Printing Photographs from the Web Dot-Com Companies

Chia-Hao (Jack) Yu

Alexan@leland.stanford.edu

 

Introduction

Background Information

As the web becoming more and more accessible to the society, a greater number of regular routines would be shifted to the Internet. Internet not only provides vast amount of information but also delivers convinience to people's life. One great example of that is the "On Line Photograph Printing and Delivering." As the number of people using digital cameras grows, there is a market that converts digital images into physical photographs. One might think that this is a "backward" solution to this digitized age. However, it should not be overlooked that majority of people in the country is still without access to the Internet, and transferring a picture of the new born body, traditional photographs is still the best media.

What these online photographic printing companies offer are the following. A registered user would upload a digital image to the company website, and then he/she could choose different sizes and finishing that he/she wishes the photographs to be printed. Then, the registered user would pay a modest fee (from 49 cents to about 2 dollars depending on the size and promotional offer) and specify the address, to which the pictures would be sent. This means that the user doesn't have to buy a high quality photo printer, and the most amount of saving comes from not having to repeatedly purchasing photographic papers, which we all know is the real money maker.

One interesting observation about the Internet is the "mushroom" effect. Once a market is identified, a flock of companies, like mushrooms, would suddenly appear into the scene. There is no exception with the Web Photograph Printing Companies. There are 6 or more different companies that provide similar service, and with almost identical pricing and promotional structure, a consumer would have a really difficult job to decide which one to pick. Since most companies take the "black box" approach, i.e. taking any digital images and passing them through a complex calibration, signal processing box, then waw-la the pictures! Thus, it would be interesting to compare how different company process different aspects of the printing challenges.

Addressing the Goals of the Project

Since we were interested in comparing outputs from different companies, it was extremely important to develop a set of standard images that we would upload to different websites. The first goal of this project was to generate a set of test images that would address different aspects of human visual system. These images should deal with color, illumination balancing, high spatial frequencies, noise, and edges. Since this was the first time that anyone has done this kind of study, there was no standard set of test images that we could use. Thus, we tried to produce test images that were as simple as possible, and ideally, they would be easily generated and analyzed in Matlab.

After receiving the pictures back from these web companies, we then had to compare them. Deriving some sensible criterions for comparison was the second goal of this project. There were a lot of papers in image quality comparisons, especially in paper printing, digital photographing, and image displaying. However, we found that they were hard to apply to this project. Image quality comparisons imply "how good the picture looks" or "how close is Picture A from Picture B," which were not the questions that we were interested in. We wanted to know "what kinds and how much of differences there were among photographs generated by difference companies." We were not trying to decide which company was better than others, but instead, we were trying to describe the different processing results from each company and let the consumers make up their own mind.

Contents of this Presentation

After the introduction, the format of this presentation would be the following. Each bullet is one individual section.

  1. The base of the soup: test images. Why, what, and how they are chosen.
  2. The opening wine. The scanner parameters and the extraction of pictures for the set of Color Balancing Photos.
  3. The tomato sauce: color processing. Does RGB become RGB or something else?
  4. The cheery in the chocolate cake: red eyes in a little girl and different color balancing. What happens if the pictures' RGB values are adjusted in different proportion?
  5. The Napolean Cake: What happens to patterns with low and high spatial frequency during the development process?
  6. The Chocolate Sprinkle: What happens to pictures with different levels of Noise?