Lightroom 3 to include Lens Correction

[Remko]

Cool stuff, Tom Hogarty from Adobe just posted a preview of the Lens Correction feature as to be included in ACR 6 and Lightroom 3. They will publish a set of predefined lenses and you are also able to profile your own! (Community shared??)

The Lens Correction works on distortion(barrel and pincushion distortion), chromatic aberration and lens vignetting.

See: http://blogs.adobe.com/lightroomjournal ... n_sol.html

This together with the (said to be much) improved noise reduction for me is a big improvement and might obsolete a set of tools I use along side Lr (Neat Image and DxO). Hope it lives up to it.

My only wishes left for Lr are: HDR, Pano Stitching and Book publishing (as Aperture).

Can't wait to Lr3 to be released!

Regards Remko

 

[CHICHORNIO]

I´ve been using LR 3 Beta 2 version for a few months and it´s great, mostly for NR improvements and printing options. I still prefer Silkypix for developing my RAW files.
The workflow I´m using is developing the RAW in Silkypix 4.0 with no sharpening or NR at all (I just keep the settings as Natural in the Sharpening/NR section), then export the TIF file as a 16 bits with 0 unsharp masking, import the TIF into the LR 3 Beta 2, do all the necesary NR and sharpening there and print the file with also sharpening in standard mode depending of the type of paper (glossy or matte). The results are great for me. I hope you can try this workflow and it works for you. The only thing I will like LR 3 Beta 2 to have is Soft profing, so I can have the print with preview option as Photoshop CS3 does emulating the paper.

 

[Detail Man]

http://www.dxo.com/en/photo/dxo_optics_ ... s_softness

 

[Detail Man]

Remko,

It looks like the Adobe functionality that you have been anticipating has arrived:
http://labs.adobe.com/downloads/lenspro ... eator.html
and see:
http://labs.adobe.com/technologies/lensprofile_creator/

But nothing comes easy - and getting it right with the lens characterizations seems critical to success:


I wonder if an Adobe/other public database of profiles would be less or more reliable than using the profiles contained within the PT Lens database?
A little further reading does not clarify the question:

"Before uploading images check the following.
1. For zoom lenses did you take 10-12 images, each image at a different focal length?
2. Were you at least 25 feet (8 meters) from the target?
3. Is there a straight continuous feature near the top that runs end-to-end?
4. Did you remember not to tilt the camera?

Prepare Images
Resize the images so they're about 2000 pixels wide. This size is convenient for calibration and upload times are minimized. If you're using Photoshop sharpen the images and save as JPEG files (medium quality). Do not crop, rotate, correct perspective, or do any other edits except for resizing and sharpening. Image EXIF information should include camera make, model, and focal length."

http://www.epaperpress.com/ptlens/

 

[Detail Man]

I found this article that is relevant to the new Adobe optical correction "lens-characterization data" (from one published viewpoint). It is (sort-of) translated from French to English at this link:
http://translate.googleusercontent.com/ ... g#josc2411

Here is a link to the original article in French:
http://www.alpha-numerique.fr/index.php ... Itemid=359

And here (in English) is DxO "tooting it's own horn" in the partial exercise of the politics of corporate competition.:
http://www.dpreview.com/news/1005/10052 ... rofile.asp

But, (in the best-case scenario, where manufacturers would/will create/add "lens-correction modules" for use with Adobe applications), I still see no credible mention of anything like DxO's "Lens-Blur" Optical Corrections being possible with the Adobe system (though many of the other most salient and desirable optical correction mechanisms are included).

 

[Remko]

Adobe is sure shaking things up with the new ACR/Lr3 capabilities and raising the bar for the other tools.

Will it reach DxO capabilties, I'm quite sure it won't with this release. But competition is good and since I'm primary a Lightroom (Lr) user this is good news at least for me. Yes I use a number of tools besides Lr (DxO, PhotoMatix, AutoStitch, NeatImage, PTLens, Photoshop) and I love to ditch some in favour of doing things within Lr.

