david_s5

The problem with mounting a camera on your chest is that on many bicycles, you lean foward to grip the handlebars. This means that a camera stabilized against your chest would get pictures of your front wheel. In addition, other cyclists are likely to be facing the same direction you are, not to mention leaning forward, so forward-facing cameras tend to get pictures of other riders' shorts.

<p>

I've generally hand-held the camera, used a fast shutter speed, and blindly photographed whatever was behind me. A lot of the pictures wind up being of the sky, roadside weeds, etc. but enough work to make things interesting.

Via <a href="http://slashdot.org/article.pl?sid=05/11/21/2316216&tid=126&tid=152">slashdot</a>, you can see some <a href="http://graphics.stanford.edu/papers/lfcamera/">examples of photos taken with this kind of equipment</a>. Barring some major technical advance, I don't see this kind of equipment coming into widespread use; from the <a href="http://graphics.stanford.edu/papers/lfcamera/faq.html">FAQ</a>:

<blockquote style="background-color: lightyellow; padding: 0.25cm; border: thin silver solid">

Are you taking a 16MP camera and producing roughly 300x300 final images?

<p>

Yes, the resolution of the final images is equal to the resolution of the microlens array, which is just under 300x300 in the prototype that we built.

</blockquote>

Scott Robertson wrote:<br>

<i>I really hope that's not what is going on.</i></p>

<p>

I've got one more guess about what is happening to your site: maybe Google is noticing that much of the text on your pages is nearly identical, due to similar formatting HTML and the standard header and footer, and therefore ignoring images on your pages.  If I do a non-image search for <a href="http://www.google.com/search?num=100&hl=en&lr=&ie=UTF-8&safe=off&c2coff=1&q=site%3Aslrobertson.com&btnG=Search">site:slrobertson.com</a> I get only a couple pages.  If I tell Google to not filter very similar pages, I get <a href="http://www.google.com/search?q=site:slrobertson.com&num=100&hl=en&lr=&ie=UTF-8&c2coff=1&safe=off&filter=0">thousands of pages</a>.</p>

<p>

Could Google be applying some sort of similar filter for the removal of images on pages with similar text?

</p>

Marc Epstein wrote:<br>

<i>I had noticed that my site was getting abundant hits from google searches, but you know, I have never had on sale that resulted from those hits.  I suspected that they were just 'people' looking for images, for whatever reason.</i></p>

<p>

I wonder if the lack of sales is simply because your visitors don't realize that you sell prints. The pages where search-driven visitors land doesn't say anything about being able to buy a print, so adding that information might result in sales.  Copywriting is an art, and one which you need to carefully hone by making modest changes to your web pages and tracking the corresponding impact on your conversion rate.</p>

<p>

Marc Epstein later wrote:<br>

<i>I would rather visitors to my website come through the front door rather than a window and browse around.</i></p>

<p>

People can and will bookmark your pages and forward links around.  You could theoretically enforce something like this using cookies, but it would be a fair bit of work.  My intuition is that you're probably better off making it easy for potential customers to buy images directly from the page they land on.  An alterative, which could work if you really are getting enormous numbers of users, would be to sell advertising.

</p>

<p>

Scott Robertson wrote:<br>

<i>A week ago several of my image pages (big images with H1 titles, long descriptions, proper ALT tags and keywords for my own site search function) were still appearing near the top of Google searches.</i></p>

<p>

You still are appearing near the top of <a href="http://search.yahoo.com/images?qry=&ei=UTF-8">Yahoo image searches</a> for many of your chosen keywords.  I suspect that the Google ranking change is because that engine now prefers a larger amount of natural prose than it used to.  Unfortunately the folks at Google have a huge incentive to keep tweaking their algorithm: it forces businesses to run a <a href="http://adwords.google.com">pay–per–click advertising</a> campaign to bring in traffic.

</p>

Bob Atkins wrote:<br>

<I>Does anyone know (or have a good guess) at what criteria Google use to index images? Do they have to be on a page with lots of relevant text. Does Google index via the "alt" tag in the image display HTML or by the filename (I notice many images don't have the search terms in the file name, so I doubt it's that).</i></p>

<p>

While Google's actual algorithm is a trade secret, and I haven't seen much in the way of experiments to figure out how their image search works. I don't think that Google depends on any one source of information, such as filenames. I think that they look at several factors in page where the image is embedded or linked to, and then make a decision based on those. For example, I notice that pages which rank high in google's image search tend to have the following properties:

<ul>

<li>The text on the page contains the search term multiple e times <i>and</i> some other text.</li>

 

