Lumière d’Argent

I am reworking my galleries. They were sorely lacking and an article from Colin Jago got me thinking. I set out to make something about it and looked at the different gallery softwares included in Fantastico. Among them Gallery seems the most appropriate to what I wanted to do. Among other things, it allows me to nicely solve the size problem. Each photo is displayed at the size of the browser window. If you make the window larger the photo grows with it. So each user can take advantage of its screen while no user will be presented with too large a photo. Nice!

There is also now the possibility to subscribe to a RSS feed for galleries.

There are other features I don’t use yet. I have first to convert all the galleries. The Luxembourg one is already done.

Stephane :: Feb.28.2007 :: Software :: No Comments »

Aperture is annoying, but… it is such a great program! LightRoom provides strong competition but does not manage to take a definite lead.

LightRoom is much better than Aperture where Aperture fails: speed and variety of RAW support. It also has generally better adjustments, much like what would be expected from Aperture 2.0.

Aperture is better at… everything else. At least that’s the opinion that’s slowly emerging of 2 days of working with them side by side. When I say better, everything is debatable, of course, but at the end of the day Aperture is the nicer application to work with.

So, where does it leave me with my unsupported Sony DSC-R1? I still don’t know, really. I’ll see if I can repair the DNG hack that used to work or if I can live with converting to TIFF before importing to Aperture. It might work.

Stephane :: Feb.27.2007 :: Software :: 4 Comments »

I have been using Apple Aperture for 8 months now. I use it on a MacBook with 2GB of RAM. A minimum machine in Apple-speak, but with respectable power by any standard. All the software I use on it is quite fast, including PhotoShop CS.

With Aperture, however, I do have performance issues, mainly with the image adjustments and external files generation. Patching and straightening are unusable. Highlights and Shadows makes the image rendering incredibly slow. Generally speaking, using the sliders is impractical because of the unresponsiveness. It does depends on how many adjustments, and which, are used. For example, once cropping is used, all the other adjustments become very slow. Patching is just usable if it is the only adjustment used and there aren’t too many spots. And external file generation takes an inordinate amount of time, the more adjustments the worse.

I expected something like that, although not as bad. I expected it because I knew Aperture adjustments are based on CoreImage. As you can see on the Apple page if you clicked on the preceding link, it is advertised as “Ultra-fast, pixel-accurate”. Pixel-accurate, I have no doubt. But ultra-fast, hmm, not on my machine. On that page , Apple emphasizes heavily the use of the GPU. The idea is that if the GPU is capable to do an operation, it is used because it will be much quicker than anything else. Otherwise, the CPU is used with a highly optimized program to still get good performance. By providing that kind of service in the OS, Apple claims software writer save time by simply using the library instead of rewriting everything.

Enter LightRoom.

Because of different frustrations with Aperture, I was waiting for the Adobe LightRoom 1.0 release to try it and see what to do. Now it is available. So I downloaded the trial, installed it, exported masters from a project in Aperture and imported the files in LightRoom. First thing I noticed was the general user interface is less intuitive than in Aperture. Nothing unexpected here, Apple is a master at user interfaces. The second thing I noticed was the speed in adjustments. Wow! This is incredible. Much faster than even PhotoShop (running in PowerPC emulation on my machine) ! Nothing to compare with similar adjustments in Aperture. Spotting is fast, interactive, works better. Very useful in LightRoom while useless in Aperture. Straightening and cropping is instantaneous, very well implemented, intuitive to use, very effective while useless in Aperture. These alone are two very important tools and would be enough to think. Then the exposure, contrast, curves, colours, all the other adjustments are between very fast and acceptably so. Many adjustment go much further than the similar ones in Aperture. But most of all, it is fast!

But wait a minute. LightRoom does not use CoreImage. It is old obsolete CPU-based programming. So, this means the CPU-based LightRoom functions are much faster than the same functions in CoreImage. It also shows Aperture adjustments could be much quicker if they did not use CoreImage!

And then, my main gripe about Aperture is not even its adjustments slowness. It is its very selective list of supported cameras. Aperture, like LightRoom, is intended to maximize the benefits of using RAW format. Problem is, Aperture only understands a very limited number of RAW formats. If your camera is not on the list, you’re out of luck and Aperture turns into a mere cataloguing application. Same thing with LightRoom, but LightRoom supports all the cameras I can think of. Big difference. But is it really an Aperture problem? No. Actually, RAW decoding is provided by CoreImage! On this front, the situation is getting more hopeless with each new camera released. Only the ones promised to be blockbusters get support. It has been expressed many times in many places how short-sighted this is. It is the first reason for an Aperture customer to look at the competition, but there is no sign of any improvement from Apple. And no communication, no plan, no roadmap, no timetable, nothing. Aperture is sold as an application for professionals. Right. I don’t know if Apple would tolerate a similar behaviour from its suppliers, but this is unacceptable. It is unacceptable to have the camera choice dictated by Apple and it is unacceptable to be kept in the dark about what to expect and when.

