In the last couple of days, two widely conflicting opinions have been published on the Leica zoom lens which ships with the new Panasonic Lumix L1 camera. Vincent Luc, writing in Réponses Photo, is disappointed with it. Not that it is bad, but he finds that the sharpness and contrast are simply not up to the expectations associated with Leica. He does, however, add that there might be some scope for improving matters in post-processing. Now, Vincent Luc is no idiot, and the review is well considered and comprehensive, nothing like the recyled PR and datasheets that most web sites pass off as "reviews". One website which certainly does not fit that in category, however, is The Luminous Landscape. Michael Reichmann, in his L1 review, has a radically different view:

"Having taken many hundreds of frames with this lens during my week in Iceland I can tell you that this is one first-rate optic. No formal tests are needed to let me know that this lens is sharp, contrasty, and quite free of any serious aberrations – at least those visible without conducting a formal test suite".

Going back to the post processing issue, it is interesting to pick up on a recent post by Colin Jago, discussion in this case the sharpness in general of Olympus E-1 images (let's just imagine that the E-1 has the fully compatible Leica zoom attached). He observes:

"(...) one of the things that you always have to bear in mind is that you only have 5 megapixels to play with. Further, these are quite soft megapixels (the anti-aliasing filter). Whilst I think that properly sharpened native resolution prints from the E1 can be fantastic, (...)".

So what is everybody actually talking about here ? First, whilst I suspect that the Vincent Luc's results are based on JPGs, I'm sure Michael's and Colin's are based on RAW. The almost diametrically opposed opinions of the lens sharpness and contrast are striking. But... is Michael talking about the results as seen (and maybe optimised) in Adobe Lightroom?

Both Colin and Vincent Luc talk about recovering sharpness lost by the anti-aliasing filter, and this where I really start to lose the plot. An AA filter is a low pass filter, usually with an abrut cutoff. It is designed to prevent the sensor from recording high frequencies which it cannot unambiguously resolve. I don't want to go into a long discussion on filtering here, but in this type of setup essentially any data blocked by the filter is gone and no amount of post-processing can bring it back. Frequencies near the cutoff frequency will be attenuated. In photography terms, this translates as an irrecoverable loss of fine detail, or more accurately, a limit on the level of fine detail that can be captured. This is obviously extremely simplistic, and people could - and do - drone on for hours about it.

Sharpening in software can give a percerption of a more detailed image, by subtle enhancement of the actual detail. But doesn't make the lens sharper or more contrasty.

The approach of evaluating the camera-lens pair using DxO's system seems to be the only consistent way to review digital systems. But when the reviewer is looking at photographic output, as the three I quote here are, then the software plays an equally important part, and should be explicitly declared.

>Perhaps we should start to talk about lenses in a different way, saying for example that on camera X, processing with software Y, lens Z does not limit resolution or inhibit contrast. Then maybe it becomes easier to understand how two highly competent reviewers can draw such different conclusions.


So, is the Leica lens a dog or a gem?

Since I have a received a good few questions about my workflow, particularly from Olympus E-1 owners, I though it was maybe time to write something about it. Note, I do all of my digital photography on Macs, so I'm afraid this workflow is Mac specific, or at least the RAW conversion part is. I use three core tools on my workflow:

• Microsoft iView Media Pro, for ingesting, preview, preliminary sorting, and cataloging


• Iridient RAW Developer, for RAW development and image processing


• Adobe Photoshop, for image post processing and output

I also use Colorbtyte ImagePrint for printing, and FixerLabs SizeFixer for scaling up. For RAW processing, I sometimes use PhaseOne CaptureOne Pro and Olympus Studio. For Lumix LX1 RAW processing, I usually use Adobe Lightroom Beta. I only use Adobe Camera Raw in special cases, for example if I want to use Photoshop's HDR tool. In all cases, you can assume I'm using the latest versions of each application.