<li>The <a href="http://www.w3.org/TR/html4/struct/objects.html#h-13.8">ALT</a> and/or <a href="http://www.w3.org/TR/html4/struct/global.html#h-7.4.2">TITLE</a> attribute of the <a href="http://www.w3.org/TR/html4/struct/objects.html#h-13.2"><img></a> tag contain the search term</li>

<li>The image file name and/or directory name contains the search term</li>

<li>The page which the image is on has links from outside the site pointing to it</li>

</ul></p>

<p>

A good example of this is when you search for <a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com++chart&btnG=Search">site:bobatkins.com chart</a>. This brings up <a href="http://bobatkins.com/photography/technical/testing2.html">http://bobatkins.com/photography/technical/testing2.html</a>, a page which contains a lot of text, repeated references to the word ‘chart’ a link to a file named ‘chart.gif’

</p>

<p>

In a similar vein, searching for <a href="http://images.google.com/imgres?imgurl=http://www.naturegraphics.net/bc502%2520great%2520egret.jpg&imgrefurl=http://www.naturegraphics.net/animals01.htm&h=383&w=250&sz=15&tbnid=luT8QH1ZU54J:&tbnh=117&tbnw=77&start=1&prev=/images%3Fq%3Dgreat%2Begret%26hl%3Den%26lr%3D%26ie%3DUTF-8%26c2coff%3D1%26sa%3DG">great egret</a> brings up <a href="http://www.naturegraphics.net/animals01.htm">http://www.naturegraphics.net/animals01.htm</a>, which contains lots of text, the words ‘great’ and ‘egret’ several times, and contains an <img> tag with the title containing ‘great egret’ and points to a file with ‘great egret’ in its name. In addition, there are several easy to find links to the page containing the image — you can get there from the home page in two clicks without needing to wade through a database.

</p>

<p>

In contrast, a search for <a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com+polar+bear&btnG=Search">site:bobatkins.com polar bear</a> turns up nothing. The obvious page <a title="Polar Bear" href="http://bobatkins.com/photography/images/lores/slides/06_IMG_0021.html">http://bobatkins.com/photography/images/lores/slides/06_IMG_0021.html</a> contains only one picture of a polar bear. Neither the file name, nor the ALT and TITLE attributes of the <img> tag contain the words ‘polar bear’. There is little text on the page, and most of what there is happens to be automatically generated. The actual words ‘polar bear’ are some of the least visible words on the page; they're in a small hard to see font, instead of in something like an <h1> tag or the title. Finally, I don't think that there is anybody in the world who is linking to your web polar bear page except you, and your link doesn't contain any descriptive text.</p>

A few other examples of searches which turn up images on your site:

<ul>

<li>

<a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com+landscape+&btnG=Search">site:bobatkins.com landscape </a></li>

<li><a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com+nature&btnG=Search">site:bobatkins.com nature</a></li>

<li><a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com+lens&btnG=Search">site:bobatkins.com lens</a></li>

<li><a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com+fisheye&btnG=Search">site:bobatkins.com fisheye</a></li>

<li><a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com+sensor&btnG=Search">site:bobatkins.com sensor</a></li>

<li><a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com+digital&btnG=Search">site:bobatkins.com digital</a></li>

<li><a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com++atkins&btnG=Search">site:bobatkins.com atkins</a> (Notice that this turns up a LOT of pictures — your name appears on every page repeatedly and in the site name)</li>

<li><a href="http://images.google.com/images?hl=en&lr=&ie=UTF-8&c2coff=1&q=site%3Abobatkins.com++new+york&btnG=Search">site:bobatkins.com new york</a>

</ul>

Notice that these are all pages with text talking about the search term.

</p>

I don't think google is removing images based on copyright notices in the exif information — my web logs indicate Googlebot doesn't retrieve the image files, which means that they can't use exif information to make their decision. My belief is that the image search is based primarilly on the value of the ALT label and text surrounding the <img> tag, as well the off-page ranking information that Google uses for text page searches.

<p>

I'm going to guess that it is one of the following:

<ul>

<li>You changed the text in some of your web pages</li>

<li>Google updated their algorithm, causing your images to appear much lower in the rankings</li>

<li>Some link or series of links to your page were removed and/or penalized by Google as a ‘link farm’</li>

</ul>

</p>

<p>

<I>why on earth wouldn't you bring all three lenses?</I><br>

Glass is heavy, and if you descend to the Colorado River, you can't bring your car — you need to carry the lenses on your back or rent a mule.

</p>

<p>

In general, it is straightforward to remove any watermark which does not substantially deface the image. This includes JPEG comments, small copyright notices within the images (you can crop), EXIF information, and schemes which work by altering low-order bits in the image.

</p>

<p>

If people can see your image, they can copy it, and all the technology in the world won't stop that. While reminding people to not copy your photographs won't hurt, the most you can really hope for is to catch infringers after they actually infringe.