DxO, PTLens & NeatImage might be the ones now. DxO probably not, NeatImage maybe, PTLens probably. Both PTLens and NeatImage also have community profiling and I happy with the results.

Anyway I can't wait to get my hands on the official Lr3 release and start testing noise reduction and lens correction for myself. And maybe, just maybe DxO will speed up and widen their camera and lens support.

Remko

 

[Detail Man]

Remko,

The (updated mid-May 2010) document entitled "Working with Lightroom and DxOOpticsPro6" for DxO Optics Pro 6.2 states compatibility with LR 2.x. I wonder if it will work with LR 3.x?
http://www.dxo.com/var/dxo/storage/fcke ... csPro6.pdf
A shorter summary of the document exists at:
http://dxo.com/uk/photo/dxo_optics_pro/ ... lightroom2

(If this is something that you do, or have tried), if/when using DxO together with LR, have/do you:

(1) Process with DxO first (for optical corrections), then export to LR? (If so, do you export in a TIFF, or the "linear DNG", format?); or

(2) Do you send "raw" image-files from LR to DxO, then exporting from DxO to LR in TIFF format? If so, (in this case) does DxO allow the full gamut of "raw"-mode processing features (such as the "Lens Softness" correction) to be applied to the "raw" image-file that is sent to DxO from LR?

For the technique number (2) above, there are indications that (at least, with earlier versions of DxO and Lightroom) the relevant EXIF data was not in all cases properly passed from LR to DXO.
http://forums.adobe.com/message/1394811
and here is a more recent indication (involving LR 3 Beta, and DxO 6.x) that problems may still exist (at least in some cases):
http://forums.adobe.com/message/2492854

 

[Detail Man]

DxO User-Tools to Create Custom Correction Data

"NEWLY AVAILABLE: Since the release of DxO Optics Pro 6, DxO Labs is now offering a new set of correction tools to let our customers create, save as a preset, and share their own optical corrections if their camera and lens combinations are not yet supported and a DxO Optics Module is not yet available. This is a powerful new feature we are happy to offer our customers."
http://help.dxo.com/faq/index.php?actio ... artlang=en

I think that they are stating that DxO Optics Pro Versions 6.x can create/export/import a user-configured preset with specific optical correction settings for the following adjustments:
Vignetting; and
Distortions (Barrel, Pincushion, Fisheye); and
Chromatic Aberration (with Purple Fringing correction).

However, the applicability of all of the above correction-settings (in any software program's applications) depend on specific data relating to the configuration/position of the camera at the time of shooting (focal length, aperture, focus-distance, etc.). This is absolutely true for DxO's unique and powerful "Lens Softness" corrections.

Without the operational functionality of any particular program being able to automatically locate, correctly identify, and utilize specific EXIF data (including "maker-notes" relating to the specific shooting parameters existing in the recorded image-file), the possibilities appear to be limited ... or at least complicated and quite tedious to compile (as well as to apply) in making corrections.

Assuming that users are happy to share and apply DxO 6.x Presets that are created by other individuals (and not camera manufacturers, or by DxO Labs), it appears that the capability exists for DxO users to be able to accomplish something similar to what Adobe is presently offering users for it's own product-line.

The limitations of the above described (DxO) user-controlled functionality appear to (also) underscore the limitations that come along with Adobe's "Lens Profile" functionalities ... ?

 

[Remko]

Hi DM,

What I typically do is use DxO as pre-processor. So feed the original RAW files into DxO and let DxO create DNG's (very big-ones I have to admit 50MB each). Then import these DNG's into Lr. And stack them with the original RAW's (Lr is my image management tool so every image goes into the Lr catalog).

Lr3 won't be any different than Lr2 in capabilities when working with DxO. Since the export and import capabilities are alike.

