I recently bought a Panasonic Lumix DMC-LX5 camera, which records video in both MPEG and AVCHD formats. The camera has some nice improvements over the previous LX3 camera. I really like the extended zoom range, 24-90mm versus the 24-60mm in the old camera. Another feature I really love is the ability to zoom while shooting a movie. The sounds quality improved significantly as well, some videos I took with the old camera had some strange sound issues.
One thing that drove me nuts however was the inability of iMovie 7 or Final Cut Express to import the AVCHD movies I made with the camera. They would import only the first few minutes of the movie and invariably stop after that. The MacOS X software that comes with the camera doesn't handle movie files, and the Windows version sucks badly.
So I was stuck with no way of converting the AVCHD movies to an MPEG file format that can be viewed on MacOS X. Adobe Photoshop Lightroom 3, the program I use to catalog the photos and videos I take, does not understand AVCHD either.
You can however convert the .MTS files from the AVCHD directory created by the camera using the open source ffmpeg tool. To do this, make sure you have MacPorts installed on your computer, then install ffmpeg like this:
sudo port install ffmpeg +gpl +lame +x264 +xvid +mp3 +aac
Once you have ffmpeg installed, you can convert the MTS files like this:
Whoa, it looks like I missed a slew of new camera announcements. The one I'm particularly interested in is the announcement of the new Panasonic LX5 camera, the successor of the LX3. I own the LX3 and it's a really great camera, especially in low light.
The LX5 camera extends the zoom range to 90mm from 60mm, it is now 24-90mm f/2.0-3.3, versus the LX3 which had a 24-60mm f/2.0-f/2.8. Sensor size is the same but it's a brand new one which supposedly has an increased dynamic range and even less noise. It looks like the sensor changes will improve movies in low light, since the minimum illumination is as low as 3 lux in movie recording mode.
Another addition is the new AVCHD Lite recording format for movies, which allows longer movies to be shot. The movies are no longer limited to the maximum 2GB file size. I haven't heard about the AVCHD Lite version before, but I find AVCHD a pain to work with. At least on Mac, you can't see the movies by opening them in QuickTime, you need to convert them first using an editing program.
Overall it looks like the LX5 is a nice incremental improvement over the LX3. I can't wait to see some side-to-side image comparison of the two cameras. For my usage, I think the LX3 is good enough. I don't shoot movies that long using the camera, so the new movie recording format doesn't affect me. With a 16GB SDHC card, I have plenty of space to shoot movies and pictures for a whole day.
Judging by the performance of its predecessor, I heartedly recommend buying the LX5 camera. Its low light performance is astounding for a compact camera.
Here are some pictures I've taken at various concerts using the LX3.
Umphrey's McGee at the Warfield, San Francisco:
Chickenfoot at the Greek Amphitheater, Berkeley:
Of course, the camera is great during daylight. Here's a picture I took last winter at Heavenly, South Lake Tahoe.
Awesome! Thanks Jeffrey for putting these together. This of course would not have been possible without Adobe's Lightroom SDK, which is pretty cool in itself.
What I'd really like to have is a plugin that emulates Apple's iPhoto Digital Photo Access Protocol (DPAP) protocol. This way the other Macs in my home can view Lightroom's albums as a regular remote iPhoto. Apparently some people reversed engineered this protocol and have an implementation available in Perl.
Hopefully future versions of the Lightroom SDK will include the ability to run plugins at all times in their own thread, as well as have the ability to use server side sockets.
I use Adobe Lightroom to organize my photographs, both digital ones as well as the scanned slide films I take. Lightroom is working great for me and I've written about my disappointment with Apple Aperture in a previous entry.
Yesterday I was looking for a way to take a subset of the pictures I have and convert them to JPG. I want to use the JPGs on a small Mac Mini to make presentation in FrontRow using a newly acquired Canon SX7 digital projector (more on this in a later entry).
Unfortunately Adobe Lightroom does not offer such a functionality. You can export all the pictures in a directory to JPG, but there's no way to take a directory, and recursively export all the pictures within it to JPG while maintaining the directory structure they are in.
I could have written a small shell script to do this, but I have a mixture of TIFF, JPG and RAW images that had to be processed. I wanted to get the image processing capabilities from Lightroom or Photoshop, instead of using dcraw. Adobe Lightroom doesn't have a way to script it, so I looked at Photoshop CS3 to do it.
