MacSpeech Dictate Profiles
Looking to improve recognition accuracy in MacSpeech Dictate? When recording, consider that ambient noise levels vary between locations. By creating a tailored profile for each location in which you record with MacSpeech Dictate, you can be assured of receiving optimum speech recognition accuracy.
Visit MacSpeech Dictate
Which of the online photo services offers the best quality prints from digital pictures? Read on for the conclusion of Alex Hoffman's comparison of a number of online photo labs. Taking a break from the intensely practical, Adam looks at some cool concepts, including a free license generator, a research project into number associations, and a test of online advertising's efficacy. In the news, we cover the release of Default Folder X and PayPal's IPO.
by Jeff Carlson
Default Folder X Improves Mac OS X Open/Save Dialogs -- For many years, Macintosh users have enhanced Open and Save dialogs with utilities like Power On Software's Action Files and StShow full article
Default Folder X Improves Mac OS X Open/Save Dialogs -- For many years, Macintosh users have enhanced Open and Save dialogs with utilities like Power On Software's Action Files and St. Clair Software's Default Folder. At long last, one of them has come to Mac OS X in the form of Default Folder X 1.0.1 (see "Tools We Use: Default Folder" in TidBITS-475). Default Folder X provides the same basic functionality as its cousin for earlier versions of the Mac OS, which let you access favorite and recently used folders easily in Open and Save dialogs. You can also rename, get info on, and delete files and folders, and open folders in the Finder. Default Folder X also shows your current location and rebounds to the last item selected in a folder. You access these functions through a toolbar attached to the right side of Open and Save dialogs; keyboard shortcuts are also available. However, the new version also includes a gem that makes it a required addition: you can use the keyboard to navigate Mac OS X's columnar dialogs properly. Unlike in Apple's current incarnation of Open and Save dialogs (which we detailed in "Apple's Dirty Little Secret" in TidBITS-601), typing a folder name with Default Folder installed highlights that folder in the list, instead of putting you in some file hierarchy limbo. Currently, the utility works only in Carbon applications, but an upcoming free update will support Cocoa applications as well. Default Folder X is available as a free 30 day trial, after which registration costs $35; owners of Default Folder 3.x can upgrade for $20. The installer is a 1.4 MB download. [JLC]
PayPal IPO Bucks Trends -- Internet stocks may never return the high-flying days of a few years ago, but Internet transaction service PayPal bucked recent trends with its initial public stock offering last weekShow full article
PayPal IPO Bucks Trends -- Internet stocks may never return the high-flying days of a few years ago, but Internet transaction service PayPal bucked recent trends with its initial public stock offering last week. The stock opened at $13 and closed its first day of trading on the NASDAQ above $20, a 55 percent gain that left the company valued at over $1 billion. Despite that vote of confidence, PayPal is facing a patent infringement lawsuit and regulatory investigations from a number of states concerned that PayPal may be running afoul of banking regulations. PayPal also is by no means profitable, with a $107.8 million loss in 2001 on revenues of $104.8 million. Still, with 12.8 million customers and new ones arriving at a rate of over 18,000 per day, PayPal remains the leader in online transaction services - see "Worthy Web Sites: PayPal" in TidBITS-562). [ACE]
Say what you like about the Internet letting the wackos out of the woodwork, but it also lets people with truly neat ideas have a chance at trying them outShow full article
Say what you like about the Internet letting the wackos out of the woodwork, but it also lets people with truly neat ideas have a chance at trying them out. The dot-com implosion could even be seen as a good thing by turning attention back toward the days when you didn't need a business plan and a cool million in venture capital to do something on the Internet - not that we're bitter because no vulture capitalists ever threw a million bucks our way!
Recolonizing the Commons -- Larry Lessig, Stanford law professor and author of The Future of Ideas, along with Code and Other Laws of Cyberspace, has come up with a humdinger of an idea he calls Creative Commons. The basic idea is to make available flexible, customizable intellectual property licenses to musicians, writers, programmers, artists, and anyone else who wants to distribute their work widely while still defining what constitutes acceptable use. The best part, for anyone who's had an uncontrollable coughing attack after reading a lawyer's bill, is that Creative Commons intends to provide these licenses for free.
