动识别地标是必须的，所以你不用担心雄伟的珠穆朗玛峰会被错过。谷歌还能识别出与你关系密切的人，例如你的孩子或女友，并确保他们出现在最精彩的照片中。你可以根据自己的需求随时调整结果，聪明的谷歌会在后台自动学习。

    谷歌还额外奉送了一些功能，并且毫不谦虚的将它们命名为“超棒”（Awesome）。如果软件察觉到一系列相似的连拍照片，将自动合成动画片，把每张图像作为一帧画面。该软件还会根据相邻的风景图片自动创建全景图像。此外，如果你给一群人拍了多张合照，该软件将把各张照片中的人脸进行混搭，尽量拼凑出一张大家都面带微笑、看着镜头的照片。霍洛维茨说道，“你冲洗底片的暗室现在成了一个数据中心。”

    周三，在于旧金山举行的谷歌I/O开发者大会上，谷歌宣布了对Google+进行一系列改进，其中包括将全尺寸照片的免费存储空间增加两倍，达到15GB，还包括经过改进、具有多栏设计的新闻流，以及新版视频群聊产品Hangouts，使人际沟通更加容易。

    Most of us will never be Ansel Adams. But Google is betting that its algorithms can enhance our snapshots with the finesse of a darkroom pro.

    On Wednesday, the company unveiled a set of improvements to the photo capabilities of Google+, its social network, that include the automatic refinishing of images. Using the same kind of machine learning algorithms it developed for speech recognition and automatic translation, Google (GOOG) will tweak your photographs' light distribution and contrast; it will soften skin tones, reduce noise from low-light photos and remove red eye.

    "We are not trying to make art out of your pictures," says Dave Besbris, vice president of engineering at Google+. "We are trying to make them look better."

    While other software packages, notably Apple's (AAPL) iPhoto, have image-enhancing features, Google says its new capabilities will be applied automatically to all photographs uploaded to Google+, removing a lot of tedious work. It also claims its algorithms deliver results that others can't match. "We're on the cusp of taking things that very very few people can do and doing them almost as well as professional," says Bradley Horowitz, a vice president of product development at Google+. Reviewers and users will have to decide whether such claims live up to scrutiny.

    In addition to enhancing images, Google's new photo service will automatically select the best pictures from a mass upload so you don't have to manually find the top 20 pictures from your trip to Asia or your day at the county fair. It removes blurry images and those with poor exposure, and picks what it think is the best among duplicates with slight variations. It can also recognize landmarks, so your sub-par shot of Mt. Everest won't get lost. And it will learn to recognize your affinity for certain people -- say, your children or your girlfriend -- and make sure they end up in the highlighted shots. Of course, you'll be able to adjust the results according to your own preferences, and Google will learn from that.

    Google is throwing in a handful of other goodies that it is immodestly calling Awesome. If its software notices a cluster of similar photos taken in rapid succession, it will create an animation, using each image as a frame. It will automatically create panoramas out of photos of adjacent landscapes. And then there's this: If you take multiple photos of a group of people, it will do its best to compose one where everyone is smiling and looking at the camera, by mixing and matching faces from the different shots. "Your darkroom is now a data center," says Horowitz.

    The improvements to the service, which also include tripling to 15GB the free storage for full-sized photos, is part of a set of changes to Google+ that the company is announcing at its I/O conference for developers in San Francisco on Wednesday. The include a revamped news stream with a multi-column design and a new version of Hangouts, which makes it easier to communicate among people.