To use, download the script, install it in /Applications/Adobe Photoshop CS3/Presets/Scripts and restart Photoshop CS3.
After restart, the script is available under File -> Scripts. When you select it, the script will ask the source directory containing the images, and the destination directory where the JPG images should be placed. The script has been tested in Photoshop CS3 running on MacOS X 10.4.10. It seems to be able to process TIFF files produced by my scanner, CR2 RAW files from my 5D and the older 20D, as well as JPGs from various digital point-and-shoot cameras.
If you stop the script, the next time you restart it on the same source and destination directories, it will not process the files that have already been processed.
Note that if you process a lot of files, Photoshop ends up using a lot of the disk space on your scratch disk. I could not figure out why this happens. Quitting Photoshop seems to reclaim the scratch disk space.
The script is public domain, feel free to use and modify it as you wish. If you do end up using it and fix various bugs, please let me know so I can incorporate them in my script.
To manage my ever expanding collection of pictures, about 4 months ago I bought Apple's Aperture 1.5. At that time I compared it with Adobe's Lightroom, which was still in beta. Aperture felt a lot more solid with respect to features, especially quick adjustments and book printing options. Lightroom seemed very promising, very fast but kind of short of editing features. In the end, I bought Aperture in the hope the speed will not be an issue on my high end MacPro 3GHz four-core system with 4GB of RAM.
Well, I soon grew disappointed. My photo gallery has about 30,000 shots in it. About 25% of these are scans of slide film, TIFF files whose sizes vary between 32MB and 132MB. About 60% are RAW files from Canon 20D and Canon 5D, with sizes between 7MB and 14Mb. The rest are JPEGs from my wife's and daughter's digital cameras. The files are sorted by year and location in hierarchical directories. The pictures occupy some 360Gb of my hard drive.
Overall Aperture is pretty good in speedy editing features. You have pretty much all the options you're going to use right there in the application. No need to start an external application to change exposure, crop and even tilt the image.
The first problem I encountered with Aperture was that there is no easy way to have it import all the pictures at once. Aperture uses the notion of projects, and each of them is limited to 10,000 pictures. So I had to make each year a project, which in itself is a pain. Then each time I'd add a new directory in my hierarchical directory structure, I would need to remember and import that directory in the right project in Aperture.
But the biggest problem I found was only after I started assigning keywords to pictures. The whole idea of this is to be able to effectively search through your pictures to find what you're interested in. However Aperture becomes increasingly slow as you keep adding keywords and pictures to your projects. At the current moment, doing any search will totally bog down my computer to the point it becomes unusable. How the heck did the Apple engineers implement this feature? Do they open each file to look at its keywords? It certainly looks like this is the case, I can't believe how stupid this solution is.
I was so disgusted by this that I went back to Adobe Lightroom, now a full 1.0 release. The application is still speedy and it seems to be doing the right thing. There are some features missing from Lightroom, that are in Aperture, most notable the picture tilt feature, and the ability to create photo albums and send them to be printed to Shutterfly or some other similar service. Apart for this, the application is much speedier and less CPU and disk intensive than Aperture. Update: I take it back, Lightroom does provide tilting, it's right there in Develop, under the crop feature. Not very well advertised.
This said, I can't wait for future updates of Lightroom and the newly announced Photoshop for MacOS X for Intel.
The film holder can hold 120 and 220 frames, but for smaller frames it uses a set of black masks to cover the unused portion of the glass. This helps with the proper exposure of the film, as it keeps light out in the unused portion of the frame. The masks are also cleverly done, as they have on one side a set of punches that uniquely identify the mask and its cut frame. As soon as you insert into the scanner the glass holder with a film strip and such a mask, the scanner reads those punches and communicates the information to the scan application. This way Nikon Scan automatically recognizes the size of the frame you intend to scan, and adjusts the crop frame accordingly. Pretty nifty!
One problem I did find with the glass holder is that on some frames I could clearly see Newton rings. I tried cleaning up the glass using a Pec*Pad wipe and Eclipse cleaning solution. This helped with some rings, but not with all of them. Luckily this was a problem for me with only one frame out of several I scanned.
Save for the Newton rings problem, scanning Xpan film strips in the Coolscan 9000 is very easy with the FH-869GR holder. In fact it appears to be the only way to scan Xpan film strips with the Coolscan 9000. The regular film strip holder that comes with the scanner has plastic tabs between the 35mm frames. My guess is that even if you cut them, the scanner will refuse to scan more than 36mm in length.