Once Creative Commons launches, which is currently planned for May of 2002, anyone with a hankering for some home-cooked license will be able to visit the Creative Commons Web site, choose the options that fit their situation, and, without even putting a quarter in the machine, get a custom license. Actually, it gets better. Every license will contain a special machine-readable tag specifying the terms of that particular license. When that license and the content it covers are indexed by the search engines, that tag will enable searchers to find specific types of content available under particular licensing terms, such as if an editor of an environmental newsletter wanted to browse through nature photos available for non-commercial use.
Another project Creative Commons plans is a "conservancy" to facilitate the preservation and dissemination of intellectual property. From what I can tell, the goal here is similar to a straw man I proposed back in 1998 under the name of the Electronic Phoenix Project - a group that would exist as a target for donation of old code such that it could be brought into the open source world and kept alive after its commercial parent was unable to justify further care and feeding. The Electronic Phoenix Project was doomed from the beginning by virtue of an overwhelming ratio of opinions to actual effort, but perhaps the times have changed. Best of luck to Larry Lessig, his colleagues, and the entire Creative Commons project - may their efforts rethink the morass that has become copyright and intellectual property.
When Numbers Get Serious -- I'm a word person, so for me, intellectual property involves writing, and numbers tend to wash over me without leaving much of a trace. But a new study called The Secret Lives of Numbers has me recalculating my digital relationships. Here's how it works. The authors wrote some custom software, ran a lot of searches through public search engines, and applied a heavy dose of statistics to the results to come up with an interactive exploration of numeric associations. What's a numeric association? The number 17, for instance, is associated with the 17-inch computer monitor, the B-17 Flying Fortress, and the Glock 17.
Notwithstanding what Mark Twain said about there being lies, damn lies, and statistics, the results are fascinating. Load the Java-based interactive visualization and you can enter any number under 100,000 and see what was most commonly associated with that number. A scrolling bar graph shows how popular any given number is with regard to those surrounding it, and you can also click any line in the graph before getting the number's associations. It's well worth a few minutes of your time, both playing around in the Java applet (which works surprisingly well on the Mac in Internet Explorer 5.1 and Opera 5.0, though a bit less so in Netscape 6.2), and thinking about how numbers both rule our lives and reflect how we live them. ZIP codes, area codes, product versions, numeric lists - numbers weave in and out of our lives in innumerable ways.
Bonus Concept -- While we're on the topic of numbers, how many items does a "couple" imply? The obvious answer is two, although in informal speech, couple can grow beyond two, into the realm of "few," which is more than one but not many, or even into "several," which is more than two but not many. I say all this as introduction to my third item in this article, which I wanted to add without screwing up the alliteration in my article's title. (And because it will irritate Geoff, our tecnical editor and resident curmudgeon when it comes to these particular usages.)
Anyway, one place where numbers have gotten very serious, as in seriously low, is in online advertising. For a long time (Internet time, which is generally measured in months), it was believed that online advertising was the second coming for advertising because such wonderful statistics were available. Never mind Mark Twain: these were numbers, and there were so many of them. Such lovely numbers! Even if they didn't want to be privacy-invading weasels, companies publishing ads could determine number of impressions, click-throughs, and all sorts of other fabulous stats. Never before had advertisers been able to add such unblemished numbers to their spreadsheets; suddenly they didn't have to work with such fuzzy concepts as brand awareness, since they knew exactly how many people saw each ad, how many people clicked through, and how many actually purchased something. Numeric nirvana!
