14 January 2010, Microsoft Internet Explorer

Have you ever tried writing CSS and HTML-code for a webpage that is supposed to look exactly the same in all the major internetbrowsers: Internet Explorer 6 (IE6), IE7, IE8, Firefox and Safari. Why does it always work the way it is supposed to work in Safari and Firefox the first time. And why does it never work the first time in IE6, IE7 and (to a much lesser extent) IE8? Microsoft is driving me nuts. They do not seem to know the meaning of words like "standards" and "consistency". Every other year they make something knew that is guaranteed not to be compatible with previous generations of their own software and does not stick to the W3C standards. In fact I am so fed up with it that I have decided that I will no longer guarantee that this site works in IE6. For those of you who still use it, perhaps it is time to upgrade to IE8 or (even better) Firefox.

14 January 2010, Some more images from Schiermonnikoog (Canon S90)

I have been using the Canon S90 for some time now. Here's what I like about that camera and what I like less about it.

What I like

• small size, light weight (175g); it's always there in your pocket when you need it;
• surprisingly solid;
• very good, large, clear LCD screen comparable to the screens of high end consumer and pro DSLRs;
• useful zoom range; from reasonably wide (28mm) to moderately tele (105mm).
• fast f2.0 aperture with decent image quality at the wide end; stopping down may be done for more depth of field, but is certainly not necessary to improve overall sharpness; together with the Image Stabilizer the fast lens enables you to shoot handheld at ISO 80-200 under almost all circumstances (avoiding the still rather strong noise of ISO 400 and higher);
• sharp and detailed 10 megapixel images that are good enough for 12 x 16 inch prints easily (but ... see below);
• almost unlimited depth of field. Nice for landscape photography and for a stunning effect in wide angle macro shots; This is something that cannot be achieved with cameras that have a larger sensor;
• hardly any distortion once you zoom away from the widest settings;
• raw capability;
• easy to use for any photographer who knows how to shoot a DSLR in A-mode; there's an aperture ring around the lens and a +/- dial at the back. Hardly any need to go into the menus during shooting especially if you shoot raw or raw + jpeg. This is a big plus; trying to figure out complicated menus during shooting just hampers the creative proces. If you prefer the simplicity of shooting a Leica M9 you can have it with the S90 (image quality is in an entirely other league though);

What I like less

• slow f4.9 aperture at the long end; but stopping down is not really necessary;
• shallow depth of field for portraits is impossible to achieve;
• no (electronic) viewfinder; I don't really miss it, but I know some of you will;
• lower corners of images at the widest setting (6mm, equiv. of 28mm) are unsharp under certain circumstances; particularly in landscape shots with foreground detail; less suitable for large prints if that bothers you;
• unsharpness in the lower corners at wide settings is worsened by distortion (and even more by correcting the distortion);
• manually correcting the distortion is difficult. The distortion is a little complicated; first order correction alone is not enough; the horizon in the upper corners always keeps drooping a little. Canon partially solves that by simply cutting of a little more of the sides in their jpegs and partially by not correcting the distortion completely.
• correction of distortion considerably diminishes the angle of view. I do not know if that means that the final result is not a 28mm image but a 32mm image instead. Canon may have designed the lens in such a way that it is a 28mm after correction of distortion and cutting of the edges;
• for fast raw conversion (automatic correction of the distortion) there is a limited choice of conversion software at the moment. Canons own software is not very attractive. Some third party raw conversion software cannot correct distortion at all;
• clearly some noise visible in the darker parts and the dark skies of the raw files, even at ISO 80. However, a little (sometimes selective) noise reduction does solve the problem entirely without really affecting sharpness. Canon no doubt does the same with the in camera jpegs.

The bottom line is that this is a very good camera but that it (necessarily I would almost say) has its limitations. It is not a Nikon D200, but it can deliver images of similar quality if you keep in mind those limitations. You have to decide for yourself, considering the type of photographs that you make, if this is a good camera for you. A Panasonic Lumix GF1 and a Lumix f4/7-14mm wide angle zoom will no doubt get you wide angle landscapes of a much higher quality (sharper corners, more resolution, less distortion). But then again, try to stuff that lens and camera combination in your pants pocket ... Do you need the extra quality if you shoot for yourself and not for a client?

The images below are all from Schiermonnikoog Island and taken with the S90. Settings: A(perture priority), ISO 80, raw + jpeg, IS continuous; raw files converted in Bibble5 (exif is in the files)

6 January 2010, Canon Powershot S90

I have spent most of the Christmas Holidays on the nearby Schiermonnikoog Island. Usually it's a quiet place in the Wadden Sea. The Island is a national park and the area as a whole is on the Unesco World Heritage List. During the Christmas holidays, however, it can become quite crowded. For the first time in many years we are experiencing a real winter here with frost and lots of snow, so there were many photo opportunities. I brought the Panasonic Lumix G1 and the recently bought Canon Powershot S90. The Nikon D300 stayed at home. You cannot take your car to the island, so the amount of luggage is limited to what you can carry on your back and in your hands.

