"Click Here to Upload Image" Building a Website around Contributed Photos
For millions of people around the world, the digital camera has equaled, or even supplanted, the keyboard as the Web communication medium of choice. Not everyone can write. But just about anyone can push a button, and if they don't like what they see on the LCD, they can push it again-at no additional cost. That formula of shoot and reshoot until you get what you want has turned many amateurs into pretty good photographers, producing both interesting and timely shots.
That's what Marc Brown had figured out back in 2003, after snapping pictures with his cell phone of the increasingly famous Coachella Music Festival in Southern California. The pictures were taken on the spot-so there was presumably a lot of interest on the Web. All that was missing was a place on the Web to post them. Brown kept a blog on Blogger, but back then, the blogging service didn't host images directly. So Brown decided to create his own website dedicated to hosting user-contributed images, and Buzznet, which he describes as a cross between a photoblog and a social network-was born.
Buzznet is part of a trend: websites that tap into the millions of pictures people take around the world. Some of these sites amount to online photo albums. A few have become serious scientific resources. And some webmasters have featured contributed images as a section of a larger website. In all cases, the driving force is two-fold: the low cost of digital images and the ease by which they can be uploaded. People are shooting digital images at unimaginable volumes compared with film. They can then sift through their collection, save the few good ones and post them on the Web. And because digital images are already digitized: no scanning is required. Photo software, from PhotoShop to Google's Picasa, can reduce resolution size for fast uploading-which means, on the server side, that an entire database of images can be stored on a single off-the-shelf hard disk. For their part, website designers have put their efforts into simplifying the uploading process for their users, thereby reducing the need for hand-holding. As a result, many of these sites are administered by just a few people, with the contributors themselves doing most of the work.
The idea of running contributed images sounds obvious now, but many Web designers didn't so much seek out the concept, as bump into it. In the case of Buzznet, co-founders Brown and Anthony Blatt had been wrapping up a project--a messaging application created for Pepsi--and were looking for more ways to collaborate. It was only after Brown was inspired by his visit to Coachella that the idea took off and keeps on growing. "Coachella and events like it are great examples of micro-communities," he says. And these micro-communities naturally coalesce around images. Brown argues that while text blogs do this too, photographs are a more universal medium: it's easier to snap a picture of a performance than to describe it. The photographs also inspire comments: on Buzznet, an average of five per photo. That's much more than found in the typical text blog entry. In 2003, Brown was one of the few people with a camera phone. Now, many people have them-and if they happen to be in the right place at the right time, they may become on-the-spot photojournalists.
"When someone comes to Buzznet, they often have a specific purpose: typically putting photos on their blog," says Brown. "But at the same time, people can visit other people's sites, add them as friends, make comments on their photographs, invite people to check out their photographs and, in general, make their presence known in this community. We get a lot of photographs: so just posting alone won't make you visible. All social networks, ours included, require some work. When people find your work compelling, or think that you are cute, or for whatever reason--they add you as a 'friend'". That friends list becomes the right side of the Buzznet page, with the most active members rising to the top.
Superficially, Buzznet resembles Flickr-one of the largest and best known of these sites, which was purchased by Yahoo! "Flickr is a large community, but I'm not sure it is very close-knit," says Brown. "There are so many photos, they get easily buried. And there aren't very many comments." But apart from size, says Brown, approach matters. "Flickr started with the concept of a 'shoebox' of images that you could share--with the idea that the more photos you have to share, the better. That's not really our philosophy. We don't have a tool where you can drag and drop 50 photos into your Buzznet account." The preference, at least for now, is to upload only the best of what you've taken.
Flickr does, however, demonstrate just how large an image-oriented website can grow. The site, which now archives more than 20 million images, is constructed as an online photo album. It's tagline: "people make their photos available to the people who matter to them." Users can specify the privacy setting for each photo they contribute: yourself only; friends and family; or the entire world. Image quality varies, but the sheer size makes browsing fun-especially as more contributors add descriptive "tags" to their images. Or you can just look through recent public photos to get a truly random view of the people, places and things people think worthy of posting. (Clicking through one summer morning brought up pictures of a silhouetted surveillance camera at sunset, a rather large woman on a leather couch holding a baby, a young couple drinking at a bar, an empty garage with a car parked out front, and a dog licking the nose of an ash-covered fireman.) Contributors also have the option of assigning a creative commons license that allows other people to use it.
Furthering science: CalPhotos
A few websites have a more serious purpose. The CalPhotos archive currently holds some 90,000 California natural history images--many of them related to botany--together with an equal number of thumbnails. All of them have been contributed. Every image can be reviewed by an expert, helping to ensure that the scientific name is accurate. The database is widely accessed: as a photo contributor myself, I've gotten queries from high school and graduate students, environmental consultants, and conservation groups. And yet, the entire database is administered by just two people, Ginger Ogle--who created the system--and her colleague, Joyce Gross.