The only problem with the FH-869GR holder is that you cannot batch scan multiple frames. The glass opening is for exactly one frame, even if 3 uncut Xpan frames can fit in the holder. Oh well...
In the previous post I was observing that scanning slide film on Epson V750-M doesn't live up to the high expectations I had. Scanning some of the Xpan slides produced fairly low quality images that I wasn't satisfied with. So I started out by producing a more rigurous test between the Nikon Coolscan 5000 and the long awaited Epson V750-M.
To illustrate the difference between the Nikon Coolscan 5000 film scanner and the Epson V750-M flatbed scanner, I scanned a Schneider SMPTE RP-40 test slide (from Adorama) - click on image to see a larger size:
I zoomed on the center portion, which displays a series of lines, very small and hard to discern on a lightbox with a 10x loupe. Here is how the slide appears on the two scanners, with different settings.
Nikon Coolscan 5000, 4000dpi, 16 bit depth, color space grayscale, 1x multiple sample scanning:
Nikon Coolscan 5000, 4000dpi, 16 bit depth, color space grayscale, 4x multiple sample scanning:
Epson V750-M, 4000dpi, 16 bit depth, color space grayscale, mounted slide placed in mounted slide film holder:
Epson V750-M, 6400 dpi, sampled down to 4000dpi, 16 bit depth, color space grayscale, mounted slide placed in mounted slide film holder:
Epson V750-M, 6400 dpi, sampled down to 4000dpi, 16 bit depth, color space grayscale, mounted slide placed directly on the glass surface:
Epson V750-M, 4000dpi, 16 bit depth, color space grayscale, mounted slide placed in mounted slide holder raised 1mm:
As you can see from the images above, the Nikon scanner has a much better contrast, as well as much better resolution. The scanner has a builtin focusing mechanism, which by default is set to automatic, but one can adjust it manually in the software. When set to 4x, Coolscan's multisampling facility provides a much sharper image - this works by scanning the slide 4 times and combining the images. In practice the scanning time increases only slightly.
The Epson scanner's high resolution doesn't appear to make any difference in the quality of the image. Several people, including Fazal Majid in a comment on my previous post, suggested scanning with the mounted slide film raised approx 1mm. I did this experiment by placing the slide holder on few paper sheets stacked up to be 1mm in height, as measured with a digital caliper. As you can see from the picture above, this doesn't increase the quality of the picture. Update: Fazal pointed to Vincent Oliver's review and how the tab can be reversed to achieve the same height increase. The obtained scanned image is essentially the same as the one above.
In terms of quality there is no doubt in my mind the Coolscan is the clear winner.
I patiently waited for the Epson V750-M scanner to hit the market, so I can scan my Xpan slides. I got it last week from B&H and ran some scans on them.
The package contains a bunch of software goodies in it. It has the Silverfast AI version 6, Adobe Photoshop Elements 3, Epson Scan, and Monaco EZ Color for calibration. It does not have a printed manual however, which was a surprise to me since the number of accessories and options is quite high. I usually don't resort to the manual unless I have no choice, and this time I really felt the need. After looking more carefully into what was installed on the computer, I did find an HTML version of the manual installed in the same directory with Epson Scan.
The first thing I did was to calibrate the scanner using targets from Wolf Faust. Then I scanned some slides made with the Xpan, using the 45mm lens with the center filter attached. The camera was placed on a tripod, so the image should have been as sharp as possible. I used the provided strip film holder, the film was not mounted. This film holder looks quite wobbly, and the film doesn't stay completely flat.
The results were quite unsharp. I scanned the slide using Silverfast, VueScan and Epson Scan, all gave pretty much the same result. On the film holder, the film strip doesn't stay flat on the glass of the scanner, but rather it's raised few millimeters above it. Taking it out of the film holder and laying it directly on the glass doesn't improve the situation either.
Unlike on my Nikon Coolscan 5000, the software doesn't have any option to adjust the focus of the scanner's internal lens. My guess is that this contributes quite a bit to the lack of sharpness in the final image. Epson provides some additional plastic tabs that can be used to replace the existing ones in the film holders. This solution however seems like a cheap way of fixing a fundamental problem.
