Thursday, November 29, 2007

Measuring Konica-Minolta DiMAGE A2 Sensor Noise

My sensor measurement series (G9 and 5D) continues. This time, I measured a Konica-Minolta DiMAGE A2, a 8MP camera which was introduced in 2004. How does it compare to a modern 12MP Canon G9?

Canon G9
ISO 64 / 80Konica-Minolta DiMAGE A2 ISO 64 f3.5 1/50Canon G9 ISO 80 f3.2 1/60s
ISO 100Konica-Minolta DiMAGE A2 ISO 100 f3.2 1/60Canon G9 ISO 100 f3.2 1/80s
ISO 200Konica-Minolta DiMAGE A2 ISO 200 f3.5 1/100Canon G9 ISO 200 f3.2 1/160s
ISO 400Konica-Minolta DiMAGE A2 ISO 400 f4.5 1/125Canon G9 ISO 400 f4.5 1/160s
ISO 800Konica-Minolta DiMAGE A2 ISO 800 f6.3 1/125Canon G9 ISO 800 f6.3 1/160s
ISO 1600N/ACanon G9 ISO 1600 f8 1/200s

Up to ISO 100 there is little difference between the cameras. At ISO 200 and above the 12MP sensor of the G9 shows more noise than the 3 year older 8MP sensor of the A2.

According to the DiMAGE A2 has a 2/3" (8.80 x 6.60 mm) sensor. The G9 sensor is a 1/1.7" (7.60 x 5.60 mm) unit. In other words: the pixel pitch is much smaller on the G9. Noise seems to be inversely proportional to the pixel pitch, so it is not surprising that the G9 does worse than the A2.

What happens when the images are run through Noise Ninja?

(Noise Ninja)
Canon G9
(Noise Ninja)
ISO 64 / 80Konica-Minolta DiMAGE A2 ISO 64 f3.5 1/50 NNCanon G9 ISO 80 f3.2 1/60s
ISO 100Konica-Minolta DiMAGE A2 ISO 100 f3.2 1/60 NNCanon G9 ISO 100 f3.2 1/80s
ISO 200Konica-Minolta DiMAGE A2 ISO 200 f3.5 1/100 NNCanon G9 ISO 200 f3.2 1/160s
ISO 400Konica-Minolta DiMAGE A2 ISO 400 f4.5 1/125 NNCanon G9 ISO 400 f4.5 1/160s
ISO 800Konica-Minolta DiMAGE A2 ISO 800 f6.3 1/125 NNCanon G9 ISO 800 f6.3 1/160s
ISO 1600N/ACanon G9 ISO 1600 f8 1/200s

Up to ISO 200, there is little difference between the cleaned up images. At ISO 400 there is slightly more noise in the G9 image, but the difference is very slight. At ISO 800 the A2 image seems definitely cleaner than the G9 image.

I am surprised that in three years there is no improvement in sensor noise. Well, there might have been if you where to create a 2/3" 8MP sensor with todays technology, but unfortunately there is no such product.

Thursday, November 8, 2007

Measuring Canon EOS 5D Sensor Noise

I continued my sensor noise measurements, this time with a Canon EOD 5D. The setup was the same as previously, the only change being the exact location of the WhiBal card.

Surprisingly (at least to me) there is a little noise evident even at the very lowest ISO settings.

Sunday, November 4, 2007

Measuring Canon G9 Sensor Noise

I recently decided that a new tool would be nice and that I needed a fairly compact camera that performs well under low light conditions. Both the Canon G9 and the Panasonic DMC-FZ18 seemed interesting from their specs. [I also looked at FujiFilms offering because in the past they performed very well at high ISO. Unfortunately, a camera with the SuperCCD and image stabilization is currently missing from their model lineup.]
Trawling the net for hard information on sensor noise for these fairly new cameras proved to be a frustrating experience. There seems to be very little information available that allows a good comparison of various cameras. Yes, I know there are many snapshots available, but the conditions under which they are taken varies greatly.
The big review sites (I personally prefer are somewhat better, but none had a review of both cameras online. In fact, dpreview had published a review on neither camera.
So I decided to create my own test setup to determine how the cameras that I have access to perform and to allow a comparison. Because I just got it, I started with the Canon G9.