As one of the first online image databases, Cal Photos' history is instructive: Ogle calls the site "a creature of serendipity." CalPhotos began around 1993 as Ogle's Berkeley graduate research project on image databases. After her advisor, the relational database pioneer Michael Stonebreaker, suggested she create a "test bed" database for experiments, Ogle acquired 12,000 photos from the State of California Department Water Resources: pictures of reservoirs, dams, and aqueducts. Ogle's project considered color and patterns as a way of searching for images--research that has proved useful in the area of computer vision.
But that database took on a life of its own-shifting away from water, toward natural history, after it acquired some 12,000 botanical photos taken by Brother Alfred Brousseau, a professor at the nearby St. Mary's College in the 1960s and 70s. Brousseau's colleague, Brother Eric Vogel, helped raise the $10,000 needed to digitize the 35mm slides. Ogle turned Vogel's Kodak CDs into JPEGs and put them online. Once that happened, it inevitably attracted botanists, who had their own images to add to the collection.
To help keep the image flow moving, Ogle and Gross wrote an automatic upload system. "Our goal has been to make it as easy as possible for people to submit, so that we don't get a lot of questions and don't have to do a lot of hand-holding. We get a lot of retired physicians who buy an expensive camera, get interested in botany and want to submit photos. And sometimes, these guys just don't get computers."
The upload system solved one big problem: how to get people who know a lot about plants, but not much about the FTP command line interface, to transport their photos over the Internet. But another problem remained: the people who were good at taking photos were not necessarily experts at plant identification. "Botanists wanted to make sure that there was a way to make the photos more authoritative," Ogle says. She rejected one idea: that experts look over every image to make sure the identification is correct. "There were already tens of thousands of images, and there was no way anybody would be able to do that." Instead, Ogle and Gross developed an annotation system. Experts can register to be a reviewer. Reviewers can comment on the photos and change the identification. The system sends a notification email to the photographer, who is free to disagree.
The system has proved surprisingly effective--providing expert review without hindering the flow of new images. Ogle estimates that about 10 percent of the photos have been reviewed by experts. Change notices are painless and professional. I've had a few of my contributions corrected by specialists. In the peer-reviewed world of science, the annotation system process has made me feel even more like a participant.
Ogle and her colleague, Joyce Gross, review the reviewers. "It's pretty easy to separate out the people that are going to be experts. We ask people about their background and look at their email domain name." A university professor of botany is automatically let in. A Tennessee fly fisherman who knows the name for one insect is not. Ogle says this system "works pretty well," and considering the volume of images and the available funding, "pretty well" is good enough. As for photographers, the criteria is much looser. The main criteria is that the photos be of decent quality and be oriented toward natural history, not a summer vacation.
One problem not encountered by CalPhotos is storage space. Because image resolution is geared strictly for monitors, the entire collection of JPEGs consumes just 70GB. The upload system enforces this resolution by automatically reducing file size. These days, many contributors have figured out how to do the reduction themselves.
Even so, a database of this size and scope seems like it would overwhelm a staff three times the size, especially when you consider that CalPhotos is not the only database they administer. Ogle and Gross save time by sharing code between systems, with each specializing in different areas of science. "I do paleontology and CalPhotos and entomology. Joyce does AmphibiaWeb and the Museum of Vertebrate Zoology. We have a document scanning system that Joyce takes care of. We are both programmers." For development tools, says Ogle, "our motto is to keep it simple. Most of it is in PERL. We've just recently changed to MySQL. We were using Informix for a long time, and before that we used Postgres." Some people have questioned whether MySQL is capable of scaling to a database of this size. "Sure it does, if you're on the Web. MySQL is totally meant for the kind of work we do, which is basically read-only, Web-based databases. We switched about a year and a half ago and our queries are much faster, easier to deal with, we don't have this ridiculous overhead of 15 year old legacy code that does 20 different things."
Ogle and Gross have also kept it simple when it comes to query lookups. While pattern recognition, the point of Ogle's original research, may have applications in robotics, text-based searches have proved far more practical. "We have a lot more metadata than many other sites, more information about the scientific name and the family." says Ogle. "The image databases that I've seen typically will let you search on a key word. Our data is structured and you can query on any of the fields. You can do something like trees in Monterey County, or redwood trees in Monterey County. But we are always looking at other databases to see what they are doing that works really well, and make changes to ours. We steal good ideas wherever we see them."
Wunder Photos: From flooded cars to dragonfly eyes
Contributed images can become the basis for an entire website, or play a supporting role. That's the case with Weather Underground weather forecast site. Weather Underground was the first Internet weather service-first going online in 1995-and is still one of the best, with forecasts and conditions reported for some 60,000 U.
"We originally wanted to grab pictures of cars underwater," says Chuck Prewitt, who heads sales and marketing. The idea: when new photographers want to show the size of a big storm, they shoot pictures of cars floating in lakes or headed down rivers. "We originally weren't confident we would get enough photos, so we contracted with the Associated Press. That was three years ago. But soon, volunteers began outpacing the professionals." It helped that the website was already accustomed to working with volunteers: Weather Underground not only includes official data from the U.
