MsgId: *emedia(8)
Date: Mon Dec 29 23:12:26 EST 1997
From: DaveThomer At: 209.109.224.165
Tonight E-Media welcomes Robert Silvers, founder of Runaway Technology and creator of "PHOTOMOSAICS" (Owl Books). Photomosaics are collections of thousands of small photographs that form a larger image, such as a group of Civil War photographs that form a portrait of Lincoln. Silvers created the computer program that assembles these mosaics, fusing technology and art into one creative process.
MsgId: *emedia(9)
Date: Mon Dec 29 23:58:52 EST 1997
From: RobertSilvers At: 18.85.12.202
Hello. I am here.
MsgId: *emedia(10)
Date: Tue Dec 30 00:01:12 EST 1997
From: DaveThomer At: 209.109.224.165
Hello, Robert. It's great to have you here tonight. A quick note to those in the audience: the chat room will open at around 9:30 tonight, at which time you'll be able to post your questions. Remember to sign your post so we can identify you.
MsgId: *emedia(11)
Date: Tue Dec 30 00:01:36 EST 1997
From: RobertSilvers At: 18.85.12.202
Thanks for inviting me.
MsgId: *emedia(12)
Date: Tue Dec 30 00:02:06 EST 1997
From: DaveThomer At: 209.109.224.165
Robert, for those people who may not be familiar with your work, how would you describe a photomosaic?
MsgId: *emedia(13)
Date: Tue Dec 30 00:03:57 EST 1997
From: RobertSilvers At: 18.85.12.202
A Photomosaic is a mosaic made from photographs instead of solid colored tiles. Photomosaic(tm) -- the way that I do it uses a bunch of proprietary techniques that I developed both as a student at the MIT Media Lab and also at my company Runaway Technology.
MsgId: *emedia(14)
Date: Tue Dec 30 00:06:10 EST 1997
From: DaveThomer At: 209.109.224.165
Could you describe the process by which a mosaic gets put together? How do you make sure that the individual photos will blend together into a larger image?
MsgId: *emedia(15)
Date: Tue Dec 30 00:09:11 EST 1997
From: RobertSilvers At: 18.85.12.202
The way to start a Photomosaic is to first have an idea of what you are trying to say, and then find content to make it from. This is one of the hardest parts -- getting thousands of images that you have permission to use. What my software then does is takes thousands of images and compares each one with each small region of an image that I am trying to represent. The photo that looks the most like each region of the larger image is used in that location. When it is done, nothing from the reference image is left -- it has been replaced piece by piece with small photographs. When viewed from a distance, the final mosaic looks like the reference image but it is no longer there.
MsgId: *emedia(16)
Date: Tue Dec 30 00:12:08 EST 1997
From: DaveThomer At: 209.109.224.165
What is the most interesting part of the process for you? Finding the representational image, the component images, or working on the program itself?
MsgId: *emedia(17)
Date: Tue Dec 30 00:16:10 EST 1997
From: RobertSilvers At: 18.85.12.202
I have spent the last year and a half perfecting the software and now I am focusing on making images. The most interesting part is when great companies call me with ideas for images. I mean, I have ideas for images, but they bring me new content that I did not have access to before -- such as frames from movies. I worked with Lucasfilm, for example, on an image of Darth Vader and Yoda made from scenes from the Star Wars trilogy. These are now in my book Photomosaics and also for sale as posters on my web site http://www.photomosaic.com This is an example of why this is exciting. Every week something comes along that is very new for me and it keeps me going.
MsgId: *emedia(18)
Date: Tue Dec 30 00:16:19 EST 1997
From: DaveThomer At: 209.109.224.165
Actually, that raises another question in my mind . . . how much of the work is done by the program itself, and how much gets done by the operator once you've assembled your images?
MsgId: *emedia(20)
Date: Tue Dec 30 00:22:54 EST 1997
From: RobertSilvers At: 18.85.12.202
Since I am very specific on how I like things to look, I designed the software to keep me in the loop. There are many ways for me to control the outcome. The first most basic way is controlled by what images I give the program to work with. I know that a certain reference image (the image that the software tries to recreate) will give certain results. So, I may try ten different versions of the reference image. Once I like the way it is basically coming out, I have tools that allow me to replace individual tiles.There are two reasons to replace tiles - either they look bad from a distance, or close up. They may look bad from a distance because the color is too strong and they may stand out to the eye. They may look bad up close because perhaps the content of the given small tile is wrong. I can click on any tile in the near-complete Photomosaic and have my software suggest the top 30 alternatives. I then settle on the one that seems best. It is common for me to make a lot of changes like this.