Overseeing the photos that I made, the conclusion must be that from a photographers point of view I made the best shots with the Canon S90. A simple camera, less than 200 grams and the size of the palm of my hand, with a limited choice of focal lengths. Consequently there was nothing to worry about but the images themselves. Photographers often carry around a lot of excess luggage (not just literally) that tends to get in the way of making pictures.

The S90 has ten megapixels on a small 5 x 7mm sensor, an f2(!)-f4.9/6-22.5mm lens (the equivalent of a 28-105mm on full frame) and at its base ISO of 80 offers the image quality of a Nikon D200 (I am not kidding). ISO 80 may not seem a lot to you but you have to remember that a 6 mm lens on such a small sensor has an almost unlimited depth of field, even at its widest aperture of f2. At apertures between f2 and f4 you ought to get fast shutterspeeds under all but the worst circumstances. And even if they are a bit slower the optical stabilizer (IS) turns out to be quite reliable. Canon is clearly targetting the market of serious photographers with this camera. It has raw capability and although it may look like a plastic toy on the pictures that you see on the internet and in brochures, it has a surprisingly solid "metal" feel to it once you've got it in your hand (it IS mostly metal). It was made to last for a long time even under intensive use.

I have done a little experiment for you with one of the raw files. I shot raw + jpeg. The first shot below is the uncorrected raw file. The distortion is incredibly high, as you can see. It's hard to imagine images taken with this camera that could do without correction of the distortion. The second image is the jpg from the camera. The Digic 4 chip in the camera has corrected the distortion almost completely, cutting off a large part of the image in the process. The camera jpeg has a magenta-red color cast, probably caused by the setting sun, that I find hard to correct. But perhaps the colorcast is not much of a problem on uncalibrated LCD screens that are usually a lot colder (bluish). Of course the jpeg result will also depend a lot on the "my color" setting that you choose in the menu. You would have to experiment with that. In this case it was simply set to "off". The third photo is the raw image converted and corrected for distortion in Bibble5. The red area indicates the angle of view that you gain compared to the camera jpeg by shooting raw (nothing dramatic). But compared to the uncorrected raw image both lose a lot on the edges (focal length/angle of view may differ as much as 2 mm).

Follow this link for a larger version.

17 December 2009, Black and White from Bibble5

Last night it started snowing for the first time this winter. And it still is snowing. Almost 40 centimeters by now. For this part of the world that is quite extraordinary. We don't get that much snow and if we do it is hardly ever more than five centimeters melting away before you eyes in a matter of hours. I took my Panasonic Lumix G1 out into the blizzard to get a few shots. I don't think it can be said to be weatherproof, so I was really careful. Changing lenses was out of the question. I stuck to the 7-14 wide angle zoom.

Bibble5 has a function that allows you to apply "presets" to your images, "B&W simple" is one of them. Since the image was almost monochrome anyway, I tried that setting. Also very useful is the similar function called "look profile" on top of the first tab. You can choose between settings like portrait, wedding, product and event. I find the portrait look very useful. One of the complaints I had about previous versions of Bibble was the "red faces". Well, the red faces in portraits are gone if you choose the protrait look. It saves a lot of time.

12 December 2009, Snow shower and Bibble5

I have been a user of the Bibblepro raw conversion software ever since version 3 in 2004. In the past 8 or 10 months I was becoming increasingly dissatisfied with Bibblepro version 4.10. Bibble Labs has been promising the release of Bibble5 many times in the past one and a half year and has postponed its release just as many times. In the meantime version 4 was no longer maintained, meaning that we got no raw support for new cameras (my Panasonic Lumix G1) in Bibble 4 and on top of that my Linux version started crashing more often by the month. I am running a 64-bit version of Ubuntu Linux on a "home built" desktop machine with a 2.66 Ghz quad core Intel processor and 4Gb of fast DDR2 RAM. Of course in Linux there are a few reasonably good open source RAW converters (RAWtherapee and UFraw) and another one (LightZone) that is not free and looks promising at first sight, but I never really managed to get used to it. My main complaint about LightZone is that it's slow. It's OK but slow. I simply missed the speed of Bibble. Just a few mouseclicks and it almost always looked good instantaneously.

The day before yesterday I downloaded the 5.0/PV3.2 beta version of Bibble5. To my surprise, I had become used to Bibble always crashing on my machine, it works very smoothly with a host of different raw files. Even with files from only recently launched cameras like the Canon S90. No crashes and very, very fast. It doesn't have all the promised features yet and it even misses some of the features that Bibble4 had, but this is a step in the right direction. Here's a screen dump. I can get tiffs and jpegs from raw files in an instant that do not need any further major tweeking. Control over the final result is very precise. Exactly what I need as there is still no Photoshop like 16-bit image editor for Linux. There are versions of Bibble5 for Windows and MAC too, so you might want to try it. The beta is free for existing Bibble4 customers right untill the final version of Bibble5 is launched (whenever that will be! Don't count on March 2010). For others it is also free, but they will only get a trial license for two weeks.