The result is that I'm not satisfied with the results of the scanner, and I'll be returning it. I'm seriously thinking of getting the Nikon Coolscan 9000 with the FH-869GR film holder.
It looks like Hasselblad is going to discontinue their panoramic Xpan cameras. The reason is a new European law that prevents manufacturers to use lead in their products. Ouch, I guess a number of cool products might disappear from the market because of this.
And I just got such a camera, it's a really cool camera. What a pitty! From here.
I just received a brand new Canon 5D! I ordered it two days ago from B&H. Even though it was not supposed to be in stores until October, I briefly checked B&H's web site to see what the status was. I was very surprised to see it in stock, so I placed an order immediately. I was lucky, half an hour later they ran out of stock, I guess other people noticed it too.
The camera feels like 20D's big brother, and is much better than the dream camera I was thinking about a little over a year ago. Few things would have been nice to have, or at least to be different:
weather sealing would be good
the depth preview button is in the wrong place. It should be on the right hand side, just like on EOS 1v and all the 1Ds. This way you could press it with the same hand used to hold the camera, without the need to have both hands on the camera.
The camera is otherwise impressive. I am going to have a really hard time reaching for my EOS 1V film camera instead of the 5D. Shooting slide film is still nice, especially when you want to project the results. But getting the results faster has its advantages, especially when you're in the field and can make adjustments to get a better image. But I cannot complain, it's a decision is nice to be able to make.
I finally made the leap in the digital photo world by acquiring a Canon 20D. Although I do own a Canon Powershot S30 for few years now, I didn't use it as much as my film camera.
The 20D is a fairly impressive camera, especially compared to the top-of-the-line Canon film camera, the EOS 1V. The camera is a bit smaller and lighter than the 1V, but feels like a solid camera. The picture quality looks impressive, although I still have to test it on a tripod and compare it to the same shot taken on film.
I've taken some shots inside using my 550EX flash, and the benefit of the instant feedback helps a lot. With the film camera I would rely on experience to orient the flash head to get different effects. On the 20D I can see immediately what I need to improve so I can get a better light.
Deciding when to use film and when digital is going to be an ongoing dilemma. For nature, I think I'll still use slide film, as projecting slides is so much nicer than viewing images on a computer screen (even on an Apple 23" Cinema Display ;). But for everyday shots the 20D will probably be the preferred camera.
Just few days ago, I was thinking Canon is due to come up with an improved version of their EOS 10D camera. And that's exactly what they did with the EOS 20d camera; dpreview.com has the scoop.
Canon also released two new lenses and a flash. The lenses work only on digital bodies. The first lens is the EF-S 17-85mm f/4-5.6 IS USM, which is the 27-136mm equivalent in 35mm cameras, and costs $600. The second lens is the EF-S 10 - 22 mm F3.5 - F4.5 USM, or the 16-35 mm in 35mm cameras; it costs $800.
I wonder what the quality of these lenses is. The first appears to be the equivalent of the 28-135mm IS USM lens. The second lens doesn't seem to have such an equivalent - I would not consider the 17-40mm f/4 L lens being equivalent with it.
The 580EX flash is an improvement over the 550EX, making it more adapted to Canon's array of digital SLR cameras. For one thing the 580EX is able to obtain information on the sensor size from the camera and adjust its zoom flash position accordingly. Another key improvement is the ability to send color temperature information to the camera, which can adjust the white balance accordingly.
The 580EX also has an optional L bracket which allows you to position the flash on the side of the camera. This is useful when taking vertical shots, when you want the flash to be at the top, not on a side where it would normally end up if you turn the camera. What I don't know is how this bracket works on a tripod setting.
My Arca-Swiss B1 ballhead has a problem locking and unlocking the ballhead. I need to turn the locking knob for a complete circle and a half (540°) to lock the ballhead, instead of the usual 180° or so. To make the matters worse, turning the locking knob back in the original position doesn't unlock the ballhead.
To unlock the ballhead I need to firmly press on the ballhead, pressing either on the camera or directly on the clamp after unmounting the camera. This situation makes orienting the camera very difficult. It's pretty sad to see this happening with such a highly praised device. Luckily it's still covered by the 3 year warranty, so I'll send it to Arca-Swiss for repair.
I came back from a four-week vacation less than a month ago. I hiked in the mountains of Romania (Piatra Craiului to be more precise), and toured through Transilvania, the northern part of the country. Very beautiful places and people, I really enjoyed it!