But there was a wee problem, which is that some of the numbers kept getting lower and lower. Click-through-rate in particular dropped through the floor because - surprise surprise - using the Web is an active process, and people are generally goal-oriented. Unless a banner ad happens to match exactly with the user's goal, the chance that they're going to click a banner is pretty low, and as the more inured users became to online advertising, the less they clicked through. Never the sorts to avoid soiling their own nests, the most aggressive advertisers like X10 immediately redoubled their efforts, introducing pop-up and pop-under ads so annoying that incidents of computer monitor abuse rose tenfold (personally, pop-up ads tend to make me speak in punctuation - @#$%!! - and hit Command-W harder than is healthy for my keyboard). Pop-up ads were effective initially, since they were novel and very much in your face, and in X10's case it probably didn't hurt that the ad implied its wireless video camera was perfect for capturing surreptitious footage of gorgeous blondes in various states of undress. (Totally fabricated response from X10: "Shocked, I'm just shocked you could think that! The clear message was that normal people concerned about security and privacy could install them, for instance, in their guest bedrooms to make sure their buddy's hot wife doesn't steal the towels after her steamy shower. Umm, and for mothers trying to catch their kids drinking milk straight from the carton. Yeah, that's what we meant!")
But I digress - it's so difficult to avoid helping X10 reap the rewards they so richly deserve for their advertising campaigns. I wanted to talk about brand awareness, which has long been a major goal of traditional advertising. No one sees a car ad on TV and runs right out to buy one - the entire point is that the next time you are ready to buy a car, the cumulative effect of all those ads will make you think, "I should test drive one of those cool new gas-electric hybrid cars from Honda." (Well, not really, since neither Honda nor Toyota seem particularly interested in actually selling their hybrids to anyone who's not willing to order one sight unseen, and advertising them would undoubtedly make it even harder to not sell them.) The problem with brand awareness is that it's hard to track, since you have to do real-world surveys and ask people questions, and you're always going to get people like Tonya, who once responded to a market research question about which painkiller she used with the helpful answer of, "The one in the round bottle."
Advertising actually is good for brand awareness, something we've been saying since we started our sponsorship program back in 1992 (possibly the first instance of advertising in an online publication), and a company called Dynamic Logic has finally proved it. They created a fake brand, a concierge-type service for people with more money than sense (err... time), and promoted it purely by running a few million ads on the Internet with the British Web sites FTMarketWatch and iVillage. When it came time to survey people, they discovered that about 4 percent of people claimed to recognize the brand but had never seen an ad for it (some people will agree to anything). Nearly three times that number, or 11 percent, recognized the brand and had actually seen the ads. And in the brand's target audience of men between 18 and 49 earning more than $120,000 per year, 19 percent of all users remembered the ads. In short, properly targeted, appropriate advertising does work, even without spawning windows all over users' screens.
by Alex Hoffman
After buying a new digital camera and going on my honeymoon, I have more than 1,000 pictures, about 100 of which my wife wants to put in photo albums. Therefore, I decided to investigate which online digital photo lab was the bestShow full article
After buying a new digital camera and going on my honeymoon, I have more than 1,000 pictures, about 100 of which my wife wants to put in photo albums. Therefore, I decided to investigate which online digital photo lab was the best. In part one of this article, I evaluated 11 different services in terms of cost, ease of use, and range of products. When I began this project, I thought I could look at the companies' Web sites, order some samples, and see which services were better than others. That was enough to eliminate a number of the services from the running, although I included almost all of them in the quality tests that came next.
I initially thought the print quality from the different services would be roughly the same. Not only did that not prove true, examining the prints afterward revealed a few flaws that I could have corrected before sending off the files. Pay attention to the areas I failed to take into account, and you'll get better results.
Cropping -- My first mistake was to ignore the issue of cropping. Most digital photographs use an aspect ratio that matches computer monitors and televisions, featuring a 1.33:1 ratio between width and height. Standard photograph sizes, however, don't match that aspect ratio, taking their cue instead from 35mm film, which uses a 1.5:1 ratio. Prints measuring 4" x 6" (1.5:1) and 5" x 7" (1.4:1) are wider and shorter than digital images; 8" x 10" (1.25:1) and 11" x 14" (1.27:1) are narrower and taller than their digital counterparts. Only PhotoAccess offers prints whose ratio matches that of most cameras and monitors and do not have to be cropped, but of course they may not fit properly in traditional photo albums and frames.