When processing some Canon 500D mid and high-ISO (400-1600) images with DxO this week I was not that content with the DxO noise reduction. Images ended up too blotchy and quite some details were lost. I haven't noticed this before with my 350D images. Maybe the resolution of the 500D (15Mp) and viewing these images 1:1 is not a good idea. In the end I decided to batch the images into NeatImage. I think I'll have to investigate further. The images were shot this weekend during a church service I'd shot (quite dark so higher ISO was necessary).

Remko

 

[Detail Man]

Remko,

Have you seen the interesting article that speculates as to the nature of the DxO NR architecture?

English version:
http://translate.googleusercontent.com/ ... Fik07Bqemg

French version:
http://www.alpha-numerique.fr/index.php ... Itemid=317

 

[Detail Man]

Remko,

I have been reading some of the Adobe information about the LR3 Beta 2 Noise Reduction:
"Lightroom 3 beta 2 introduces a much more complete solution that includes an outstanding luminance noise reduction control and we’re excited to hear your thoughts on the improvements."
- LR3 Beta 2 Release Notes

I was wondering if you have tried it out? And (since you are also a DxO Optics Pro user), wondering how you may find it comparing to DxO's Noise Reduction (with it's separate Luminance and Chrominance NR controls)?

 

[odklizec]

Jeremy, you should check this post...
viewtopic.php?p=21877#p21877

Download the full resolution files I processed with LR3 beta and compare them with Carl's originals. What I did in these photos was turning the Chroma noise reduction all way up and in ISO400-3200 sample I did a light touch of Luminance NR (about 10-15%). I previously noticed that LR3 Beta 1 did great job with Chroma noise reduction. But now it's really superb and the Luminance NR works great as well. I'm sorry to say this, but other RAW editors I tried so far are nowhere as good in NR, including dedicated noise reduction tool Noise Ninja! I'm really impressed with LR3 NR results.

 

[Detail Man]

... now it's really superb and the Luminance NR works great as well. I'm sorry to say this, but other RAW editors I tried so far are nowhere as good in NR, including dedicated noise reduction tool Noise Ninja! I'm really impressed with LR3 NR results.

Pavel,

That does look pretty good, indeed. Interesting.

BTW - Have you (in your testing) had occasion to be able to evaluate DxO Optics Pro's Noise Reduction functionality and effectiveness? From using it myself, it appears (to me) that the Luminance NR control is more effective than the Chrominance NR control. I initially worked with a philosophy of increasing Chrominance NR more than Luminance NR, but I seem to find that the Luminance NR is somewhat more effective in achieving a desired NR effect. DxO typically (automatically) selects a (slightly) higher level of Luminance NR relative to Chrominance NR. With DMC-LX3 RW2 image-files, I typically divide DxO's "automatic" settings by a factor of at least 2 (and, more commonly a factor of 4, 5, or 6) for ISO Sensitivities in the range of 160 - 400. After reading the article (previously linked in this thread) speculating about DxO's NR-architecture, it is (alleged) that the absolute level of the Luminance NR control makes a large difference in the proportional weightings of the (speculated to be) pre-demosaicing NR, and the kind of "detail restoration" information derived (in some manner) from image-sensor information prior to the (speculated to be) pre-demosaicing NR. Thus, it should be noted that the (Luminance and Chrominance) NR control-settings that I have found necessary to apply (in processing DMC-LX3 RW2s up to ISO=400) has seldom been more than a setting of around 10 or so (out of a total control-range of 100 for each slider-control). From what the published article speculates, this would (largely) correspond to applying the results of the (alleged pre-demosaicing) noise reduction applied. At any rate - I (personally) find that DxO (in general, and without rigorous testing) appears to do a better job (relative to Silkypix) of preserving a sense of image-details relative to the amount of image-noise reduction (a general perceptual impression from my memory, rather than a rigorously tested and demonstrated assertion).

Perhaps Remko (who actively uses both DxO and LR3 Beta 2) may do some comparisons on (DxO supported) "raw" image-files - or, at least, share his general impressions with us as to how DxO NR compares with LR3 Beta 2 NR. It's always exciting/sobering to find out whether one's "preferred tool of choice" has been "outclassed" by competing efforts from other manufacturers ...