The screening process might seem competitive in this laid-back world of digital photography, but it also may have improved the overall quality. Some of the images are breathtaking, especially the telephoto shots of birds and close-ups of insects--so close that a dragonfly eye appears as faceted as a diamond. Another factor pushing quality is the Wunder Photos thermometer, found on every photo, that allows viewers to assign a rating of 1-10. Some contributors take the system very seriously, writing angry rants when someone appeared to be scoring unfairly low.
Vimeo: videos online
While most sites archive still images, videos can be uploaded almost as easily. That is the specialty of Vimeo, which started as a personal project of its inventor, Jakob Lodwick. Just two years out of college, with a major in information technology and a minor in psychology, he has nonetheless been doing Website development work for nearly ten years. "The idea popped into my head in October 2004. For a year and a half before that I had been making 'vidblogs,' which were different from the video blogging you generally see now. If I went on a trip or to a party, I would shoot clips and interview people--basically documenting my life. The goal was to edit these in Adobe Premier into a short film that I would put on the Web." At the time, Lodwick had just graduated from college and moved from Rochester, New York to San Diego with his two business partners. Their company, Connected Ventures, had launched CollegeHumor.
"After a couple of weeks I started getting a little more ambitious and would spend several weeks on each video--until what were supposed to be little diaries were turning into full-fledged documentary short films." Looking for a broader audience, Lodwick bought a cheap camcorder that records directly to disk. The imported videos, which he called "vidbits," were uploaded to his personal homepage. Lodwick traded the name "vidbits" for Vimeo, and a new website was born: a place where people can upload video clips.
"I try to set as few rules as possible," Lodwick says. The only real rule is that you are only allowed 8MB of disk space a week." That file size can hold about 10 minutes of low-quality video, though most submitted clips are much shorter, averaging about three minutes. Vimeo runs on two Linux servers, and while file size is not inherently a problem, the limit serves as an artistic constraint. In video especially, brevity is a virtue. Beyond that, Lodwick hasn't set any rules on content. "Right now people can do whatever they want and experiment."
I asked Lodwick if he was surprised by the response to his site. "So far, I'm actually a little bit underwhelmed. We think that it has to do with video technology being so new. A lot of people don't own digital video cameras, and those that do I think are pretty intimidated by it." Lodwick says that people don't always have a firm grasp of the technology. He recalls one friend who couldn't upload a 30 second video to Vimeo--because the 40MB file was over the weekly limit. She didn't understand that even a short film could create a large file size if the resolution is high enough.
People also need more narrative skills to shoot a compelling video--it is a trickier business than shooting a still. "The acquisition of skills is steeper, but the payoff is bigger," Lodwick says. "I think video itself is inherently more captivating than text, or just audio, or just still pictures. If you upload a clip that's at all interesting, immediately people will start seeing it and commenting on it. This little community is starting to build up and people are getting to know each other through the site."
In growing Vimeo, Lodwick is a great believer in starting small, putting something up, and revising often. "I used to design sites by putting so much time into these monumental releases, so that it would sometimes take months before I made a change to the site. The approach that I've taken more recently is to release early and release often, knowing that no improvement will be the 'end-all' new version of the site. A website is a living thing; it's never going to be finished. I try to update it when possible, but I'm not trying to be the world's best programmer and I don't take myself too seriously. I'm just trying to make a site that my users enjoy." He says that listening to users--primarily about the problems they encounter, is probably the reason why Vimeo has been able to grow as it has so far.
Lodwick says that the key to building a successful website is to pay attention to what your users want, even if you only have three users to begin with. As the site grows, "you'll have an extremely loyal base of people who are in love with the site." He says it's important to resist the temptation to push some features that he thought were cool, but that his users didn't. "I realized that if my users don't want it, there's no point in doing it." And what users often want is not a fancy new feature, but ease-of-use. "Josh Schacter, the creator of the del.
Sidebar: Some guidelines for accepting user-contributed images
Establish a specialty.
Few websites will achieve the volume or visibility of a general-purpose site like Flickr. But there are plenty of specialty niches that have yet to be fulfilled. CalPhotos, for example, is best known for its California botanical pictures. Perhaps Japanese botanists would like a comparable service.
Figure out whether you want quantity or quality.
For the former, put as few guidelines as possible and don't encourage comments. For the latter, add rating systems and ways for people to comment on the quality of the photos. As a general rule, the more competitive the site, the fewer the images, but the better they look.
Encourage or discourage cellphone images.
Image quality is lower, but cell phones are often in the right place to capture something unusual.
Put your development efforts into making uploading easy.
Great photographers can fumble when it comes to exporting images from their camera to your site.
Start small and innovate as you go.
Most image-oriented Websites have grown slowly by word-of-mouth. This is a good thing. The leisurely pace has meant that the site didn't have to grapple with a lot of data traffic before it was ready.
Think about licensing.
Flickr's use of a voluntary Creative Commons license allows people to contribute their photos. Other sites give visitors the rights to use all photos for their personal use. CalPhotos permits photos for educational and non-commercial uses. In many cases, the photographer maintains at least some control: they retain the original high-resolution image.