MsgId: *emedia(19)
Date: Tue Dec 30 00:18:27 EST 1997
From: DaveThomer At: 209.109.224.165
How long does it take to put an image together once you have the idea, or someone brings an idea to you? What are the challenges and potential obstacles?
MsgId: *emedia(23)
Date: Tue Dec 30 00:25:17 EST 1997
From: RobertSilvers At: 18.85.12.202
It takes me from a few days to a few weeks to put an image together once we know the idea. The challenge is gathering thousands of photos. The more I have, the better the result. The challenge is when someone wants an image and they do not have enough photos. Sometimes I have to make the mosaic anyway.
MsgId: *emedia(22)
Date: Tue Dec 30 00:25:12 EST 1997
From: DaveThomer At: 209.109.224.165
What are the factors that the software computes in selecting tiles? I'd assume color and brightness would the big ones . . . are there any others? Do you look for anything special in selecting the tile photos, or is it just a matter of finding a large enough quantity?
MsgId: *emedia(25)
Date: Tue Dec 30 00:28:22 EST 1997
From: RobertSilvers At: 18.85.12.202
The software considers hundreds of aspects of every photo when selecting tiles. Not just color and brightness but also shapes and texture. This is very important because the fact that it considers shapes allows each photo to be worth more than a pixel. This allows Photomosaic to have fewer tiles than you would expect. I can make a recognizable face of Abe Lincoln with only four photos across by five up and down. This is way less than the number of pixels that Leon Harmon the Bell Labs researcher used to make the face of Lincoln back in the 70s (there was a mention of this in Omni back then).Also, if you look at the Marilyn Monroe cover that I made from LIFE Magazine's 60th Anniversary issue you can see a sparkle in her eye. This sparkle was not an entire tile but rather just a light spot in an existing LIFE cover. My software found that and it greatly enhanced the result. Also, because shapes are considered the subject of the mosaic transcends the boundaries of the tiles, for example, Marilyn does not look blocky but rather "floats" behind the grid.
I look for special photographs because if they are related to the subject that I am forming, the result takes on more meaning. I made a portrait of Bill Gates of money from around the world. I made Abe Lincoln from Civil War photographs. The correlation does not have to be direct but can also be ironic. I prefer to find the largest group of related images possible.
MsgId: *emedia(28)
Date: Tue Dec 30 00:33:23 EST 1997
From: DaveThomer At: 209.109.224.165
What fascinates me about photomosaics -- besides the fact that they just look really cool -- is the way that you integrated the creativity of computer programming with the creative process of making images. How did you first imagine that computers could help you create these images? And how did people first respond to the idea, before you had any mosaics to show them?
MsgId: *emedia(29)
Date: Tue Dec 30 00:38:11 EST 1997
From: RobertSilvers At: 18.85.12.202
We can go way back to the fact that I always was interested in computers as a means of creative expression rather than number crunching. This seems obvious now but way back when I was forced to be good at math and stuff. In 1991 I saw a portrait of a man made from sea shells by Dr. Ken Knowlton. Ken did other things like made portraits of people out of physical objects and complete sets of things like Dominos (tm). Being a photographer (I am both a photographer and computer programmer), I though of doing something similar but with photographs.The time was not right for me then, but at MIT many years later I had access to 400,000 photos and I need a project for a class called "Digital Image Processing for Hard copy" taught by professor Mike Bove. This is when I gave it a go. I played around with it but recognized that it had commercial potential so decided to make it my master's Thesis to further develop it.
There were many projects on my plate but I made myself focus. I knew I wanted to start a company when I graduated and thought this could be it. Professor Mike Hawley at the Media Lab was my advisor and guided me through this and also helped me turn a collection of the resulting images into the book - of which he wrote the introduction.
People responded well with the idea before I showed them the mosaics. They responded with ideas when I showed them the first mosaics - ideas for better software and also ideas for images they would like to see. It was a long process to get it where it is today. I mean, it takes a long time to make something look simple. This is true for industrial design as well. I guess the best example is the wheel.
