Lens corrections

During pre-launch briefings for the T, Leica was very keen to stress the optical quality of the new lenses. Most interestingly, we heard that they relied on optical corrections, rather than software, so as to project the best possible image onto the sensor.

So we were a little surprised to note an interesting little message when processing the camera's DNG files through Adobe Camera Raw: "Built-in lens profile applied: Vario-Elmar T 1:3.5-5.6 / 18-56 ASPH.. Click for more info." Clicking indeed gives more details:

According to Adobe Camera Raw, the Leica T's files include lens profile information for software correction of distortion and lateral chromatic aberration.

Because the profiled corrections are mandatory in ACR, we decided to look at the underlying data using a raw converter which won't apply them by default. For this we chose RawTherapee (4.0.12): a nice GUI around built around the DCRaw converter. The results are pretty interesting: Leica is clearly applying in-camera software corrections to the image projected by the lens onto the sensor.

Barrel distortion correction (18-56mm at 18mm)

Here we're looking at distortion, using the 18-56mm zoom set to wideangle (where we'd expect any distortion to be strongest). We're comparing the camera's JPEG with the corresponding DNG file processed through Adobe Camera Raw and RawTherapee.

Camera JPEG DNG + ACR DNG + RawTherapee

Roll your mouse over the buttons of the table above to see how the geometry compares between these three versions of the image. The result is pretty striking: processing through RawTherapee reveals plenty of barrel distortion, that's being automatically corrected in the JPEG (and indeed during normal live view).

We've looked at the full range of focal lengths, and have found that the 18-56mm's distortion characteristics are quite typical of a zoom of this range: ranging from barrel distortion at wideangle, through neutral in the middle of the range, to slight pincushion at 56mm. But the camera auto-corrects all the time, including when you're composing your pictures using the camera's LCD or EVF. So the user experience is of a lens that shows very little distortion at any time.

Chromatic aberration correction (18-56mm at 18mm)

Here we're looking at software correction of lateral chromatic aberration, by examining color fringing towards the corner of the frame with the zoom set to 18mm (again, where we'd expect it to be most noticeable). The camera JPEG gives a distinctly different result to the Raw conversions: both RAWTherapee and ACR show obvious (although not necessarily excessive) red/cyan fringing, while the JPEG gives some distinctly unexpected yellow and magenta fringing.

Camera JPEG 100% crop
DNG + ACR 100% crop
DNG + RawTherapee 100% crop

It's not quite clear exactly what's going on here - we have to assume that RawTherapee gives the most accurate depiction of the lens's characteristics, which then suggests that Leica does some processing to suppress the visibility of the fringing. We're just a bit surprised to see some residual yellow turn up in the camera's JPEGs. (Note that objects in the RawTherapee crop appear smaller because we haven't applied any distortion correction here.)

Vignetting (18-56mm at 18mm F3.5)

Finally, let's take a look at the possibility of Leica applying vignetting correction. In the example below we're comparing a camera JPEG to the corresponding DNG converted in ACR and RawTherapee, but now with distortion correction applied manually to the latter to attempt to match its geometry to the JPEG.

Camera JPEG DNG + ACR DNG + RawTherapee,
distortion corrected

Here the camera JPEG and the DNG converted with Adobe Camera Raw match quite closely. The DNG converted with RawTherapee has a very different rendition in terms of color saturation, which complicates the interpretation. However it still appears to show some degree of vignetting that's not present in the JPEGs.

It's worth noting that the 24mm prime would probably give much less ambiguous results when assessing the possibility of vignetting correction being applied, unfortunately we haven't yet been able to shoot with one to test this out.

Summary of findings

Our quick tests looking at the T's DNG files converted through DCRaw (via RawTherapee) show that Leica is in fact employing a thoroughly up-to-date lens design philosophy, similar to that used by pretty much every other mirrorless system. So the 18-56mm lens isn't fully corrected for distortion optically, but instead integrates software correction into the overall system design. This may not please purists, but frankly it counts as standard practice with modern lenses, and in our experience has little, if any, negative impact on the final image compared to fully-optical correction.

When it comes to lateral chromatic aberration, the lens is at least as well-corrected optically as we'd expect for a small zoom, and in fact probably better than most. But Leica is also taking the opportunity to reduce the visibility of color fringing via software correction. Again, this is simply standard practice in modern camera systems.

We're less certain about vignetting, but on the balance of probabilities would suggest that Leica is also reducing lens shading in software. We're less keen on this, and generally prefer it to be left to the user's discretion. But again, other mirrorless systems do the same thing (from, for example, Fujifilm and Sony).

Overall, we have no problem with Leica using this technology - incorporating software corrections into lens designs has enabled the creation of a range of very good lenses that wouldn't have been feasible if only glass were used. The only thing that we'd take issue with is the company claiming not to use this approach, when it so clearly is.