There are three solutions to this problem. The first is to resize the picture disproportionately, but that's unacceptable in almost all cases, as it would make people look as though they were reflected in fun-house mirrors. The second is to shrink the photo proportionately, which works fine, but means the image won't completely cover the paper. The picture ends up looking like a letterboxed movie, not using the very top and bottom of the print for narrow sizes, or putting white borders on the left and right for wider sizes. That's the safest option. Third and finally, you can crop the photo, which eliminates some of your image. That may be fine if you're doing the cropping yourself, but it can be disastrous if a service does it automatically and gets it wrong.
Naively, I assumed that the first photo service I used would be smart enough to compensate for the aspect ratio differences. But because the process is automated, the results weren't great. I should have manually cropped each photo. Most sites offer cropping tools, and I also had the option of cropping the photos before uploading them. If done on the services' Web sites or with their uploading software, as they recommend, cropping takes less than 30 seconds for each photo. Had I taken the time to do this, I would have been much happier with my initial prints.
Shutterfly and Apple stand out as having the most versatile cropping options. Customers get total control over what is printed. Other services feature less control, with Ofoto simply offering the option to print extra borders to fix the shape, or to crop the image automatically by keeping the center and cutting off the edges equally. Shutterfly's upload application and Apple's iPhoto let you crop your images to a specific aspect ratio to fit the different photo sizes perfectly. Once a photo is uploaded to Shutterfly, you can change the cropping even if you cropped earlier.
Cropping also affects how large you can have a photo printed. When you remove portions of a photo, you're reducing the image's overall number of pixels. If you crop too much of the image, there may not be enough information to make a picture look good at larger sizes. Cameras with more than two megapixels of resolution can produce decent-quality photographs at sizes up to 8" x 10", but if you crop too much, you may not be able to print at the size you want. Fortunately, all of the services I tested give some kind of warning about which images will print well at what sizes, though some are more obvious than others.
Gamma -- Digital color is tricky to do right. Different monitors display colors differently, as do different cameras, scanners, software applications, and operating systems. This happens because they all have slightly different assumptions about which combinations and intensities of red, green, and blue should be used to represent any given color for each pixel. One aspect of this situation is gamma correction, which controls the overall brightness of an image.
Macs are usually set to a gamma of 1.8, and PCs are set to a gamma of 2.2, which explains why an image created on the Mac will look darker and will have more contrast when viewed in Windows, and images created in Windows may look washed out on the Mac. Since most computers run Windows, photo services seem to try to match their output to a 2.2 gamma setting to provide what most of their customers expect. Unfortunately, that meant almost all of my test prints came back looking darker than I expected, and pictures where the color was just right on my computer weren't as good in print.
Apple's ColorSync technology helps resolve the variations in colors that result when an image is reproduced using different devices, applications, materials, and printing processes (and is used by iPhoto when printing directly from your Mac), but it doesn't help with any of the photo services. I learned from the president of Ofoto that they often try to achieve a similar goal by examining a JPEG image's EXIF (Exchangeable Image File Format) information. Ironically, iPhoto strips out EXIF data whenever you modify an image, preventing Ofoto (which prints pictures for iPhoto) from using this technique when printing from iPhoto.
This area is where Apple has the opportunity to stand out, and part of why Apple was wise to introduce its own photo printing service. Because Apple knows that all the photos it gets are coming from Macs, it alone has the potential to calibrate the output to match typical Macintosh monitors. Unfortunately, this isn't yet the case, although there is hope for the future: Ofoto's president seemed receptive to the idea of applying a common correction to all prints ordered through Apple. Right now, however, the only way to do this seems to be editing the images in a separate application such as Photoshop or GraphicConverter, which realistically is more than what most users want to deal with. And even using Caffeine Software's free (and utterly cool) PixelNhance to edit every photo may be more trouble than most people want to take.
I've spoken with a few of the services, and none of them yet have an answer for this problem. Shutterfly was also responsive to the issue, and is considering offering a setting for platform in the customer's profile. Unfortunately (a word that pops up a lot with this topic), they can't promise anything.
The Most Important Test: Quality -- I sent the same six files to every service so I could examine the results. It turns out that for any given picture, these different services sometimes deliver rather different results. And none of them quite match what I wanted to see, in part because of gamma issues and in part because the brightness of my PowerBook G4's beautiful screen simply cannot be duplicated on paper.