That GXR P10 module looks pretty "slick" ... ... I've been scratching my head a bit as to why they would choose a 1/2.3" image-sensor (as opposed to and approximately 1/1.7" image-sensor) - though it does help in the Depth of Field "department" ... There seem to be hints that the NR is applied prior to demosaicing:

"... noise reduction processing is immediately performed on the sensor's signal output."
http://www.ricoh.com/r_dc/gxr/unit3.html
and
"Performing noise reduction processing on the signal immediately after it is output by the back-illuminated CMOS sensor makes it possible to reduce noise while maintaining resolution, tone characteristics, and saturation."
http://www.dpreview.com/news/1005/10050702ricohp10.asp

I wonder how the signal/noise ratio of their back-lit 1/2.3" MOS image sensor "stacks up" compared to the state-of-the-art for 1/2.3" CCD image-sensors? I have read conflicting information about the relative signal/noise ratios of these two technologies. And (of course) each (new) image-sensor to emerge may (or may not) achieve signal/noise performance that may exceed what has come before - thus superceding what has been previously published about the relative merits of each of these two (CCD and CMOS) image-sensor technologies (when the same pixel-pitch as well as the same active sensor-area is utilized in each separate case).

It's nice to see Ricoh adding image-stabilization functionality to the P10 Module. Though (as reported by Ricoh owners, including yourself, on this forum), Ricoh's "sensor-shift" image-stabilization may be ineffective (or even problematic enough as to worsen, rather than improve, results ... ), perhaps improvements have been made in their implementation of this technology in the P10 module? Image-stabilization (when effectively implemented) is a very beneficial feature, indeed. Rather than having to increase ISO Sensitivity (invariably leading to the application of increased amounts of noise reduction as a result), effective image-stabilization is like having a "faster" lens-system - without the loss of Depth of Field that a larger aperture causes. A gain of 2-3 "stops" is nothing to "sneeze at". I like it so much that I will not purchase a camera that do not include effective image-stabilization functionality ...

 

[odklizec]

Hi Jeremy,

I'm sorry, but since there is no support for Ricoh cameras in DxO, I'm not interested in it. I'm just not willing to add yet another step to my existing workflow. Converting the DNG files to huge TIFF with messed EXIF is really not an option for me

LR3 works surprisingly well in NR department. From what I saw, it works better than Noise Ninja, which I always considered to be a gold standard in noise processing. LR3 simply produces more appealing grain and color noise-free images and this all with much less effort.

As for why Ricoh used 1/2.3" sensor in P10, it's simple. It allows to create very small 10x zoom lens. P10 with 1/1.7" sensor and 10x zoom lens would be ridiculously big. Not to mention there is currently not even similarly fast 1/1.7" sensor than the 1/2.3" one used in P10. Additionally, it's much easier to repack the existing CX technology into GXR module, than to design whole new module and lens. With P10, Ricoh finally converted last of their existing cameras into GXR module (except GRD) and following GXR camera modules will be designed as whole new modules with no history.

I was very skeptic about P10 module. But with current P10 kit price and so great results delivered by LR3, I think it's very appealing entry into GXR system? I guess Ricoh will get many new users thanks to P10 kit.

 

[Remko]

Hi DM,

i have to admit that i haven't used Lr3 beta. Although the improvements seem to be very worthwhile i don't have the time to spend on beta stuff.

The good news is Lr3 just got released i just read. Since i have some busy weeks i won't come around testing it on short term. If you can't wait you should be able to download the 30 day trial by now from the Adobe site.

Remko

 

[Remko]

DPReview just posted their Lr3 review: http://www.dpreview.com/reviews/adobelightroom3/ interresting stuff.

They've also included a range of RAW processing comparisons with SilxyPix, Bibble, DPP (Canon only) and NikonCapture (Nikon only) and for a number of cams (including G11 and LX3, so not only DSLR).