Stage 1: Ingesting
The first part of the workflow involves getting image files off the card and into the computer. I use MediaPro to import and rename the files, and to apply a basic personal metadata template. I don't really have a solid naming scheme, but I at least rename files to include a code indicating which camera the file comes from, and the date. I let MediaPro assign a serial number. Actually, the best renaming software I've seen is Olympus Studio, which allows you to use EXIF fields as components of the file name. The important thing is to have a unique name, and to keep it throughout the pipeline. In this way, it is possible to sync metadata in between different catalogues and different filetypes (so for example the captions from E1_20060603_0091.ORF and E1_20060603_0091.JPG can be synchronized. My filing system is quite simple: I have a master hard disk for RAW files, with folders for each month. Within these folders, I create subfolders whose structures depend a bit on the nature of the shoot. Usually, I create a folder for a single day, e.g "2006_06_03" (that's 3rd June…I'm European), but I might create a folder for a multiple day "shoot", e.g. "Iceland March 2006". No hard and fast rules, just whatever makes sense. If I end up using CaptureOne, it will require its own sub folder structure to be created inside this folder – at least, if you prefer to keep things simple it will. Other RAW developers will leave settings files inside these folders too, so it is important to ensure these remain safe. I generally back up each folder to DVD – one or more, as necessary, although this is not ideal, really, since the backup only store settings files created up to that point. I also keep an automated working backup using Retrospect. I maintain two "master" RAW catalogs for Olympus ORF and Lumix Raw (converted to Adobe DNG) respectively, and these automatically update. I plan in the future to manage archive backup creation from these catalogs (using a script to record the archive volume names in each image's metadata), but I haven't got around to that yet.

Microsoft iView Media Pro

Stage 2: Preview
Once the files are organized, I create a local catalogue using MediaPro – by this I mean a catalogue of just the RAW files in the particular folder. I set MediaPro to produce the largest, highest quality previews it can create. I then use the comprehensive previewing and rating tools to decide which images I want to work further with, and how I want to categorise them (for example, I might well have "fun", "family" and, er, "art" shots from the same shoot. MediaPro 3's new Lightbox tool comes in extremely handy here, although it is very important to note that you are working with an 8-bit preview here, so the histogram for example is of the preview, not the 12bit RAW. However, it is easily good enough to show if an image is irrecoverably over- or under-exposed, and the method is fast and effective. At this point I will delete any files which are really trash, although if in any doubt, I keep them. Disk space is cheap. I end up with sorted, rated files, and I can even print contact sheets if I wish.

Stage 3: RAW Processing
For RAW Processing in the vast majority of cases I use Iridient RAW Developer (IRD). IRD, thankfully, has the good taste and common sense to be "just" a RAW developer, and has no pretensions to act as a full blown "workflow manager" (no, I'm not going to get sidetracked). IRD has drawn praise from a number of reputable sources, and is possibly the most full featured product in its category. Certainly it can't be beaten on sharpening options. Initially IRD seems very complex, but the complexity is only there if you need it, and it doesn't get in the way. It also often provides several ways to reach the same end, for example tone curves and tone sliders (I prefer curves, because Photoshop forced me to understand them, and Bruce Fraser's books explained them). IRD is especially highly rated by black & white aficionados, but, typically, I don't use it to B&W conversion. I tend to open RAW files in batches of related images, by selecting them in MediaPro, right-clicking and opening in IRD (this works as well for Olympus Studio or Camera RAW, but not very well for CaptureOne – ironically, since there is some sort of marketing alliance between CaptureOne and iView…well, there used to be, pre-Microsoft). Once the photos are open in IRD (note how I use "photo", "file" and "image" interchangeably), they can be selected from the open files drawer, and processed in turn. I won't go into detail on how I use IRD here, but my default settings for E-1 ORFs include using light "Difference of Gaussians" sharpening as "capture sharpening" – although I'm beginning to wonder if multi-stage sharpening is necessary considering the lack of artifacts IRD introduces – and using ProPhoto as the working colour space, and Joe Holmes' ExtaSpace as output space. My principal output files are 16 bit TIFFs, which I save in a temporary holding folder called "IRD Output". I also sometime process directly to JPG for print or web, but for my web galleries I use a set of Photoshop actions (coming later).

Iridient RAW developer

Stage 4: Post-processing
Currently I use Photoshop (.PSD) format as my final archive format, for fully processed images. There are several reasons for this, number one being layers, although even I'm not using any layers, I still save as PSD for consistency – if I see a PSD file, I know it is "finished", or at least it has been worked on. The second reason is that at least 25% of my output is still film-based, and my film workflow always culminates with Photoshop. I don't know of any compelling reason to change this practice. Therefore, I open all files from "IRD Output" in Photoshop, and at a minimum save them as PSDs on another disk volume, which is my "finished work" repository. Here, I simply maintain a folder for each month, and save the file into the current month's folder (I use MediaPro, not the filesystem, to catalog, so it doesn't matter if a photo I took in March 2005 ends up in the January 2006 "finished" folder). Depending on the photo, there are several things I might do in Photoshop. Generally, I will have sorted out tone in IRD, but I might run a local contrast enhancement action to see if it adds anything (I have evolved a variety, which act on different tonal ranges as necessary – usually I find excluding highlights is a good idea). Noise reduction is often required on E-1 files taken at 800 ISO or over, sometimes at 400 as well. For this I use the Noise Ninja plug-in. If I decide to convert to black and white, I use the Convert to BW Pro plug-in on a layer. I save files unflattened, although I might compact things a bit if they get out of hand. The "finished work" gets cataloged in my "reference" MediaPro catalog, where I add detail metadata, and construct various sub-catalogs and sets.