I shot some 55 rolls of slide film, about 2000 frames during the whole trip. I'm now working on scanning them using a Nikon 5000 ED with the SF-210 auto slide feeder. I scan about 3 films of 36 positions per day, which is pretty good, considering I do this only in the morning and late evening. I just feed the slides to the scanner and let it do its job unattended, which is very convenient.
Although this may look like a lot of work, it's not too bad. Except that it takes a lot of time to scan all the films I have. I'm seriously considering switching to digital for things other than snapshots. I'm waiting for Canon to come up with an updated version of the EOS 10D camera. While EOS 1D Mark II looks really impressive, it's a heck of a camera, quite heavy to be carried in a backpack in the mountains.
What would be best is for Canon to come up with a digital version of the EOS 1v, just like the EOS 1d Mark II but without the power drive booster. An 8Mb pixels camera with 3 fps, with an 1.3x magnification factor, and all the features of EOS 1v: such a camera would really rock!
I'll post some of the shots hopefully sometime later this month.
The past few weeks I went through my slide folders and scanned some of the ones I liked the most.
Shooting film is a lot of fun. Especially the slow film - ISO 50 or ISO 100 - forces you to shoot from a tripod. This in turn forces you to slow down and do the right thing with exposure and composition. The equipment is bulkier and heavier than carrying a small digital camera, but the results are much nicer.
To eliminate most of the costs associated with shooting film, I decided to do my own film development, which so far has provided very good results, with only few failures. I shoot and develop color slides, since it's a lot easier than with print film. With slides I can look at them on the lightbox or project them on the wall screen, without going through the complicated print process.
Few months ago I updated to a better Canon camera to have the ability to download the shooting information. I've written a small Java program that takes this information and generates a PDF file for printing slide labels, that I attach to individual slides.
I've setup few albums in my picture gallery. The pictures in Valley Of Fire were made in the (very) early morning of the first day at ApacheCon 2003. The ones in Red Rock State Park were made the next early morning, when I convinced Steven to come along for a hike.
If you feel inclined to critique, please leave a comment.
Few weeks ago ReallyRightStuff released a new "L" universal plate for a multitude of cameras, including for my lightweight Elan 7E. I ordered the plate and got it in mail few days ago.
Having tried the similar "L" bracket from Kirk Enterprises, I can say the ReallyRightStuff is really the winner. Their plate is surprisingly lighter and of much better quality than Kirk's metal. The RRS plate attaches directly to the body, as opposed to Kirk's "L" bracket which requires an additional plate to be mounted on.
As far as usability is concerned, not having to flop to tripod's ballhead into the drop notch of the Arca-Swiss is really nice. Especially since this in many cases the tripod is not placed in a perfectly horizontal plane. In such cases, you'd also need to adjust the tripod legs to align the camera horizontally, which is a real pain. With the "L" plate, you just unmount the camera from the horizontal position and mount it vertically. Adjusting the camera on the ballhead is then a usual operation.
I now have a spare RRS plate B24; if you're interested in getting it drop me a note.
I've decided to buy an used Jobo processor so I'll be able to process my own slide films, instead of sending them to Fuji via an emailer or using a local shop which charges $10/roll. I got a pretty decent Jobo CPE-2 on eBay, which is able to process up to 5 rolls at once.
Last night I developed the first two rolls, a Fuji Velvia 50 and a Provia 100, using the Tetenal E-6 3 bath chemicals. One thing I've discovered by carefully reading the documentation of the CPE-2 processor is that these chemicals are not recommended for processing such films. The results were not too bad, although it's hard to say how since most of my slides were shots of animals at Zoo, and I had to use a fast aperture.
The overall process is a bit time consuming, but fun to do it (at least the first time). The chemicals have to have the right temperature (38 degrees Celcius or 100 Fahrenheit) before the processing can start.
You load the processor with water in the bath, prepare the chemicals and put the bottles that hold them in the bath, so they can reach the right temperature. In a separate plastic graduate, I put water at room temperature and placed it into the bath. I used it to measure the temperature of the water inside, and only started the processing when it reached the needed temperature.
To time the individual steps of the processing, I used FotoTimer, an excellent program for Palm. In it you can setup the steps for your process, and their times. The times for each step are described in the chemicals instructions. You need to input these in the program, but you also have to account for the time it takes to switch from one step to another, and enter these into the program.
