menu

MIT and Adobe Researchers Developed Image Mimicking Technology


Researchers from MIT alongside colleagues from Adobe and the University of Virginia, have developed an algorithm that could allow you to use a portrait image shot with a particular style and implement this style onto any image you like.

Everybody uses filters these days – it can be those simple ones on Instagram that millions have been using for the past couple of years or the more advanced ones built into Adobe Lightroom or as 3’rd party plugins. A filter is just a simple easy way to change the properties of an image (contrast, brightness etc.) however what if you could change the image more fundamentally and make it look like a different image?

The video above (which has no sound) as it is part of a presentation for the SIGGRAPH 2014 conference shows a demonstration of the new algorithm capabilities to create different “filters” based on different input images. It also demonstrates a short video filtering technique based on the same algorithm.

The original photos (far left) have three styles transferred onto them: (from left) low-key and high contrast, warm and soft lighting, and high-contrasted black-and-white (Credit: MIT-Adobe 5K dataset)

selfie-copycat-1

According to Robert Bailey, senior innovator at Adobe’s Disruptive Innovation Group: “One of the things we’re exploring is remixing of content. You can’t get stylizations that are this strong with [existing] kinds of filters. You can increase the contrast, you can make it look grungy, but you’re not going to fundamentally be able to change the lighting effect on the face”. However, the new technique “can be quite dramatic [and] you can take a photo that has relatively flat lighting and bring out portrait-style pro lighting on it and remap the highlights as well.” Bailey said.

There are still some limitations at this point – the technique seemed to work best when the source and target images where relatively similar — when they’re not, the results might appear bizarre, like the superimposition of wrinkles on a child’s face. But in experiments involving 94 photos yielded consistently good results. Another issue has to do with eyes – apparently ins some cases using the filter can distort the original eye color so the prototype algorithm  offers the user the option of turning that feature off.

You can find more information on the new algorithm on MIT’s website.

More photography related technology videos can be found on our photo-tech section here on LensVid.

(Visited 5 times, 1 visits today)
Go to top
Shares