Saturday, November 3, 2007

Using Lightroom with Expression Media to Add Metadata to RAW Files

In the past, I have used Capture One Pro (C1) to edit the RAW files (which I exclusively shoot) and then output JPEG files. I would then employ iView Media Pro to add metadata to both the RAW the JPEG files. I exported the metadata to a XMP sidecar file for each RAW file and archived the RAW with the sidecar XMP.

It turns out that this workflow is far from ideal: both the RAW and the JPEG files' metadata needs to be kept in sync. If I go back to create another JPEG from a RAW that has already be processed, I have to manually copy the metadata from the RAW to the new JPEG because C1 3.x looses most of the metadata during processing.

When Adobe Lightroom (LR) came out, I started using it and eventually replaced C1 with it because (among other reasons) it supports XMP metadata very well. I can now add metadata to a RAW file and it will be preserved no matter what I do with the file in Lightroom (or Photoshop, for that matter).

But here the problems started:

Tuesday, September 11, 2007


I saw some fireworks yesterday and snapped away.

[caption id="" align="aligncenter" width="800" caption="Fireworks XVIII"]Fireworks over a church tower at night[/caption]

Monday, August 13, 2007

Open Source Consumer Electronics Platform

Thinking about the business of consumer electronics, it seems that most devices are now knocked out in China by the (tens or even hundreds of) millions. It appears that a device is developed and then the entire batch is manufactured in one go. The resulting pile of devices is stored in a warehouse and pushed into whatever supply-chain exists to move the product. Once all the inventory has been moved, a new device with an updated (hardware) specification is produced, repeating the cycle.

In this sort of product cycle development is at best a nuisance the costs of which which are minimized by utilizing standard components. This can be seen with cheap MP3 players or low-end digital cameras which all seem to sport the same features at the same time as a new chip set is released.

As an aside note, the story is slightly different for a recognized brand which may build upon past devices to strengthen the brand itself. Take Apple, for example, who have a clear platform strategy for the iPod where one generation builds upon the previous generation. The iPod firmware has now reached a level of maturity where few if any updates are required and the (software) feature set seems fairly stable.

But at the bottom of the pool development carries little return past the current device. There is little evidence that the software of these devices improve over time. For an example, take low-end MP3 players, all of which seem to share a common bug and have been for years: the playlist is sorted by the order of file creation on the device. It looks like there is a chipset out there which likes sorting this way ...

The limitations introduced by this approach to development may be tolerable in a simple MP3 player or digital camera, but what about a DVD recorder? Do you remember when programming a VCR was pas the ability of many (if not most) consumers? If you want to experience deja-vu, go an purchase a cheap DVD recorder and try to reliably record shows off air using the timer. Assuming you get the media compatability mess sorted out programming the device will prove to be a challenge. And even if that works, you may find that after you have missed the third show because something inexplicable went wrong with the recording you will want your VCR back. At least I did.

Clearly there is a steep learning curve for the makers, even though other people solved all of the issues with VCRs years ago.

It would be easy to bash the manufacturers for these shortcomings (and often I do). I do think, however, that these companies are really (and I mean REALLY) good at knocking out hardware at the lowest possible costs. They are not (and most of them probably never will be) consumer device developers or software developers. That is simply a different problem domain.

So what can be done? Looking at Apple again, Apple do their own user-experience development and then farm out the software development to a specialist, at least partially. If you don't believe me, look at the credits on an iPod. Manufacturing is handled by a number of companies who specialize in this.

This is, of course, a league in which many of the class of manufacturers mentioned above may aspire to but will never reach. There is a solution however: open source the device.

Imagine this: a company designs an MP3 player with the usual software, just like everyone else. Then they sit down and write a complete manual on the hardware and the registers of the chipset that is used suitable for a programmer. They sell the player and release the documentation for download along with the tools required to upload new firmware to the device. What will happen?

My answer: how fast can you say Linux?

It is almost guaranteed that someone will create a new (and better) firmware for the device and release it as open source. Now people buy the cheap player to run the cool firmware. Sales go up and more people become interested and contribute to the software. Pretty soon the little company is selling players like the proverbial hot cakes with no software development effort on their own.

This works for a number of router and access point manufacturers, so why not for consumer electronics manufacturers?

Skymaster DVR 7400

I purchased a Skymaster DVR 7400 digital satellite receiver with time-shift capability. In the box is the receiver itself and a plastic remote control plus the batteries for the remote as well as a user manual with instructions. The receiver appears well-built and sold from the outside. The remote is a bit on the light side of comfortable and doesn't feel pleasant in my hands. It is perfectly useable, however, so I won't complain.

