Mini bake-off: DPP vs. DxO

[neuroanatomist]

Some time back, I posted a poll question on RAW processing software - one reason I posed the question was that I was curious about DxO Optics Pro but surprisingly (to me, anyway), no one indicated that they used DxO.

I was specifically interested in DxO because of the lens corrections it offers, which I think is a real benefit of DPP (in addition to the cost benefit!). LR3 now offers lens corrections as well, but I don't think LR3's corrections are specific to the camera used, whereas DxO's are specific. Another reportedly good DxO feature is better noise reduction.

So, I downloaded the trial version and gave it a test. It's pretty intuitive to use, IMO, and I was up and running quickly without tutorials or the manual.


I'll give a brief pro and con summary, then some general impressions and some specific examples. The comparisons are all done using default options, with no adjustments to the histogram (except a significant EC adjustment in one case, mentioned below).

Pros of DxO compared to DPP:

• Faster/easier image editing workflow
• Customizable interface
• Better image corrections (noise, distortion, sharpness,
  highlight recovery, etc.)
• Can rescue images DPP cannot

Cons of DxO compared to DPP:

• Not free
• Less flexible rating system
• No clone/stamp tool
• Default color is 'cooler' and less saturated (a con for me,
  but maybe not for everyone)
• Slower RAW conversions

The workflow for DxO is reasonable, although I don't like that fact that you need to add images to a project before you can apply a rating (edit - no longer necessary). DPP allows what I found to be a more efficient triage of incoming images. But I wasn't used to DPP at first either, so this is something that should work itself out quickly.

DxO applies corrections automatically based on a preset, and makes it easy to see the effects of those corrections. They can be altered at will, and the interface is very customizable, allowing you to design a workspace with only those corrections you'll use frequently.

I started by comparing a 'good'; image - a daytime shot with good lighting, well-exposed straight from the camera, reasonably low ISO, etc.

<not shown because of the limit of 5 images per post>

In this case, there's not much difference between the two programs, although the color may be a little better with DxO.

Next, I looked at white balance. The WB results with DxO are different than DPP, particularly where flash is used. I normally shoot with Auto WB and select the best option in post (sometimes best = most pleasing, and I frequently use Cloudy for a little boost in warmth). With a Canon flash and a recent body, the flash passes color temperature data to the body which is used in setting white balance. When Flash WB is chosen in DPP, those color temperature data are used, whereas DxO's Flash WB
setting is unaware of the additional data and uses a constant 6000K. Here's the result for an Auto WB shot set to Flash WB in DPP vs. Flash WB in DxO, where DxO is cooler:

Leaving DxO on As Shot produces results identical to DPP, and DxO does make it very easy to tweak the WB, since a temperature slider is
available alongside all of the preset options (it's also nice that DxO showsyou the color temperature of each of those presets). So, it's something to be aware of but something that's easy to correct with a slight modification to the normal workflow.

Color also appears a little cooler and saturation a little lower with DxO. I think this results from Canon's Standard Picture Style being more warm and saturated, as
many users find that more pleasing (I am one of them). Here' an example:

Like the white balance issue, this is something that's easy to correct in DxO.

Next, I looked at sharpness and distortion correction. I chose an architectural image shot with the 24-105mm at 24mm, where there is substantial barrel distortion on FF (photozone.de calls it 'massive'). The top row are the full image, and the bottom row are near-100% crops of the planter at the lower right of the steps. Going from left to right are the DPP image without distortion correction, the DPP image with distortion correction, the DxO image, and finally an optically-corrected image (shot with the TS-E 24mm f/3.5L II with a +8 shift).

DPP does a decent job of correcting the distortion, but not quite as good as DxO, which gets closer to the 'ideal' image from the TS-E lens. Looking at the crop of the
planter, you can see that there is still some distortion in both software-corrected images. The DPP correction results in a softer image (some detail on the planter and flowers is lost), whereas DxO actually improves the sharpness in the corner, bringing out more detail in the planter and the stairs, as well as improving the color and contrast. As you'd expect, the TS-E offers the best sharpness and contrast, and the least distortion (which is why it's better to 'get it right in-camera'). The WB is set to daylight throughout, so in addition to the lack of distortion, you can see that the TS-E lens also has better color and contrast

Next I looked at noise reduction. This set of shots is from a PowerShot S95, where ISO noise is a bigger problem. This was shot at ISO 640 (the crop factor for the S95 is 4.6, so this is approximately equivalent to ISO 3200 on FF or ISO 2000 on a crop sensor).

The improvement in noise is detectable even when looking at the whole image, and pretty evident when looking at the 100% crop of my
daughter making friends with a Madagascar hissing cockroach. Sharpness is better with DxO, too. You can start to see the striations on the cockroach's antenna that are not evident in the DPP image.

Finally, I wanted to see how DxO would do at 'image rescue'. Over the summer, a red-tailed hawk spent some time right next to our house, from late afternoon to
after sunset. At one point after sunset, he flew across our yard and landed on a branch underneath some trees behind our house. I had my 7D on manual with Auto ISO and a fast shutter speed (1/800) to catch him in flight, resulting in an ISO 3200 shot that was very underexposed. He perched on the branch for only a second or two, and was gone before I could change the settings to a slower shutter and bring down the ISO. So what I was left with was a noisy, underexposed shot which was unusable. The image on the left is pushed to +2 EC in DPP, with strong NR. The image on the right was processed in DxO using the High ISO Noise Reduction preset and was pushed to +2.5 EC (DxO allows up to 4 stops, vs. 2 stops with DPP).

The DxO result is far from perfect, but is much, much better than with DPP - the DxO version would probably be usable as a small print.

At the back end, there's a noticeable difference in the time it takes to convert the edited images from RAW to JPG - DxO is quite a bit slower. However, this is not a big
deal to me since that's unattended time (I can let it run while I'm busy during the day, or even overnight). The actual hands-on editing is faster with DxO (I think DPP applies more 'on-the-fly' which slows it down, or it's just not coded as cleanly). I suspect the total time from start to JPG is similar, but with DxO less of that time is attended.

DxO is an image correction program only - there's no clone/stamp/heal tool, etc., although for simple image artifacts the dust tool does a decent job. ersonally, I only rarely do selective editing like that and will just Photoshop the JPG prior to printing when I need to remove a blemish, etc.

So, overall I'm quite pleased with DxO, and I think I'll be using that instead of DPP going forward.

Hope this helps anyone interested in DxO or wondering where to go from DPP. DxO offers a free 30-day trial, and right now the software is on sale (through December 25th) for 33-40% off - $99 for the Standard version and $199 for the Elite version (same features, but the Elite license is required for use with 1-series and 5D/5DII bodies).


--John