Like many of you, I’ve often relied on Google Image Search when I need to quickly find an image. It’s simple – you type in what you’re looking for, and you get results that match your keyword. It’s always done the job, so why look elsewhere? Well, I recently came across GazoPa, a new website dedicated to image search that provides some pretty fantastic features when it comes to finding specific (or non-specific, we’ll get in to that) images. Instead of relying on keywords to get search results, GazoPa can take a sample image and return results that match things like the color, shape, and layout of the original.
GazoPa’s main attraction is the ability to find images by similarities, not just by keyword. This opens all kinds of possibilities. Since “similar” can mean so many different things, GazoPa lets you choose exactly what you mean. For example, when it comes to desktop wallpaper, I tend to prefer abstract or fractal images, usually on the darker side of the color spectrum. With GazoPa, I can upload a sample of the type of wallpaper I enjoy, and it will search for other images that have the same color scheme, shape pattern, etc.
Match by Color
When I upload an image to GazoPa, I can choose to get images results that match my source image’s color scheme. So if I use this image…
… GazoPa will return pictures that match the color scheme, more or less, regardless of the actual content of the image.
Match by Shape
If i didn’t care at all about the color scheme, and just wanted something of a similar shape, GazoPa will let me do that too. That same source image from the section above will give entirely different results if I tell it to sort by shape instead of color. GazoPa sees the somewhat starbust-like pattern of my source image, and gives new results that match that fairly well. Not too surprisingly, flowers seem to be a major theme for that shape.
Match by Facial Features
This was one of the things I was most curious about. How well could GazoPa match faces? In my tests, I wasn’t as impressed with that as I was with the Color and Shape searches. I first tried matching a picture of Raul Julia, and ended up with lots of Harry Truman. Then I tried Tom Waits, and got the results below.
It found more Tom, but only the images nearly identical to my source. All others are different people. On the plus side, all the faces returned are male, so it does seem to be doing an ok job, just not perfect.
Match by Layout
Somewhat similar to the Shape option, you can choose to match images by layout. In this example, I used a landscape image with grass, mountains, and sky. GazoPa did a pretty good job of finding similar images.
One of the most unusual features of GazoPa is that you can edit the image directly from the website. It’s certainly no Photoshop, so don’t expect to make any particularly beautiful changes. To me, this seems most useful when searching by color. I can upload a photo with a color scheme I like, and choose to add additional colors without having to open a desktop editor, insert the color, and re-upload.
Here, I’ve taken a wallpaper I like and chose to add some red to the palette so that GazoPa would include red in the results.
I could easily replace the red with blue, green, or anything else to see what comes up.
Combining all of the above with standard keyword search and size filters, GazoPa seem like a great way to find images by style. For specific images of specific people or things, I’ll probably still use Google Image Search, but for wallpapers and other images where style is more important than substance, GazoPa is my new home. It gives me flexibility that Google can’t match and allows me to discover things I never would have found otherwise. It may not replace my old standby, but compliments it wonderfully. As the image database grows, and search algorithms improve, things will only get better.