</p>

<p style="text-indent: 3ex; margin-top: 0">

Road closure information for Vermont is posted at <a href="http://www.511vt.com">www.511vt.com</a>.  According to that web site, route 108 is "Closed due to deep snow and packed snow - until 5:00 PM, 05/10/04"

</p>

<p style="text-indent: 3ex; margin-top: 0">

I don't know what conditions are like this year, but there is usually still snow on top of Mount Mansfield in early May.  Snow on trails ⇒ ice on trails.  Spring temperatures are highly variable from one day to the next.  Get a local weather forecast before you go, and don't be afraid to turn back if you encounter conditions beyond what you have prepared for.

</p>

<p>

I've been using a Minolta Dimage Xt while cycling for some time now, and have found it to be a useful point and shoot camera. It is not a replacement for your SLR, as there is no manual aperture or shutter speed control.

</p>

<p>

So far as I can tell from looking at spec sheets, the Xg is both more expensive and less capable than the Xt. In particular, it seems to have a lower resolution liquid crystal display, and appears to be unable to produce lossless .tiff files like the Xt.

</p>

<p>

For both models, you will want a larger memory card than the one the camera is sold with. If you take large numbers of pictures, or ride in cold weather, you will also want a spare battery.

</p>

<p>

The major optical tradeoff with these tiny cameras is the lens quality. Despite the small sensor, there is very noticable fading in the corners of the pictures. I've posted some <a href="http://www.employees.org/~dsacer/weekend/03282004/index.shtml">example cycling photos taken with the Xt</a> so that you can get some sense of its suitability. If your goal is pictures for the web or small prints to share with friends, these cameras are a reasonable choice.

</p>

<p>

Spec sheets for comparison:<br>

<a href="http://www.minoltausa.com/eprise/main/MinoltaUSA/MUSAContent/CPG/CPGProducts?NS=true&cname=dig&fname=dig_ps&Mname=DiMAGE_Xt&mDetail=Specifications">Minolta Dimage Xt</a><br>

<a href="http://www.minoltausa.com/eprise/main/MinoltaUSA/MUSAContent/CPG/CPGProducts?NS=true&cname=dig&fname=dig_ps&Mname=DiMAGE_Xg&mDetail=Specifications">Minolta Dimage Xg</a>

</p>

<p>

Is your website intended to provide information to existing clients, or to bring in new ones? If the latter, it might help to do a little keyword research, and to create pages which include the keywords people are using to look for photographers in your area.

</p>

<p>

For example, I notice that your wedding page mentions Stratford prominently, but not other towns.

This may be some sort of statistical anomaly, but a quick check with <a href="http://www.digitalpoint.com/tools/suggestion/?keywords=ontario+photographer">WordTracker and the Overture keyword suggestion tools</a> indicates that there are many web searches for "wedding photographers in cambridge, ontario", some for "London Ontario photographer" but relatively few which are targeted at Stratford. I'm not sure about how practical travel time is from your location, but it may also pay to produce additional pages which describe your willingness to work in locations such as Kitchener, Guelph, St. Thomas, and Waterloo, as each of these seems to get a couple searches per day.

</p>

<p>

If the objects are small, and you aren't allowed to bring large objects in, try using aluminum foil and ordinary white paper as bounce targets. Depending on distance, you may need to use or two stops of exposure compensation.

</p>

<p>

I suggest using the foil to send the flash to a piece of paper above, and to one side of your subject. Use the paper to then send difuse light back down onto the object. Tape and ring stands are ideal for controlling this kind of setup.

</p><div></div>

<p>

My understanding is that right now, we do a poll of the database every time somebody looks at the unified view. This eats up a lot of CPU cycles.

</p><p>

The classic software strategy for solving this problem is to do a database poll once every few minutes, and serve a single cached database view to all visitors between polling events. This cuts down on CPU utilization quite dramatically. Style sheets can be used to provide different settings-based appearances for individual threads.

</p>

<p>

Steve Chan wrote:<br>

<I>Drawbridge is supposed to be interesting - I live virtually next door to it in Fremont, but haven't gotten around to sneaking out there (it is a reserve and people aren't supposed to wander in).</I>

</p>

<p>

The problem isn't that the town is in a preserve. The preserve is open, and there is convenient parking at the <a href="http://desfbay.fws.gov/directions.htm">environmental education center</a>. The problem is that you need to cross a somewhat busy railway bridge to get from the wildlife refuge into the ghost town. For this reason, I suggest joining one of the infrequent public tours.

