A clash of titans!
In my opinion, DXO PhotoLab and Photo Ninja are the best RAW converters. Others are good. But, as pure RAW converters, these two are head and shoulders above the others. Things, of course, change over time. Skylum is about to update Luminar. It’s already very good. Who knows just how good it will be? DXO have just released Photo Lab V2 so now it a good chance to see how it stacks up against Photo Ninja.
New In Photo Lab 2
Photo Lab 2 features the following improvements:
- Improved U Point Technology
- Better Noise Reduction
- Clearview Plus
- Photolibrary
- Support for DCP colour profiles
U Point tech is DXO’s brilliant approach to making local selections. First seen in NIK’s Viveza tool it is the most effective way to make local selections in an image. No messing around with masks, no fiddling with selections. Just click and size. Done!
Clearview was one of the main reasons for using DXO – it does what it says. It removes haze from images without making them look unnatural. And now they claim to have improved it…
Photolibrary is DXO’s first foray into Digital Asset Management (DAM), thus really encroaching into Lightroom’s territory.
DCP color profiles are the way to ensure that the colours you capture with your camera truly reflect what you saw. As DXO point out, if all your software supports colour profiles then you can pass the images to different programs without corrupting the colours. Not all software supports them, but Photo Library and Photo Ninja both do, so it’ll be interesting to see how they stack up.
RAW processing
This review will focus on RAW processing in the following areas:
- Colour rendition
- Noise reduction
- Highlight and shadow recovery
- Sharpening
- Image clarity and colour enhancement
Both DXO and Ninja offer tools in these areas. Ninja has no DAM capabilities so they won’t be compared. Neither does Ninja offer local selections and editing. Ninja is 100% a RAW development tool that can also process regular images. DXO can do the same along with local edits. However, DXO can only develop RAW images for supported cameras. Even converting the image to .DNG doesn’t help. Ninja, in contrast, is happy to process .DNG files from unsupported cameras. So, if you have the excellent Canon M50, which produces .CR3 files, then DXO will refuse to perform RAW conversions. Pity.
Colour profile support
In my Ninja review, I pointed out how well it supports these. Here’s an image with a colour profile target in it. Let’s see how they both handle it:
Ninja can build a colour profile from this image. DXO needs the profile to be built from 3rd party software (which comes free with the Colorchecker product. Once built and applied here are the outputs with all other processing options switched off:
[URISP id=533]
This is a bit weird – Ninja’s results are warmer and richer than DXO and its tones are closer to the original image. So, I put the same image into Luminar and applied the RAW profile. Its result was much closer to Ninja’s but less saturated. Hmmm. The image, in theory, should be identical in every program but it is not. I have no idea which one is ‘right’ and I’m sure all of them provide a decent starting point for further work. I really like the way Ninja organises profiles and I love the warm, rich colours. But a colour profile is supposed to standardise the colours and this isn’t happening.
Score: impossible to tell… Ninja’s colours look nicer but are they right?
Oh well, back to the test…
Noise Reduction
This is something we can easily test. I’m going to use my standard test image for this one:
Original Image, no processing
This image has some subtle noise in it, which is most easily seen in the sky:
It is not bad, so neither program should have any problems dealing with it. But what effect will this have on the rest of the image?
Here’s DXO’s result, using Prime Noise reduction:
And here’s Ninja’s, using just Colour Noise reduction:
Both are noise free and both have retained the same level of detail in the mountains.
Score: Even, although DXO took a very long time to process the image, whereas Ninja is almost instantaneous.
Maybe this was not a stern enough test…
So, here’s a horribly noisy 1600 ISO image for them to try on:
Don’t Panic indeed! Can they handle it?
This image could indeed induce panic into any noise reduction program. The noise is horrible and the detail in the image requires that it be maintained despite the noise reduction. After much fiddling, here is the best result I achieved from each program:
DXO
DXO has managed to remove most of the noise but at quite a cost – there’s a nasty new noise around the yellow text. I couldn’t get rid of this and keep the image noise free.
Ninja
Ninja has, in contrast, done a much better job. The Noise has gone without too much loss of detail and no nasty artefacts. I was quite surprised by this result. DXO’s Prime Noise reduction is very highly regarded. It is touted (by them!) as ‘often imitated but never equaled‘ yet has been surpassed by Ninja. Then again, Photo Ninja’s noise reduction engine is very mature indeed, being the Noise Ninja engine that has been in existence for years and years.