As for lens corrections, DxO is still the tool to beat and Lr3 seems to have some challenges.

I have to admit I've not upgraded myself yet.

Remko

 

[Detail Man]

DP Review:
"Adobe's new 2010 process delivers excellent detail and relatively low noise at the expense of a noticeable 'grittiness' which we don't find objectionable, but can of course be reduced using the 'luminance' noise slider in Lightroom 3 (which by default is set to zero).
http://www.dpreview.com/reviews/adobeli ... /page7.asp

I took a look at the "grittiness" visible in the DMC-LX3 samples on the web-page above. DP Review states that "defaults" were (for the most part) used when employing the "raw" processors. The Lightroom 3 default Sharpening control settings are: Strength=25; Radius=1.0; Detail=25; Masking=0.0.

Processed this DMC-LX3 full-telephoto RW2 (ISO=100, F=2.8, Ts=1/400) crop of a building at least 100 Meters from the camera (and a church-steeple probably around 200 Meters from the camera, at the far-right edge of the cropped image-frame) using DxO Optics Pro (using available adjustments and all of the automatic optical corrections, without using DxO "Lens Softness" corrections:


Processed same crop using Lightroom 3 with all default settings [except for reducing the (output) Brightness control from it's default value of 50 to a value of 25]. No optical lens corrections were used. Note that the level of visible image-detail (in the brick structure of the building, as well as the structure of the church-steeple) appears to be improved (relative to previous image):


Take a look at what is possible when the DxO Optics Pro "Lens Softness" corrections (which Lightroom/ACR does not do) are enabled (with no other sharpening tools used in the processing):


My experience using Lightroom indicates that (even with a Sharpening Radius as low as 0.5) increasing the value of the Sharpening Strength and Detail controls in an attempt to provide more visible image-detail rapidly leads to an unsightly "grittiness" (which I have described as a kind of "quantized graininess"). Further, these processing artifacts exist even at lower settings of the Sharpening controls, and are readily amplified by any subsequent USM operations that may be performed on the image (even an image down-sized by re-sampling prior to applying USM).

These findings concerning the relative abilities of DxO Optics Pro and Lightroom/ACR to resolve image "micro-details" without producing unsightly processing artifacts are consistent with those experienced when processing "raw" image-files recorded by the (also DxO-supported):

Panasonic DMC-GH1 at ISO=100:
viewtopic.php?f=24&t=4780&p=22781#p22390

Canon G11 at ISO=800:
viewtopic.php?f=24&t=4780&p=22781#p22758

It appears to me from my experience that the Lightroom/ACR Sharpening tools are not capable of preserving critical image "micro-details" to an extent that compares with the Lens Softness corrections of DxO Optics Pro, and that Lightroom/ACR produce significant levels of unsightly processing artifacts (the "grittiness" or "quantized graininess") when the user attempts to match the "acuity/clarity/sharpness" provided by DxO Optics Pro Lens Softness corrections (which, like all of the DxO Optics Pro automatic optical corrections, are automatically applied based on the camera's actual focal length and aperture used when the camera recorded the "raw" image being processed). No fuss - no muss!

To "file your plea" with DxO that they add support for your own favorite "raw" recording camera(s)/len(s) combinations, it only takes a minute to do so at:
http://www.dxo.com/intl/photo/support/m ... ailability
I have made DMC-LX5 support requests from all of my email addresses ...

 

[odklizec]

Jeremy, I always admire your enthusiasm when it comes to DXO. So please, don't take my following words wrong. As few times before, I have to disagree with your findings. In my opinion, the DXO result you showed us today appears to be seriously oversharpened with visible sharpening artifacts on the edges and diagonals? In my humble opinion, the so called "Lens Softness" corrections is nothing else than a clever USM or something similar. The LR3 is maybe softer and unsaturated, but generally, it appears much better and gives much better start for further enhancements.