</p>

<p>The ghost town of Drawbridge contains an abandoned train station and is just across a bridge from SF bay national wildlife refuge in Alviso. You can get a brief glimpse of it if you ride the Caltrain from Pleasanton to San José. There are also sporadic and infrequent public tours of the ghost town — see <a href="http://www.sfbws.org/train.htm">this announcement</a> for a past example. Other local examples such as Wright or Loma Prieta are either harder to obtain legal access to, or much less well preserved.</p>

<p>

If you're interested in photographing old railway artifacts, you might also want to check out the <a href="http://www.roaringcamp.com/">Roaring Camp Railroad</a> near Felton.

</p>

<p>Unfortunately, we seem to only be having a middling wildflower season this year. A bit more rain would really have helped.</p>

<p>

That said, flowers are blooming along the coast ranges at lower elevations and the altitude will increase steadily over the next few weeks. You might try <a href="http://www.coepark.org">Henry Coe State Park</a> -- I saw fields of yellow violets there this past weekend, and there were lupines blooming along the access road between route 25 and the east side of the <a href="http://www.nps.gov/pinn">Pinnacles National Monument</a>. That access road is also a reliable place to spot <a href="http://www.mbr-pwrc.usgs.gov/id/framlst/i2940id.html">California Quail</a> and black-tailed deer.

</p><div></div>

Ease of use is something which is really quite subjective. I urge you to find a local camera shop which will let you spend some time experimenting with different cameras so that you can find one you enjoy using. In the ideal world, you'd bring a laptop to the shop so that you could have the full picture-taking to image transfer experience. A camera which <i>you</i> don't enjoy using will just sit in a drawer somewhere.

<p style="text-indent: 1.5em">In general, the lossy compression in the <a href="http://www.jpeg.org/">JPEG format</a> causes areas of a single color to appear more homogenous than they are in the original image. This is consistent with the disappearance of slight variations in a solid red area. If you really want to post an image without distortion due to the <a href="http://www.theimage.com/web/graphic/jpgvsjpg/gif2A.html">JPEG lossy compression</a>, you'll need to use an image file format which doesn't use lossy compression, and post the image someplace <i>other</i> than <a href="http://www.photo.net">photo.net</a>.

</p><p style="text-indent: 1.5em">

If you want to be certain of not having JPEG lossy compression distort your image, I suggest using the <a href="http://www.libpng.org/pub/png/">PNG format</a>, and posting a link the the image in question in your photo.net post. Virtually all web browsers in use today suppor the PNG format, although your image file will be somewhere around 5 times larger than a top quality JPEG. Because the PNG format uses lossless compression, you don't lose any detail as a result of using it.

</p>

<p>

I can't say for sure how long it will take to get there as I managed to get quite lost along the way. There are no signposts at the intersections along dirt roads in that area. I suggest that you allow 90 minutes from the edge of the asphalt, and bring a 4wd vehicle, detailed maps, water, and plenty of gas.</p>

<p>

Don't expect to see anybody or have cell phone service on your way — you're on your own if anything goes wrong.

</p>

If you're really worried about JPEG induced artifacts in web postings, you can use a PNG file instead. The <a href="http://www.libpng.org/pub/png/">PNG file format</a> uses a lossless compression scheme, and won't introduce the artifacts that JPEG does. Most decent photo manipulation software supports this format, as do 99%+ of web browsers. The downside? For a given image size, PNG files are typically four to six times larger than an equivalent high-quality JPEG.

Getting skin tones to render correctly on a digital camera is generally an issue configuring the camera for your lighting conditions.

<p>

The default configuration of most digital cameras causes them to choose a lighting-based white balance automatically. Unfortunately, this automatic white balance scheme is often wrong, and doesn't always handle things like mixed sunlight and florescent lighting very well. By explicitly telling your camera what kind of lighting you're using, you can greatly produce the accuracy of skin tones. Alternatively, you can go and correct these issues using photoshop once you've taken the picture.

<p>

I've posted some example images <a href="http://www.photo.net/photodb/folder.tcl?folder_id=377291">here</a>. Each image was taken using a different white balance setting on an inexpensive digital camera. The actual light was a mix of twice-reflected sunlight, florescent, and the output of my CRT. As you can see, some settings produce reasonable results, while others produce waxy or oversaturated color.

<p>

If you want to guarantee that you purchase a camera which does a good job with skin tones, I suggest visiting your local camera dealer with a laptop, and trying any digital camera you are considering. That way, you'll be able to get a sense of whether white balance control is something which your camera's interface lets you adjust easily before you buy. Try taking pictures with:<br>

* White balance set to auto, in both florescent light and daylight near the store window<br>

* White balance set to each possible setting in florescent light<br>

* White balance set to each possible setting in daylight<br>

and make sure you're pleased with the results.