Hooking the receiver up was a snap: connect the satellite feed to the satellite input and one of the two SCART outputs to the TV. There is also an S-Video output as well as an optical S/PDIF digital and RCA analog stereo audio outputs all of which I did not use.

Upon turning on the receiver it automatically found the Astra 19.2°E satellite that my dish is aligned to (this is a no-brainer since it is practically the default satellite here) and tuned in the channels. Wonderful!

The EPG (electronic program guide) implementation is strange: it seems to download all data in real-time, so that it often takes a while to actually get information. There is a hard drive in the device, so why isn't the information cached there? And while I am wondering: why is the program description slowly scrolled in a separate window automatically every 5 seconds or so instead of allowing me to scroll manually at my own pace?

The EPG is optimized to show what is on a single channel (or up to 4 different channels) for the next few hours. This works well and could be improved by a mode that shows what is on all the different channels at the same time.

It is fairly easy to program a show for recording: a simple button press and after that a confirmation is all that is required. It is less easy to see what was actually programmed afterwards, you have to scroll through pages of individual timer settings which do not seem to have the show title set reliably.

The recording seems to consistently start 2 minutes late. The only way to work around this that I could find was to manually set the start time earlier.

Playback works fine some of the time. I had one show which started skiping about 25 minutes into the recording. There where short (1-5s) pauses in playback every minute or so. After 30 minutes the receiver froze and had to be rebooted to accept any input at all from the remote. This behavior was consistently repeatable.

I had recorded the same show off the same satellite feed using the ElGato EyeTV and everything was fine. The problem was clearly not with the feed. I suspect a hardware or software bug in the receiver.

The manufacturers website did not offer any helpful advice. There was a kind offer to call a premium-rate number for support, an offer which I politely declined, as I always do. The receiver was bought as an entertainment device. I do not find bugs to be entertaining, quite on the contrary. So I gave the device back and will spend my money on something else.

2 stars out of 5.

Monday, June 11, 2007

Country Road

A snapshot while taking a walk. This picture (or at least the full resolution image) illustrates why it is not good enough to have a camera with you at all times, it really should be the best camera that you have. If you don't have a clue what I am talking about, have a look at the three birds in the picture. What three birds? Exactly! With a better camera you would see the birds in their full glory.

I'm still posting the image because I like the light summer feeling which it captures.

[caption id="" align="aligncenter" width="480" caption="Country Road"]Country road with a tree and blue sky[/caption]

Thursday, June 7, 2007

Review: Elgato EyeTV 310

I purchased an Elgato EyeTV 310 which is a digital satellite tuner and personal video recorder for the Mac. It will decode DVB-S which is what most(*) digital TV and radio satellite feeds are. The feeds can be saved to the hard drive and time-shifted or edited and/or converted and burned to CD or DVD.

(*) Most are DVB-S. Most high-definition feeds are DVB-S2 (which uses MPEG-4 instead of MPEG-2 for compression) which the EyeTV 310 does not decode. If you want to receive HD satellite feeds, this is not the product for you.

The EyeTV 310 is a small box that has a connector for the satellite feed, power, and two Firewire 400 (IEEE-1394) ports. Plug it all in, install the software from CD, and get going ... Except that when I did, nothing happened. The EyeTV did not find any channels.

I knew the satellite feed was good because it worked with a stand-alone receiver. I tried various settings, but nothing helped. So I contacted Elgato tech support using the website. Within less than a day, I had a knowledgeable, friendly reply. No boilerplate, no automated junk, a real human tech-support contact. I was pleased.

What the tech suggested did not work, so I made some screen shots and saved some log files, wrapped them up in a .zip archive and sent another email. Within a minute of sending the email, my message had been entered into the tracking system on the Elgato website. This time, I was amazed.

I though Elgato would fumble the ball on a same-day response, but I got an answer later that evening from my original support contact. No playing hot potato, the guy who knew the case kept it. The message was friendly and again very helpful. Someone clearly knows his stuff here!

The tech suggested that I try eliminating all the pieces of equipment starting at the LNB to find out what was causing the problem - he thought that despite the stand-alone receiver working the EyeTV was not getting a signal.

