Wednesday, 22 October 2014

The Grid Website Platform Automatically Adapts Design To Make Your Content Shine

Are electrical designers dream websites? Web contents You can know itself? The grid, a startup in San Francisco, they are changing the way websites are designed fundamentally changing the role of the designer. This new platform uses machine learning is combined with design and flow based on constraint-based scheduling to dynamically adapt the form to content. To understand how this revolutionary technology is, you have to compare with how most websites today are made.

Barriers to entry for web design keep getting lower. From WordPress to Squarespace to "no code" website builders drag and drop designer's life seems to be getting easier. Only it is not.

By simplifying all the tools, the complexity of the web environment is growing like kudzu. Desktop, laptop, tablet, phone, clock and eventually 4K TV screens-the number of destination is dizzying. The most common way to deal with this mess of pixels practice is to use templates and design response for the same flow of content (many) different size containers.

This is an effective strategy compared with building websites tailored for each target device. Ethan Marcotte (the designer who invented the term "responsive web design") and Karen McGrange (content strategist who has championed the concept of "adaptive content" to flow sensitive designs) well packaged this concept for the enterprise market . What has always attracted me for the approach response (or RWD) is trying to reduce its assumptions about the user of your site. Like the movement of web accessibility and web standards (led by Jeffrey Zeldman) from which it evolved, RWD implies a Zen detachment pixel perfectionism.

The problem with these approaches is that there are only so many pixels to perfect! It involves a lot of work to conceptualize and test every possible permutation. This sounds like a job for computers!

Dan Tocchini, CEO and co-founder of the network, has a high level of design and code. His team includes Brian Axe, a former product manager of Google AdSense, Henri Bergius, a pioneer in the engineering of open content management code and Leigh Taylor, former lead designer on Medium. In the new platform used responsiveness own code to simulate real process designer. The results as you can see in the picture below, they are very sophisticated.


As with responsive web design, this is the code in service design. Unlike RWD, The Grid figures things difficult for you.

Most of the network coverage so far focuses on the tactic of using artificial intelligence (AI) to create the company web sites. I think this may be only half true, but does not diminish the importance of what is building this team. To be technical, I think that the grid uses a lot of machine learning (ML) for users to perceive the process of design and artificial intelligence.

This difference may seem subtle, but important. The true AI is emerging. A contemporary artificial intelligence system not only perform sophisticated combinations of rules-invents new rules. I have not yet seen any evidence that the "filter design" (as The Red calls its version of content management "themes") are fully capable of spawning new filters. To see what these filters are capable of (which is considerable) we will see for the first time in color.

One of the interesting things you can do with the network is to create a restriction that will generate the color of a text box to go with each photo you upload. On the inside, the grid measures the color data of image and calculates the values for a perfectly tuned panel. The effect is that the photo is combined with a pleasantly box in color with no two being exactly the same. The role of the designer is to establish the restriction. One would choose the average color of the image, or a shadow 50% of that color or a complementary shade of whole color.

This is just one way that the data on the content of the reports that display the content. Another related feature scans the color of an image to identify areas of low contrast. The software can then use this geometry suggest optimal position for the text boxes color-keyed on top of the photo itself. Again, the designer creates the filter and specify what type of data can be controlled display settings. But for the non-technical user, everything is magic.

Take another example that solves one of the most pressing problems and time-consuming for contemporary web designers. For each of these different screens for design, images need different sizes and shapes. And 2x and retinal images 3x now, the image stack gets pretty deep for every bit of content. Finally, there is some commonality in terms of the new html image element, but that only defines what image is loaded in the format of the screen, not what those images are.

The grid solves this problem with some sophisticated algorithms. The not only to crops and resizes all necessary images, solution also recognizes features in an image (eg a face) and these features properly positioned within the frame of each image on the screen. For a simple example, imagine you have an image with a figure standing places a third of the distance from the right edge. A simple linear function would be to keep that proportion in all sizes and shapes of the images. (You might consider this AI rule-based rather than ML because there is no mechanism for adaptation.)

But what if the designer of a page filter observes their behavior and realize they do not always follow strict linear consistency when re-crop images from the web? Instead of a linear rule, the designer could provide some sample image sets as learning training data and the use of teaching machine to make the filter more nuanced adjustments to each image.

When the simulation of the human may be more critical is the own choice of content. If the software can automatically tune the shape with the content, then it is the choice of the content itself that becomes the only generator design. The grid has some automation features in this area also. Press release from the company says that users can "Collect pictures and text throughout the internet through a browser or a mobile extension to incorporate into your site." This is very convenient and will further reduce production time.

There is a possibly harmful content with this automation aspect, however. Part of the pleasure of looking at Tumblr, for example, is trying to imagine the person who posted this very particular content. What if that "person" is an algorithm? By combining web scraping tools and scripts, it is certainly possible to create "content filters" that analyze large amounts of data and select the output highly particularized. This may be where the promise of the Ritual of artificial intelligence kicks in-but also it could be derailed. It will be interesting to see how the balance of ease of flow against identity assurance platform evolves over time.

No comments:

Post a Comment