Stage 5: Output to Web and Print
Output requires sharpening and sometimes scaling. If I'm outputting to print, I take the PSD file, and set the output size as necessary. If the resolution drops a bit below 240dpi, I will scale up using Photoshop, but if it is well below, I use SizeFixer SLR. Once I get to the target size, I use Photokit Sharpener to apply output sharpening (note, for SizeFixer, it appears that sharpening before scaling is quite successful, but how Photokit's algorithms react to this, I'm not sure. Whatever – if it looks right, it is right). I then flatten the file, and save it as a copy to a temporary print folder, where ImagePrint picks it up. For my web galleries, I run an action which converts to 8bit, sRGB, then creates three different sizes of the file, appropriately sharpened, in an output folder hierarchy. The largest files are managed in a MediaPro catalog, and this is synchronized with my online mySQL database using a set of Applescripts, which glue MediaPro, MacSQL and Transmit FTP together. There are ways of using scripting additions to do the SQL and FTP parts, but they are complex, and not worth the trouble to me. The Applescript is very specific to my configuration, but I'm happy to send it to anybody who would be interested to see if they can adapt it.

So that's basically it. It takes longer to write about than to do it. The foundation stone is obviously MediaPro, which is a very powerful, but subtle application. The fact that Iridient RAW Developer constrains itself to doing one thing very well makes it very easy to introduce into a composite workflow. And Photoshop remains Photoshop... At some point I will follow up with my film-based workflow, but it isn't really so different.

For some time I've been meaning to write something here about using the Zuiko Digital 8mm Fisheye lens, specifically for landscape. This is the first true fisheye I've ever used. Many, many years ago, I used a fisheye adapter on Canon FD lenses, a combination which puts most Holga photos to shame.

The lens is very well built and the focusing ring rotates smoothly with just enough friction, making it a pleasure to use. The huge front element is very impressive, and a bit exposed. A fixed lens shade is included, and a large lens cover fits over the barrel.

Conventional wisdom claims that fisheye lenses cannot be used for landscape photography, but I think that images here demonstrate that this is not always true. Using a fisheye for landscape work requires a certain approach. First, you need to have a suitable subject, usually with a range of subjects from very near field to very far. Second, you need to make sure that any feature which you do not want to distort is centered vertically and horizontally. The “normal” rules of composition do not work with fisheye photos. Exposure is always critical, especially as it is likely that there will be a wide contrast range. Don't trust automatic exposure with this lens - check the histogram. It is best to under-expose slightly, to reduce the danger of localised flare. As with any fisheye chromatic aberration or localised flare (purple / blue fringing) in high contrast areas is going to be a problem, but with care and attention this can be avoided. You also need to decide if you're going to 'correct' the fisheye effect, using Olympus Studio v1.5 for example. Other options exist, but, for example, the lens correction filter in Photoshop CS2 is nowhere near as good as Studio. Personally, I tend to approach subjects with the idea that I am looking for a fisheye view. Although the correction tool is very impressive, inevitably the corners are very soft, and a lot of cropping takes place. However, this itself gives a certain feel which works well with some subjects.

The first shot is of the Strokkur geyser, at Geysír in Iceland. The first version is geometrically corrected and processed from RAW using Olympus Studio 1.5, and the second processed without correction. The shot directly into the sun shows how well flare is controlled by the 8mm fisheye if carefully managed. Exposure details f5.6, 1/3000th sec, -1.5EV, ISO 100, tripod mounted. In this particular case, I find that the uncorrected version works better.


geometrically corrected...


uncorrected.

The second shot is of Gullfoss, also in Iceland. Again, the first version is geometrically corrected and processed from RAW using Olympus Studio 1.5, and the second processed without correction. Exposure details f5.6, 1/90th sec, -0.5 EV, ISO 100, tripod mounted.


geometrically corrected....note very soft foreground edges.


uncorrected.