The reason of unsaturated look of LR3 file is that LX3 RAW files are generally unsaturated. Try convert any LX3 file with dcraw or rawtherapee and you will see nearly true appearance of the LX3 RAW (including the huge barrel distortion). The reason why it looks more colorful (LX3 JPEG-like) in other RAW editors (for example in Silkypix and as it appears in DXO too) is, that these RAW editors apply Panasonic recommended/requested RAW enhancements. So you in fact don't have a chance to see how the LX3 RAW really appears.

As for the details provided by DxO "Lens Softness" correction, I personally see nothing appealing in this exact example. I'm sure it may improve the sharpness of many RAW files, but in my opinion, it's not this case. The image look seriously oversharpened to me? Not to mention the ugly color noise. But I guess it's because you did not apply the color noise reduction? So this may be further improved. In any case, for me personally, the LR3 file looks better.

Here I did some some 1:1 crops from your LR3 and DxO exports and even a crop from LR3, where I applied a touch of luminance sharpening. But because the source file was JPEG, it can hardly be called an ideal source material for sharpening experiments? I guess it can be noticeable improved in RAW?

 

[Detail Man]

Odklizec,

Leaving personal tastes as subjective and rightfully personal as they may be, there are a few factual and procedural things for me to point out and clarify:

Jeremy, I always admire your enthusiasm when it comes to DXO. So please, don't take my following words wrong. As few times before, I have to disagree with your findings. In my opinion, the DXO result you showed us today appears to be seriously oversharpened with visible sharpening artifacts on the edges and diagonals? In my humble opinion, the so called "Lens Softness" corrections is nothing else than a clever USM or something similar.

Dxo Optics Pro Lens Softness corrections are known (stated in writing and confirmed by DxO to me) to include "deconvolution deblurring" within the "Lens Softness" corrections process:
http://www.dxo.com/en/photo/dxo_optics_ ... s_softness

DxO themselves also emailed me the links to this long-published information relating to their earlier Version 3.5 of DxO Optics Pro (as well as relating to their newer Versions 6.x):

"Lens Softness correction
Lens softness’ is the intrinsic degradation of sharpness introduced by an imaging device (camera body plus lens). In image processing terms, this is a local, color-channel dependent, anisotropic convolution of the original image, which results in a 'blurry' image. In terms of image spatial frequencies, ‘blurriness' refers to how well low spatial frequencies are reproduced in the image. You may be familiar with these concepts if you are familiar with MTF (Modulation Transfer Function) curves. DxO Labs has developed a unique unit called the BxU (Blur eXperience Unit) which is a mathematical way of describing this 'blur'. Reducing the ‘lens softness’ or 'blur' or 'lack of sharpness' means performing local, color-channel dependent and anisotropic deconvolution of the image produced by the camera. Furthermore, DxO deblurring uses a complex contextual approach taking into account both local noise and local detail level in the image. As a result, deblurring will be automatically reduced in uniform areas (like a pure blue sky.), but increased in a detailed zone ...

... DxO Lens Softness is different from Unsharp Masking. DxO Lens softness correction is automatically adapted to the lens intrinsic softness of each shot, which is field, color and captures parameters dependent (for example, image corners will be sharpened differently than image center). It brings each image to a nominal and constant level of softness.

http://www.beautiful-landscape.com/Thoughts35.html


"Deblur-Lens softness
One of the main reasons I use DxO, and one of the main reasons behind the quality of my images is the "deblur/lens softness" feature of DxO. This feature really separates DxO from the crowd IMO and plays to its unique camera/lens combo analysis. This deblur is far more than an image sharpener as is evidenced by the incredible scalability of the deblur feature ...

... What is the difference between a debur/Lens softness correction and the sharpening of an image? (My definition):
Deburring is similar to removing a thin film of oil off your lens - realizing the accutance potential of what your unique combination of camera/lens captures to the sensor, correcting for inherent flaws in your optics (just as was done for the Hubble telescope after its flawed mirror was sent into orbit) ...