O.K., I will admit it: I was skeptical. After all, I had seen it work, right? But the next day, I sat down and did what I should have done from the start, I eliminated the possible causes for the error one by one.

Guess what? It turns out that I was using a bad cable. Swap the cable and the box works like a charm. At this point, I was ecstatic, and I let the tech know.

So how does the EyeTV perform? On a lowly Mac mini Core Duo with 1 GB RAM, the CPU utilization is just above idling when I watch a TV feed or listen to a radio feed. The picture quality is excellent, as you would expect from a digital feed. There is no discernible lag in frame rate, decompression artifacts, or other weirdness.

Most DVB-S feeds come with program information. In a view slightly resembling iCal, all the channels and programing are listed. A simple click on an entry brings up additional descriptions. To schedule a recording, a single button press is enough. There is nothing else to configure.

EyeTV also comes with a one year subscription to an EPG (electronic program guide) service which should list all feeds, even if they do not broadcast program information. I have not tried this because the Mac mini is offline.

Once a feed has been recorded, it becomes available for playback, editing, or archiving. If you happen to have Toast, EyeTV will pretty much automate burning a feed to DVD. That's a nice touch.

There are some things the EyeTV software could do better: there seems to be no way to show all program entries of a certain type (say all movies or all documentaries) in the EPG view, for example. I would also have liked the level indicator to correctly show that the EyeTV was not getting a signal when I was troubleshooting. It would be nice to have more options in organizing the channel list.

But by and large, this is not just a good product, it is excellent and though much more expensive than a Windows solution well worth the money. I was (and still am) impressed by the support I got. Good support is hard and not cheap, but Elgato seems to go the extra proverbial mile to deliver it.

Saturday, March 31, 2007

Canon and the Customer Experience

I read an Article on the Canon ipf5000 printer that details problems the author and members of the canonipf5000 wiki have experienced with the printer. It comes to the conclusion that the print quality is excellent, maybe the best of any current printer, but the software, documentation, service, and warranty terms are somewhere between mediocre and unacceptable. The recommendation is therefore not to buy the printer in the USA.

This got me to thinking about my experiences with Canon products.


I have owned (and still own) several Canon cameras in the last 25 or so years, most of them SLRs. I own no compact digital cameras from Canon. The documentation has been and is excellent, leaving nothing to be desired. The warranty terms are fair. I can not comment on the service as I never had a problem with any of the cameras that required that I contact Canon. In short, I am extremely happy with my customer experience for the cameras.

The software for Canon DSLRs is a different matter, however. DPP is simply horrible for me. The user interface and the non-support of a decent workflow are both show-stoppers. There are better alternatives.

Interesting also is the Canon EOS Utility, which allows you to connect a Canon DSLR to a computer. From Mac OS X 10.4.4 to 10.4.7 the utility simply failed to work. It appears that Apple changed something in the way the OS handles USB which in turn broke the Canon software. I do not know who is to blame here, but I do know that I would have expected Canon to find a solution. They did not. Telling me to call some support hotline to talk me through the installation was not helpful either.

And while I am griping: the following is in the release notes to EOS Utility Updater 2.0, released in March 2007:

EOS Utility is not compatible with Intel-based Macs (Mac OS X 10.4.4 to 10.4.6).

What is that supposed to mean? Does it work on Intel-based Macs using an OS >=10.4.7?


All of the Canon printers that I have owned have worked more or less as expected. The hardware is definitely good. I have never used anything but the included printer drivers, all the fancy additional software is lost on me.
The manuals vary in quality, fortunately I do not need them often.
In short, I am happy with what I got but not thrilled. Canon used to be my default printer brand simply because they did not try to lock me into buying Canon inks. But now that Canon has started chipping their ink cartridges to prevent me from buying OEM inks, I will start looking at other manufacturers in earnest again.


I have owned several consumer scanners from Canon. The first, a CanoScan LIDE 30 worked fine, but the OCR software did not install on my Mac. No amount of fiddling would do the trick. The support information from Canon did not help.

I bought another scanner to use at my home office, a CanoScan LIDE 25. Unfortunately, if the software for the 30 is installed, the software for the 25 can not be installed in parallel. It simply will not work. It is also not possible to uninstall the software for the 30. This basically forces you to re-install the operating system when switching from one scanner to another and prevented me from using both scanners in parallel, like I had planned. No help from Canon on this.