Fisheye photos are obviously on the borders of tastefulness in many cases, and in others are just illegible. But to capture really dramatic views, such as the geysir shots (I have about 100 variants of this...), they can really work. The Zuike Digital 8mm is a very expensive lens, considering that it will be used rarely, but sometimes specialist lenses give special results. Just don't try using them everywhere 😊

Following earlier posts about this, today I managed to find time to evaluate Iridient RAW Developer 1.5.1 against CaptureOne Pro 3.7.4, for Olympus E-1 RAWs. The results are clear: RAW Developer is extracting more detail and more neutral colour than CaptureOne. As a long time CaptureOne user, I'm a bit shocked... There are plenty of RAW developers (lower case) out there, but a strong motive for my moving to Iridient RAW Developer (let's call it IRD from now on) is that it has no pretence to be a workflow tool. It doesn't even try to read images from newly inserted cards, thank heavens. Therefore I can use efficiently in tandem with iView MediaPro, which is exactly what I want. To be fair, Adobe Camera RAW could also be used in this way, as can Olympus Studio. CaptureOne cannot, easily, because of its sessions concept, and obviously do-everything solutions like Aperture and Lightroom cannot. Concentrating on RAW development alone has allowed Iridient to focus effort on delivering what must be the most extensive range of adjustment tools in any program of this type. It is complex, and it does take a bit of getting used to moving from CaptureOne (for example there are 4 different sharpening methods, of which 3 are pretty esoteric). You can do an enormous amount of fiddling, but, fortunately, you don't have to - the default settings are pretty good. Although the workflow issue is important, image quality is still a prime factor. From what I've seen so far it seems that IRD has little to fear here. I settled on a test image which has detail at various scales, and also has a sample of the dreaded, Tetris-inducing bright red in it.

The full image

100% crop, processed in C1Pro v3.7.4

100% crop, processed in IRD v1.5.1

Looking at these two crops, the first thing that has to be said is that the IRD version is basically "out of the box", although I did decide to use the R-L deconvolution at low settings, as this seems to to restore sharpness lost in the demosaicing processing / anti-aliasing filter. In C1Pro, finally I turned sharpening off, because even using the very low settings I had settled on as "capture sharpen" defaults, it seemed over-sharpened. I also turned "pattern suppression" on, but I was really surprised to see the infamous "tetris effect" in various places, and not just in red. It isn't clearly visible in the JPEG here, but one area is outlined in green. As for reds, well C1Pro for whatever reason did not tetrisise (new verb - you saw it here first!) them, but it certainly over-saturates at default settings. Here I am using -20 on the saturation scale, which is quite drastic - and it still looks wrong compared to IRD. To my eyes, IRD brings out more detail, including highlight detail, and delivers a more natural colour. The R-L deconvolution sharpening lends itself very well to different levels of output sharpening. The IRD versions shows a little low level tetrisisation in the reds, but it doesn't show up in prints. I've been considering switching to RAW Developer for some time. Now that I've finally found a few hours to evaluate it, I've decided to do so. It's a great piece of software at a very fair price. I'm not saying that CaptureOne is no good. I've been using it happily for 2 years. But I doubt that PhaseOne see much point in improving Olympus RAW performance for the tiny user base which all indications show they have. I think they are more likely to focus on the higher end DSLRs and their own medium format backs in future. CaptureOne can still sometimes be a very obscure program, and the session concept is too limited to provide a full image management solution, whilst getting in the way of programs like iView. It saved me from being stuck with Olympus Studio, but for now, it is probably going to be put out to pasture...

Following Bernard's comment to the previous article, I tried giving RAW Developer and CaptureOne a better chance. Here is the IRD version, using the new R-L deconvolution sharpening (aside, I wonder if this is what FocusFixer uses ?), set at radius 1, iterations 10, as Bernard suggests. Noise reduction is off: It's certainly better, but I wouldn't say that the Tetris effect is gone, exactly. I was unfair to CaptureOne, as I'd left "pattern noise suppression", designed precisely for this problem, switched off. So here it is switched on, with my standard light capture sharpening (amount 10, threshold 1, standard look) and noise suppression off. Everything else is left to defaults. Well, it is better than before, and actually setting noise suppression up a bit helps even more. But it is still there. NOTE: in both cases, the JPG compression on these images is artificially enhancing the effect, but about 20%. Finally, does it matter if you can't see it on a print ? Generally speaking, no. In 99% of cases, no. But it can limit the scope for enlargements, and since resolution is not the E-1's strongest feature, it is worth considering. It still seems for absolute quality, Olympus Studio is best, but with a lot of caveats. Raw Developer seems to be about at the level of CaptureOne, and at a considerably lower price, with a much more dynamic release schedule, it looks like a very worthy candidate. But CaptureOne is still my first choice. Until further notice.