The image below was converted using Bibble5, exported as a jpeg and only resized and slightly sharpened after that. In order for the snow flakes to draw these white lines in your image you need a slow shutterspeed and a tripod (wich I didn't have with me; image stabilisation saved the day).

22 November 2009, New Cover for November

I have updated the cover photo to November. It was shot last year in the Hoge Veluwe National Park on a rainy day with very little light (as dull as dull can be). So I decided to experiment a little. I pressed the shutter and then slowly moved the camera up and down during the one second exposure. The result of this is hard to foresee exactly. Just try, evaluate the result on the camera LCD and try again adjusting the movement of the camera. What I can say is that there is a clear difference between moving the camera already before you press the shutter and moving it after you have pressed the shutter (which is what I did in this case). Follow this link to have a look.

20 November 2009, Canon S90, raw versus jpeg

After a short deliberation I got myself a Canon S90 (and not the G11 I originally intended to buy). I decided that in terms of size and weight the G11 came too close to my Panasonic G1. I wanted something pocketable. The S90 has the same sensor as the G11, but a 2.0-4.9/28-105mm lens instead of the 2.8-4.8/28-140 of the G11. Canons new 10 megapixel sensor ought to be much better at high isos than the 12 megapixel sensor of the G9 and the almost 15 megapixel sensor of the G10. I have no experience with the G10, but the iso 400 images of the G9 were practically useless.

I had read a lot of confusing reports about this new sensor on the internet. Varying from "not much better than the G10" to "it looks like a DSLR". Well, at iso 400 it certainly does not look like my Nikon D300 at the same speed, but it is much, much better than the G9. In most circumstances at iso 400 it is probably good enough for an A3 print. With the G9 anything larger than 4 x 6 inches was too much really.

Last weekend I took the S90 on a short trip to Münster in Germany. The image below was shot in the Münster Dome. I used the canon S90 at 28mm (equiv.), f2.8 and 1/13 sec, iso 400. The S90 has an optical stabilizer, so images of stationary objects ought to be sharp at such slow shutterspeeds. I took three shots and they were all sharp. It certainly helps that this lens, at least at the wide end, has a very fast aperture of f2. I used f2.8 in this case but I don't think using f2.0 would have had serious implications for image quality.

Speaking about image quality. I shot raw + jpg because I wanted to compare the two. Since the camera is so new, it wasn't easy to find a raw converter for my Ubuntu Linux desktop capable of converting the raw file. I had to download the latest version (0.16) of UFRaw. The documentation said it supported the G11, so I figured it would probably be able to convert the S90 files as well. And it was. I could have used the Canon Software on an old Windows laptop of course, but I wanted to see the raw results without the automatic manipulation Canon has built into its own raw conversion software. The image here was converted with UFRaw. It shows quite a lot of distortion compared to the jpeg from the camera, confirming that Canons jpegs are corrected for distortion in the camera.

Is Canon cheating? Some people may think so. You make a bad (cheap) lens and then correct the image digitally (sacrificing resolution in he process). However, I am not so sure. This is a very small and also very fast lens (f2.0) with a very short focal length (6-22mm). Making such a lens without the distortion would most likely require a very complicated construction (a lot more glass!). I doubt if the camera could be as small as it is with such a lens and I think the lens design would almost certainly degrade the image quality in other respects (less sharpness). So, perhaps it is not a bad idea to make a small, very sharp lens, accepting that it has a lot of distortion that has to be corrected digitally. It may very well be that the sharpness sacrificed by correcting the distortion in software is less than the sharpness sacrificed if you try to correct the lens itself. Assuming that the latter is possible at all in such a small design and against a reasonable price.

You can download full size samples of the image (caution 2,5 MB files! Use a fast connection):
1. camera jpeg
2. jpeg directly from raw (UFRaw)
3. jpeg from raw (UFRaw + corrected, sharpened and moderate chroma noise reduction)

The first one is the camera jpeg, the second one is the uncorrected UFRaw image and the third one was corrected and sharpened in Raw Therapee. The corrected raw image is clearly a little sharper (and noisier, but it's mostly luminance noise in the dark parts), exposure is better balanced (the highlights are not as blown out as the ones in the jpeg) and it has lost less of the original image at the edges. The in camera correction of distortion also seems to cut off quite a lot of the edges of the image, making the lens considerably less wide (about 15%!). Move your mouse over the image above to see just how much you lose of the window on the right side of the image. You should keep in mind that the raw workflow I used was very unusual. The updated version of Raw Therapee will likely do a better job than UFRaw and so will ACR and Apple's Aperture, getting even more quality out of the raw file.