So, the two biggest flaws in Aperture are actually CoreImage flaws. One could even argue the other main part of the application is top notch. Indeed, its cataloguing functionality is, for me, the best on the market. Very well designed, implemented and.. fast!

Some might say that’s very good. The next OS version will surely bring a brand new CoreImage that will solve its problems and Aperture will inherit the improvementS. Maybe. But, somehow, I doubt it. Because on the OS side, there is no perception of a big CoreImage problem. Why? Because no one uses it. Indeed, why would a third party developer inherit slowness and/or very partial RAW support?

So, Aperture team, when will you dump CoreImage?

Stephane :: Feb.25.2007 :: Software :: No Comments »

Aperture is driving me nuts. It is so good, so full of good ideas, so well thought out, it is almost miraculous. Yet, Apple misbehaves so badly on important details, it is almost killing it. The problems I have are RAW support, performance and printing.

RAW Support

Aperture was designed and is marketed primarily as a RAW files management, workflow, do-it-all solution. Emphasis on RAW files usage is very strong. From there you’d think the list of supported cameras would be at least as good as the one supported by PhotoShop, right? Wrong. Aperture’s RAW support is very limited. Compared with Adobe’s list, it shows that while Adobe tries to be very current and neutral, Apple prefers to support what they think are the main cameras. This makes Aperture is the RAW tool with the most limited support on the market. Consequence is that photographers cannot really trust Apple on that regard. Many photographers use Canon and Nikon cameras but many among them have secondary cameras, often not best sellers. For those photographers, Adobe’s approach is much more appealing.

That would not be such a problem is Apple would support the DNG format. Apple says it supports DNG. Well, that’s corporate Newspeak at best. It support DNG files that are the result of conversion from RAW files it supports natively. In other words, what Apple calls DNG support serves no purpose other than posturing like supporting something without actually supporting it. It is either colossal corporate stupidity or a bad example of the “Not Invented Here” syndrome. It brings no advantage to Apple nor to its users and it brings misery to both. Apple sells less software (and related hardware) because of it, and users are restricted in choice. Besides the 5D, I also have a Sony DSC-R1, another excellent camera, very well suited to landscapes. Alas, the R1 is not on the Aperture list. Never mind it is on the list of all its competitors. The very sad thing is that the R1 will probably never be supported by Aperture because it is now discontinued and Apple is facing a growing list of more recent cameras to support.

Actually, when I complain about Aperture limited RAW support, I’m not really explaining what is going on. Aperture does not support RAW itself. It relies on the MAC OS X operating system to do it, more precisely a component called CoreImage. The idea of putting RAW support in the OS has many advantages, like offering it as a service to any developer who might be interested. Problem is, Apple updates the OS much less often than what would be needed to keep pace with the digital cameras market. Never mind that in the process it is killing the rationale for putting RAW support in the OS: What developer would be interested in having those limitations in his application? Especially when a nice free alternative exists! So, in the end, putting that function in the OS does not work. Please Apple, put that back where it belongs: in the Aperture team.

It could solve 99% of the problem by offering true DNG support like the excellent LightZone and RAW Developer programs. But no, too simple an idea, probably.

Performance

The second problem is performance. Here again, Aperture relies on CoreImage for its image alteration functions. Another great idea. CoreImage is wonderful in what it can do and in the quality of results it provides. It is also wonderful when supplied with proper hardware, namely a powerful GPU. The idea in CoreImage is that, when the GPU can process the function, it is used. Otherwise, conventional CPU-based processing is used. The choice is transparent, proving the user with the best of both worlds. Theoretically. The way CoreImage is implemented, it is indeed very fast when it can use the GPU, but it is much slower than the competition when it has to rely on the CPU, which is the case on my machine, the MacBook. There is no other reason for this other than laziness. Apple programmers are at least as capable of others to program fast graphical routines, as they have repeatedly shown in the past. End result: Aperture is slower than it should be on my hardware.

Printing

This is different. It used to be nice. Not as well featured as the LightRoom printing module, but enough to meet my needs. Used to. Since release 1.5.1, a bug has appeared. When you reduce the size of a picture on the page by telling Aperture to put a white border around it, instead of downsizing the picture, Aperture print now crops it! In a flash, they have rendered their whole printing module useless. This shows bad quality control, for a start. Not too bad, it is probably a simple bug to correct, so by now, one would expect Apple to have acknowledge the issue promptly and provided a fix. No. Not even an acknowledgment. This is the most serious issue, for me. Not just because it means I can’t really print with Aperture, but because it shows an attitude incompatible with the professional market Aperture is targeted at.

Conclusion

I keep using Aperture, partly because it remains the best in its class despite the problems, and partly because it is still in its early stage and I’ll see how it evolves. However, yesterday I downloaded the LightRoom Beta 4.1 to check where Adobe is in their development. It looks better than before. Apple has some time before Adobe fully catches up, but when that time eventually expires, I’m afraid many Aperture users will give a long hard look at the competition. This is so stupid.

Stephane :: Nov.30.2006 :: Software :: No Comments »