I also have a Canon Pixma Printer-Scanner device. Yes, you guessed it, it requires yet another different software. The Pixma supports some of the features of the stand-alone scanners but not all.

My mental picture is that of a different software version for every device with no thought to building an integrated solution using a single code base.

Concluding Thoughts

Ideally, a consumer device just works. Take it out of the box, attach some wires, maybe install some software and everything else is milk and honey.

If you do not reach this level of quality, you must build a safety-net system that compensates. Such a safety-net is your documentation, tech support, and - if all else fails - warranty terms.

If you do not get this safety-net right, you will have unhappy customers which eventually cost you sales.

Examples of products that fit into the first category (just works) are: Apple iPod, most Nokia cell phones, Canon 35 mm SLR cameras, and - mostly - Canon DSLR cameras.Canon printers that I have used are also in this category, but clearly some people feel that the ipf5000 belongs in the third category.

Canon consumer scanners are in the second category. Most Canon software seems to be in the last category.

I wonder if there is a pattern here? Is it possible that Canon is really good at making devices but less than stellar at writing software?

Perhaps this could be the basis of a rule as to which products one should buy and which ones to avoid:

"Buy anything where you can get decent 3rd party software. Avoid anything where you will be stuck with the software it came with."

And there is an upside for Canon to this: partner with someone who writes great software and increase the bottom line.

Disclaimer: I have a large number of products from Canon and I really love a lot of them. I think that Canon is a great company and they do well at what they do. I also think they could do better wherever software is an essential part of the product.

Tuesday, March 6, 2007

Parallels Desktop for Mac and International Customers

In order to run Microsoft Windows software on my Intel Mac, I use the fine software Parallels Desktop for Mac (referred to as "Parallels" from now on). It runs everything I have thrown at it up to now and it is getting beter and faster with each new release.

The only annoying technical limitation is that Direct X does not work, so modern games are out of the question. But who needs games on a work machine, right? :D

There is a problem with Parallels, Inc. and the way they handle their overseas business, however: we are being treated as second rate, at best.

Let me explain: "Paralles" (the software) is not localized in a technical sense. Customers anywhere in the world get the same American English version of the software. This is no problem for me and it should not be for anyone because Parallels, Inc. makes no claim of language localization.

However, the activation key used to activate the software is different in the U.S. and elsewhere in the world. I have no idea why this is so, but I assume that it is because of some business problem this approach fixes. Again, I have no issues with this.

What I object to is that beta versions of the software, which Paralles, Inc. makes available months before a new release (again, a move that I applaud) require an activation key which must be a U.S. activation key. International keys do not work. International users must request a new demo key every 30 days until the system locks them out for requesting too many keys.

Do international customers have less of a need for new features? Have international users paid less than U.S. users to justify this treatment? Are international users challenged in some way when dealing with beta software?

But wait, it gets better: you also can not upgrade to the latest final version of the software until several days after the U.S. customers get to do so. And (as opposed to the beta version) Parallels will not tell you that this is so.

I downloaded the newest version (1986) right after it was released on 27-FEB-2007 because an email message from Parallels, Inc. urged me to. The new software simply rejected my activation key. So I contacted Parallels,Inc. support using the form on their website. I made it clear that I am a paying customer (I gave them my key) and requested assistance. I am still waiting for a reply today (05-MAR-2007).

A few days later Parallels, Inc. quietly published a different version of the software that is called "Parallels-Desktop-3186-Mac-uk-AQ" and only available using the built-in update function, it seems. This version installs and activates without a hitch. Unfortunately, I was not told about it by Parallels, Inc. like I was with the U.S. version and I could not find this version on the website.

All of this leads me to the conclusion that for some unfathomable reason, international customers are only second rate customers in the way Parallels, Inc. treats us. I can only hope that this is due to a flaw in the way the company handles it's processes and not a deliberate decision.

I'd really like to hear from a representative of Parallels, Inc. on this issue, but - alas - I have not received a reply yet.

Sunday, February 25, 2007

Early morning fog

It's been a while since my last post. I have been busy doing other things, non-photography-related, I am afraid.

I had the opportunity to take a morning walk down the Isar river. The sun was just out and fog wafted over the water.

[caption id="" align="aligncenter" width="800" caption="Morning Fog"]Morning fog over the river[/caption]