LX7 awful edge resolution due to distortion correction

Started Apr 30, 2013 | Discussions thread
Shirozina
Regular MemberPosts: 306
Like?
Re: LX7 awful edge resolution due to distortion correction
In reply to Detail Man, May 2, 2013

Detail Man wrote:

Shirozina wrote:

Detail Man wrote:

Shirozina wrote:

Just 'upgraded' from an LX5 to an LX7 and noticed very poor edge resolution at wide settings using RAW developed in ACR - worse than the LX5. When processing them in Capture one where you can turn off the distortion correction you can see why - the native lens distortion is absolutely horrendous and at the wide end it looks like a fisheye lens. All that loss of resolution is solely from the distortion correction stretching the edges of the image back into shape and looking at the undistorted image the resolution is superb right into the corners.

The last image had some noise reduction applied to the processed image so here it is without and also the thumbnail showing what the distortion correction is actually doing.

The rectilinear distrortion corrected one (here the lower image) does look pretty bad - though you have here displayed it at a higher magnification (which is bound to make the upper one look better). I take it that it looks as bad using LR as it does using C1 ? The Imaging Resource reviews state the following uncorrected RAW full wide-angle (24mm) distortion percentages:

LX7 (4.0%):

.

LX5 (2.7%):

.

LX3 (2.9%):

The inter-camera differences in barrel distortion do not appear to be very significant. Perhaps there is (somehow) something else going on ? The following are just ideas, just thoughts ...

Where you using the physically widest utilization of image-sensor photosites (16:9 aspect-ratio) ? Is the effect equal in the right-hand sections of your images ?

A fair number of LX5's had an asymmetrical softness effect that would often (but not always) manifest around the left-hand edges/corners.

You (might) have a defective lens ? The image-file meta-data driving your RAW processor corrections generates statistical correction data (which is not based on individual lens tests).

Yes, I know that you find the the uncorrected versions looking better - but without a closer look at equal (and higher) magnifications it is hard to discern a lot from your example images.

Both images are at 100% magnification - the top one looks bigger due to the image being stretched and thus resampled / upressed during the distortion correction process which is what the issue I have is. It needs so much that it destroys image detail. Look at the thumbnail lower left to see the huge pincushion shape needed to create the lower corrected image. They were taken at 16:9 and the effect is equal both sides. Image stabiliser was off which I know on the LX5 can give one edge a softer look than the other.

Was not aware that using the POIS system can cause lens-system de-centering effects to occur.

When the aspect-ratio is 4:3, the effects that you note using the LX7 are not nearly as much (with the LX3, having 2.9% barrel distortion at wide-angle). Using DxO Optics Pro 7.23 below:

No Distortion Correction

Distortion Correction - Aspect Ratio = 1.333

Distortion Correction - Extra Wide Aspect Ratio = 1.393

.

However, at a recorded aspect-ratio of 16:9, the LX5 shows a fair amount of the same effects. Bottom-right corner of an LX5 RW2 processed using DxO Optics Pro 7.23 (no sharpening, no NR):

DMC-LX5_DxO-7.23_No Corrections

DMC-LX5_DxO-7.23_Distortion Correction

.

You probably are not in the mood to use a different RAW processor as a work-around - but using the "Lens Softness" corrections of DxO Optics Pro along with the distortion corrections (and no other sharpening processes) yields better results than in either of the two cases displayed above.

DMC-LX5_DxO-7.23_Distortion and Lens Softness Corrections

.

Even when one uses DxO's ablity to further extend the aspect-ratio of the processed image (when barrel distortion is present) - from 1.778 to 1.910 in this particular case, the "Lens Softness" corrections still manage to do (what I think is) a pretty good job of depicting the extra image-data:

DMC-LX5_DxO-7.23_AR=1.910_Distortion and Lens Softness Corrections

.

The extra 1% (or so) of barrel distortion measured with the LX7 (wide-angle at 4:3 aspect-ratio) may translate to more than a 33% difference in the case of a 16:9 aspect-ratio. DxO Optics Pro "Lens Softness" corrections (might) serve to lessen the extent of the problem with the LX7 somewhat (similar to what is demonstrated above when processing LX5 RW2s). Perhaps something to try ?

Anyway, I own and use a LX3, and have no particular interest in promoting any particular camera model. The LX7 has nearly one EV more Dynamic Range than the LX3/LX5, but appears to use a somewhat stronger AA filter assembly, resulting in something less in the way of fine-details.

DM

Are DXO's distortion figures quoted additional to the in built distortion correction or from the native image that is projected onto the sensor. Even after correction there is still some distortion on the LX7 which if you were a perfectionist you would probably want to correct further.

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow