Dxo 9.5 vs. LR 5.5 - horrible result

[kirschm]

Perhaps I am stupid regarding DxO... or DxO is really horrible compared to Lightroom.

As an example I took a RAW ISO3200 image (Samsung EX1, Olympus MFT same result). I processed it in LR5.5 as well as in DxO9.5 (regarding noise and moderate sharpening). In DxO I used 'Prime 40%' and already reduced the quite agressive sharpening sliders (Lens Softness).

Here a simple 200%-Monitor-View-Screenshot (LR on top, DxO below, both exported to JPG)... Not only that LR appears much better... what are those horrible vertical and horizontal 'pixel-stripes' on the DxO image?

Click: https://www.mediafire.com/?8pc3921rr0xl2r9

here you'll find all relevant files in order to reproduce (or hopefully get better DxO results) the results:
https://www.mediafire.com/?aji6b91399ar75i
Dxo9.5 vs LR5.5.jpg -> quick&dirty screenshot
SAM_0395 EX1-I3200-F3,0-1_750s-5,20mm.SRW -> original RAW-file
SAM_0395 EX1-I3200-F3,0-1_750s-5,20mm ACR84.jpg -> LR export
SAM_0395 EX1-I3200-F3,0-1_750s-5,20mmDxO40.jpg -> DxO export

 

[randyspan]

Yes, I had similar results. Lightroom is the ONE for me! Won't make it to the end of my Dxo free one month trial: out it goes!

 

[Matero]

Here a simple 200%-Monitor-View-Screenshot (LR on top, DxO below, both exported to JPG)... Not only that LR appears much better... what are those horrible vertical and horizontal 'pixel-stripes' on the DxO image?

Click: https://www.mediafire.com/?8pc3921rr0xl2r9

Sorry if this is a stupid question, but what you mean by this?

If you took screenshot from your photo before 'developing' it and exporting it to real eg. JPEG then you're comparing previews of your picture. I've found that actual developed JPEG is remarkably better than screen preview of it inside DxO. Just my 2 cents...

 

[kirschm]

Sorry if this is a stupid question, but what you mean by this?

If you took screenshot from your photo before 'developing' it and exporting it to real eg. JPEG then you're comparing previews of your picture. I've found that actual developed JPEG is remarkably better than screen preview of it inside DxO. Just my 2 cents...

I am talking about Screenshots of JPG-files, exported from DxO in highest 100% quality.

I am not talking about screenshots taken from within DxO. Posting 'real' exported JPGs vs. screenshots of those exported JPGs here wouldn't make a difference because both look equally horrible.

 

[Matero]

Well, I couldn't reproduce your result with my samples. Again stupid question, your image hints that DxO one has worse resolution than LR. Are you sure you haven't left 'enable resizing' option on when exporting JPEG from DxO? (Done that many times, nowadays I remember to check)

Otherwise I can't reproduce such a dramatic difference.

 

[kirschm]

unfortunately not... both JPGs (LR vs DxO) have exactly the same pixel size...

 

[kirschm]

I have now attached all original files in my first post... in case you want to reproduce or not reproduce my results....

 

[Matero]

I'll give it a try later today. I'm curious to see because I'm just changing my workflow from Aperture to Lightroom or DxO. And this made me thinking...

 

[OzRay]

I also tried the trial version of DxO and thought the results were pretty woeful compared to LR. If I had to change RAW processors, I think I'd go for Capture 1 again (Lightzone is still a slow work in progress).

 

[Matero]

Yes, I can now see what you mean. With Prime NR in DxO you get very mosaic results in pixel peeping level. Very poor, IMHO. And no matter how you tweak settings, mosaic stays there. I wonder where one could really use this Prime noise reduction?

However with the High NR in DxO you can get pretty much comparable results with LR. They are just different. DxO maybe little bit smoother or fuzzier and LR some times sharper but has artifacts. So maybe you can't get best of the both worlds, then

Below are my samples, I also made comparison from the center to avoid lens correction to add to results. Left upper corner is kirschm's original LR export, right lower corner is mine export from DxO without any other correction than High NR with my own tweaks. Right upper corner is auto correction with High NR from DxO and lower left corner auto correction with Prime NR from DxO. Please notice that both autocorrections from DxO includes lens distortion corrections, that's why the pictures are different.

Just to satisfy my curiosity I made one with 50% zoom level to avoid pixel peeping. Not much difference then, what you think?

[clicking the image you go to gallery and can view little bit larger version]

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

 

[kirschm]

Many thanks, Matero... for reproducing my findings. DxO-Prime was my only pro-dxo argument. It's a pity.... Need to stay with LR... Was looking for an alternative in case they introduce cloud for LR.

 

[Dave Lively]

I downloaded DXO Optics yesterday. My one major complaint about Lightroom is the lack of lens profiles for m43. While the built in correction data for distortion works OK there are times when I need to manually adjust it when a picture has a lot of straight lines. LR is supposed to automatically correct CA when using a Panasonic lens on a Panasonic body but while that may be true the correction is less effective than an OOC JPEG. And vignetting is always a manual adjustment.

Since DXO made it easier to work with Optics as a supplement to LR I am hoping that I can use LR for most of my photos and Optics only when I need to correct for some lens defect. I do not plan to pass the results of Optics back to LR. My workflow would be a fork, not a loop. Any picture processed in Optics would go straight to to JPEG.

Since it uses lens profiles I was hoping DXO would automatically adjust for distortion, CA and vignetting. And it does. I have not tuned Optics to my tastes so it is way too early to reach any final conclusions in general I like LR output better so far. Optics has some artifacts at 100% I do not like but I am not sure if they could be eliminated by tuning Optics. What it really comes down to is whether tuning Optics settings to match my tastes is less work than manually adjusting for lens defects in LR. Since I do not adjust for vignetting, CA or fine tune distortion in most of my photos I doubt I will be purchasing Optics when my trial expires. The expense of owning 2 different raw processors and added effort of tailoring both of them for the results I want is a pretty big hurdle.

I had someone on DPR ask about Lightroom and CA correction so I took a test shot. The results can be seen at http://www.dpreview.com/forums/post/53928195 . While DXO did the best job of correcting the CA there are some black specs in areas where the CA used to be. Manually correcting the CA in LR took about 30 seconds and gave better results.