MsgId: *emedia(32)
Date: Tue Dec 30 00:43:22 EST 1997
From: DaveThomer At: 209.109.224.165
Have you given any thought to what you'd like to do next? Are you going to focus on photomosaics, or are you looking for more ways to use computers creatively?
MsgId: *emedia(34)
Date: Tue Dec 30 00:51:10 EST 1997
From: RobertSilvers At: 18.85.12.202
I will run with Photomosaics for several more years. I have plenty of other ideas and I think 3-5 years is the most I would want to do anything. Photomosaics will get more sophisticated as far as what I choose to create for images. At first I was sketching - learning what the tools can do. Now that I know I will be very specific. I am lucky that many companies are calling me and offering content and ideas for images. I will pick the best ones and make them happen.The Mastercard ad - Washington made from 5,000 Mastercards was a big one for me. The LIFE and WIRED covers were very exciting. Like these, I have a few more to go that I can't wait until they are printed -- and by the time my next book comes out (next year), you will see that I am starting to get much more focused on the art aspect.
MsgId: *emedia(36)
Date: Tue Dec 30 00:54:05 EST 1997
From: DaveThomer At: 209.109.224.165
Can you give us a hint as to what some of those future images will be?
MsgId: *emedia(37)
Date: Tue Dec 30 00:58:14 EST 1997
From: RobertSilvers At: 18.85.12.202
I can't say too much only because the companies paying for it like them to be a surprise. I will be part of a famous ad campaign and also will be doing some more posters - some of the posters will be very highly publicized. Also I am working with two difference companies on millennium celebrations. The two companies are my two first choices - I am not sure which party I will go to when the century turns!
MsgId: *emedia(38)
Date: Tue Dec 30 00:59:26 EST 1997
From: DaveThomer At: 209.109.224.165
I just remembered something that struck me when I checked out the web site. I had seen the Abe Lincoln portrait before, but watching it download slowly on my monitor, it was amazing how all of a sudden Lincoln just appeared out of nowhere from the mosaic. Have you thought about how you might be able to make more viewers experience effects like that?
MsgId: *emedia(39)
Date: Tue Dec 30 01:02:57 EST 1997
From: RobertSilvers At: 18.85.12.202
I try to surprise people by making the overall image a challenge to see. The way to do this is to use fewer tiles. When people see a face form out of a handful of photos they can be very surprised. When they learn that I do not alter the photos - and that I simply picked them and arranged them - they are more surprised still. This is the effect that I try for. Once in a while an advertiser will request that I make the resulting image easy to see, hence the problem with commercial work. Design goals are not always p to the artist.
MsgId: *emedia(40)
Date: Tue Dec 30 01:05:14 EST 1997
From: DaveThomer At: 209.109.224.165
Unfortunately, we're just about out of time for this evening. Any final thoughts?
MsgId: *emedia(41)
Date: Tue Dec 30 01:11:44 EST 1997
From: RobertSilvers At: 18.85.12.202
I think a lot of people are getting creative satisfaction from computers - like you can from building castles from blocks or cities from Lego. Art and technology go together well - and always have. When photography was first developed "artists" did not take it seriously because you were "just capturing reality." When pointillism was developed it was called a visual gimmick. Photomosaics is just a tool that I am developing and it is part science and part art. As it gets further developed the art aspect will take over more and more but I hope that history will show it as something that was really figured out when computers became powerful and digital images became plentiful -- and I hope that it leads to a new creative outlet for lots of people at some point in the near future.
MsgId: *emedia(42)
Date: Tue Dec 30 01:14:53 EST 1997
From: DaveThomer At: 209.109.224.165
You're certainly off to a great start . . . best of luck to you in the future. That's it for this week's E-Media; I'd like to thank Robert Silvers for being here tonight, and everyone in the audience as well. This is Dave Thomer saying good night, and be sure to check out www.omnimag.com for the complete schedule of this week's prime time programming and much, much more. We'll see you next week!
MsgId: *emedia(43)
Date: Tue Dec 30 01:15:41 EST 1997
From: RobertSilvers At: 18.85.12.202
Thanks Dave.
Home || Prime Time || Live Science || Machine Dreams || Project Open Book || SF-Fantasy-Horror
Continuum || Antimatter || Mind-Brain Lab || Interactive IQ || Gallery || OMNI ToonsQuestions, comments and suggestions can be mailed to the webmaster.
Copyright © 1998 by Omni Publications International, Ltd. All Rights Reserved.