... While certain localized areas of an image can be USM'ed to match DxO's debur/lens softness feature, I have found this same USM setting applied in a global fashion will induce halos in other areas."

http://www.beautiful-landscape.com/Thou ... titch.html

The LR3 is maybe softer and unsaturated, but generally, it appears much better and gives much better start for further enhancements.

It is important to understand that when I stated in my previous post:
"using available adjustments and all of the automatic optical corrections"
the phrase "available adjustments" refers to color saturation and lighting adjustments made on both of the posted DxO processed images at a previous time many weeks ago. My apologies if that was not made sufficiently clear in my post. Thus, the comparison was intended to relate to the issue of image-detail (and not provide a reference point for comparison of other characteristics of the two individual processes) ...

(In this particular case), when I processed the image in Lightroom 3, I wanted to apply the same "default" Sharpening settings that DP Review implies that they also used. I considered modifying other Lightroom 3 parameters (in order to try to more closely match color and lighting effects), but I decided to leave the settings in the same configuration that DP Review used at:
http://www.dpreview.com/reviews/adobeli ... /page7.asp

As well, the examples already posted (with very complete data describing all processor settings) for the:

Panasonic DMC-GH1 at: viewtopic.php?f=24&t=4780&p=22781#p22390

and the

Canon G11 at: viewtopic.php?f=24&t=4780&p=22781#p22758

previously endeavored to more closely match all of the parameters (by adjusting the various "raw" processors used in those tests). (You may not personally like the overall appearance of DxO output), but (I think that) both of the above links test results reveal clearly significant differences in the amount of visible image-detail between DxO Optics Pro and Lightroom (in all 3 separate comparisons).

The reason of unsaturated look of LR3 file is that LX3 RAW files are generally unsaturated.

See my text directly above.

Try convert any LX3 file with dcraw or rawtherapee and you will see nearly true appearance of the LX3 RAW (including the huge barrel distortion). The reason why it looks more colorful (LX3 JPEG-like) in other RAW editors (for example in Silkypix and as it appears in DXO too) is, that these RAW editors apply Panasonic recommended/requested RAW enhancements.

We have both been parties to speculation that ISL may do such things in Silkypix (over and above an attempt at barrel-distortion correction). I do not know of any evidence that DxO Optics Pro performs any "silent" operations (other than offer the option of "As Shot" WB color temperature, WB Tint, and automatically correcting barrel/pincushion distortion, vignetting, chromatic aberrations, and lens softness). All of the above functions can easily be disabled completely by the DxO Optics Pro user.

So you in fact don't have a chance to see how the LX3 RAW really appears.

True as well in the cases of Lightroom and ACR. We can always use Rawnalyze to see the ugly truth ... but we suspect that (some) cameras and (all) processors take liberties with "raw" content, so I am not sure as to the intention of the point that you are trying to make ...

As for the details provided by DxO "Lens Softness" correction, I personally see nothing appealing in this exact example. I'm sure it may improve the sharpness of many RAW files, but in my opinion, it's not this case. The image look seriously oversharpened to me? Not to mention the ugly color noise. But I guess it's because you did not apply the color noise reduction? So this may be further improved. In any case, for me personally, the LR3 file looks better.

OK.

Here I did some some 1:1 crops from your LR3 and DxO exports and even a crop from LR3, where I applied a touch of luminance sharpening. But because the source file was JPEG, it can hardly be called an ideal source material for sharpening experiments?

That would surely seem to be the case.
On the subject of attempting to perform processor comparisons under conditions that are performed under as close to equivalent conditions as possible, did you read and think about the many valid points about equivalency of test conditions that I made in my post at:
viewtopic.php?f=24&t=4780&start=10#p22319
and my expressed interest in the performance of more representative comparisons at:
viewtopic.php?f=24&t=4780&start=20#p22781
which concerned your Dxo/Lightroom Noise Reduction comparison post at:
viewtopic.php?f=24&t=4780&start=10#p22276

 

[deejjjaaaa]

this might be interesting for the parties in this topic :

http://deejjjaaaa.blogspot.com/2010/07/ ... scape.html