After I received the first (less than satisfying) set of prints from the services, my goal became to figure out which service delivered the best results. With all these prints, surely I would be able to compare the quality of the different services! To help, I enlisted my wife, my mom, and a few friends. I simply asked everyone to select the best version of each picture, and tallied the rankings using broad categories of good, medium, and bad.
Every service delivered prints which offered plenty of detail and were printed on glossy stock, just like the ones I've received from the drugstore all these years. The problem was always the color in the prints. In addition to being darker than what I expected, some came out a little more golden (making everyone look like Oscar statuettes), or a bit more brown (giving my palest friends a nice tan, and my Indian wife and in-laws a dark muddy complexion). One horrible set from dotPhoto (which lacks a Macintosh application for uploading, but which I've left in the competition because of its inexpensive pricing plans) made everyone green (or, to use iMac colors, a sage that's somewhere between seasick and Kermit the Frog). Still others looked washed out. In every set, details of my wife's black coat and our friend's tuxedo were lost due to color problems.
As surprising as I found this to be, there was a greater surprise in store. For the same picture, with the same digital file, prints from Ofoto, Apple, and ImageStation (all of which are actually printed by Ofoto) are often quite distinguishable from each other. It wasn't just Ofoto - I accidentally placed the same order twice at dotPhoto, and the two sets of results couldn't have been more different. One was by far the worst set overall, while the other came close to being the best set. Why was this? Even though the original files are digital, most of the processes used by these services are chemical and analog (the same RA4 process used to develop and print conventional pictures). Plus, although at least Ofoto recalibrates its equipment multiple times each day, temperature and humidity variations lead to slight, but noticeable, differences.
In fact, the big problem was that print quality from all the services varied widely. Each service had its share of good, medium, and bad results. None of the batches clearly stood out as being the best. Having finished what felt like our one millionth examination of the results, my wife asked which service I planned to use for my next order. Based on the vast range of quality, I couldn't give her an answer. No service was definitively the best.
Overall, I am rather disturbed by these results. I truly wanted someone to offer the best pictures. I even wrote a draft of this article assuming that Apple's service through Ofoto would take care of the gamma issues, but the prints did not show that. The only results that matched the color I saw on screen were in the book I printed from iPhoto (which uses a laser printing process which does not bring out the detail that the photographic process shows off so well). I wanted Shutterfly, with its beautiful Web site, to be the best, but its results were no better than the others. I was also rooting for PhotoAccess because it offered prints in sizes comparable to the images' actual ratios, but it too failed to offer consistent quality.
No One Is Picture-Perfect -- No single service stood out. Apple's iPhoto makes ordering the easiest, but offers the fewest additional products. dotPhoto offers the best price, but is a pain to use and delivered the worst results. Ofoto prints the quickest. PhotoAccess offers 1.33:1 aspect ratios and the widest range of merchandise, but its output (like the others) ends up too dark. Shutterfly has the best Web site, great customer service, and has supported Macs the longest, but they suffer the same color problems as the other services.
Almost every service offers free prints when you sign up, leaving you to pay only for shipping. That certainly makes it worth trying multiple services. Due to the ease of ordering prints through iPhoto, and because I think the Apple/Ofoto combination is most likely to adjust its output for Macintosh users, I plan to order most of my prints from Apple in the future. However, when I need a product that Apple does not offer (such as mousepads, mugs, or other extras), I won't hesitate to order from Shutterfly or PhotoAccess.
Despite the uneven results I experienced, I still think it's worthwhile to use an online service to print your best digital photos. The alternative, which I touched upon in part one of this article, is to print the photos yourself on an inkjet printer. I received a lot of feedback from TidBITS readers about this topic, pointing out the cost savings for large prints plus the capability to produce comparable results in image quality. With some help from a reader who has a lot of experience printing photos at home, I intend to explore printing at home in a future article.
[Alexander Mishra Hoffman is an IT Manager in New York City, a Red Sox and Pats fan, and a newlywed.]