I found the steps before and after the processing to be quite interesting. Before the processing, you have to take the film out of its case, mount it in a special reel, and place all the reels in a film tank. This operation has to happen in complete darkness. Getting it right is quite challenging, so I practiced loading a scrap film on a reel few times before doing it in darkness.
After the processing is finished and you open up the tank, you see the delicate images on the film: it's magic!
After the film is dried, cutting and mounting it in slides is quite tricky, especially when you have lots of film to mount. I tried the Gepe manual mounter, but it's no good, especially with their slide mounts. The problem is the difficulty of aligning the slide within the mount. I found the Pakon mounts to be much better, since the opening in the slide mount easily directs the film in the right position.
Depending on the chemicals you use, the processing takes between 30 and 40 minutes. I'll try the Tetanal E-6 6 bath chemicals to see the difference.
I noticed on eBay that Canon EOS 1Ds is being sold for prices as low as $1,200, which is way too low to be true. The lowest price I was able to find in a respectable shop was around $7,200. What is even more amazing is that some of the kits being sold on eBay also include a 1Gb IBM micro-drive and the Canon 16-35mm L lens, which alone is more than the asking price for the camera. Anybody knows what is the deal with these offers?
Update (October 30, 2003): Bob Atkins has a great article on photo.net about avoiding scams when buying photo equipment. He mentions not only eBay scams, but also "Scum of the Earth" photo stores with prices too good to be true.
Update (November 15, 2003): Somebody setup an auction on eBay explaining all about the camera scams. He's actually selling a photocopy of the instructions to conform to eBay's policy of listing actual items. Very nice!
Last week I got my Arca-Swiss Monoball B1 ball head, a truly amazing piece of machinery. I also got a quick release plate from Really Right Stuff, which perfectly matches my camera body, an EOS 7E. I wish they had L plates, they seem to be so convenient for taking vertical shots with a tripod.
On Saturday, July 4th, I took the new toys out in Yosemite for a day on the Misty Trail (Vernal and Nevada Falls) to try them out. It was quite a long day, woke-up at 4:30am, drove there for 3 1/2 hours, hiked up for 5 hours, and head back home for another 4 hours. The weather was very nice, although a bit too hot for hiking. Pictures should be coming shortly.
So far I haven't been too happy with any of the film processing labs in the San Jose area. For the past week I was contemplating getting my own film and print processor, so I can develop and print my own films without having to rely on untrusted photo labs.
My investigation revealed Jobo CPP-2, a semi-automatic film and print processor, which is about $1,600 new. The processor needs special drums for processing films and prints, but with a $500 rebate going on right now, these would come for free.
With such a tool, the chemical's cost for developing a negative film is about $1.60, and the price of a 4"x5" print $0.11 per print. The best part is that printing a large 8"x10" is only $0.44 per print and you can go up to 20"x24" for only $2.73, which is really economic. I read lots of positive comments about the quality and consistent results you can obtain with this processor.
The only concern I have is time, and I wasn't able to figure out how long is the whole process going to take. As I understand developing 5 negatives using an "expert drum" should be less than an hour in total. What I couldn't find is how exactly are the prints produced. It appears that with the additional VarioFormat Easel tool you can print up to four 4"x5" at once. The process however seems to be somewhat manual, and the time it takes is not clear.
None of the photo dealers in South Bay seem to carry the Jobo processor, the closest one I could find is Calumet in San Francisco. I think I'll pay them a visit later this coming week.
National Geographic published in the April issue an article by Rick Ridgeway about a trip to the Tibetan Chang Tang plateau, to find the calving grounds of chiru, the Tibetan antelope. The article is illustrated with beautiful pictures made by Galen Rowell, who unfortunately died in a plane crash few weeks after he returned from this trip.
A truly great story about altruistic people trying to save an endangered specie.
I've setup a gallery of pictures made since I bought my Powershot S30 digital camera. I uploaded some of the pictures I've taken in the past 6 months. Older pictures taken on film still wait to be scanned, as I'm too lazy to take the time to scan them.
Update (June 2010): I disabled the picture gallery hosted on this site after an attack exploiting some PHP vulnerabilities in Gallery. After the post-mortem I've decided PHP is not safe to be run on a web site, and I removed all the PHP code I had on this site.