So, for basic noise both programs work well. For truly nasty noise, Photo Ninja is better. Much better. And Ninja takes a few seconds to perform the noise reduction. DXO takes a very long time.
Point to Photo Ninja.
Highlight and shadow recovery
Back to the original image:
This image’s dynamic range was too great even for the excellent Samsung NX1. The clouds are over exposed and the bottom left corner is too dark. What can the converters do to help?
Photo Ninja
Given that Ninja has a limited set of available adjustments, and has no local edits, I will just present one image from it in this section:
This is a decent result. Remember, no colour enhancement has taken place yet – the image needs more contrast, etc. But there is some detail recovery in the clouds and the shadows have been nicely lifted.
The settings I used for this were:
- Smart lighting
- Highlights: -1
- Shadows +0.70
- Illumination: +9
DXO
DXO offers both global and local adjustments. Here is my best result using global adjustments:
This is similar to Ninja’s. Ninja’s image looks a little more detailed and the main differences are how the clouds look.
I used the following settings to achieve this result:
- Smart Lighting: Slight
- Highlights: -55
- Shadows: +15
- Blacks +13
But what about local adjustments instead? Here I applied Smart Lighting: Slight globally and locally adjusted the clouds, sky and dark areas:
I prefer Ninja’s highlight recovery in the clouds but DXO has pulled more detail out of the dark area bottom left, simply because I was able to target just that area.
Result: Hard to judge. Both converters have done a good job. Ninja has better highlight recovery, DXO had better shadow recovery when applied to a targeted area.
Sharpening
DXO has image sharpening to die for in that when it ‘knows’ where a lens is sharp and where it is soft it sharpens accordingly. If it doesn’t have a profile for the lens then it offers basic Unsharp Masking.
Ninja offers its own custom sharpening algorithm, which is intended to compensate for the effects of an anti-alias filter. It also has a feature called ‘Detail’ which provides additional sharpening.
Let’s see how they did.
DXO
Ninja
Evaluating these results is interesting. As we are only sharpening, I’ve turned off colour and highlight/shadow adjustments and just left on Noise Reduction and Sharpening. DXO’s image is a little lighter as a result but, as can be seen in the rocks and water, is a little noisier. DXO also has a slightly painterly look to it. All in all, there doesn’t seem that much difference to me in the sharpness of these samples…
Image clarity and colour enhancement
The real test, I suppose, is to see what happens when the full capabilities of these tools is thrown at the test image. Ninja has a set of colour enhancements and DXO has a huge array of settings that can be used to tweak the image. Without extending this already long post too much, let’s look at the final results:
[URISP id=558]
It’s really, really hard to pick a winner. With DXO, I switched on its enhanced Clear View Plus option and this removed a lot of haziness from the image. I used local adjustments for the sky, the shadow areas and the highlighted ridge on the right hand side.
Ninja only offers global adjustments, but it has some unique tools for controlling exposure and setting colour saturation and brightness.
When it comes to the quality of the resulting image, I think Ninja has a tiny edge in terms of noise reduction and sharpness. DXO has not been able to keep the image totally noise free, but this new release produces far more detail than DXO Photo Lab V1. Detail wise, there is little to choose between the two programs. Clear View Plus has dramatically enhanced the image with no discernible drawbacks.
Conclusion
DXO Photo Lab V2 is a big improvement on V1. My previous, albeit brief, comparison of DXO V1 vs Ninja showed Ninja to be better. Now the results are much closer. Both have their strengths and weaknesses which I summarise below:
Photo Ninja – Pros
- Very fast processing
- Excellent sharpening and detail – probably best in class
- Excellent noise reduction – best in class
- Excellent chromatic aberration removal
- Excellent colour adjustments and enhancements
- Results have a crispness to them that is hard to describe but easy to appreciate
- Easy to use
- Supports .DNG files even from unsupported RAW formats
- Best colour profile handling of all the software I have tested
Photo Ninja – Cons
- No local edits (although this is somewhat mitigated by its excellent ability to edit individual colours as well as exposure, highlight and shadow recovery)
- No Digital Asset Management facilities
- Lack of built in lens profiles for distortion and vignetting correction (although the software can be ‘taught’ to do this)
- Lack of built in noise reduction profiles for individual cameras (although the software can be ‘taught’ to do this)
DXO Photo Lab V2 Pros
- Good image sharpening, distortion removal and vignetting correction for supported lenses
- Massive range of available adjustments
- Excellent, class leading, local adjustments without need to draw masks
- Support for colour profiles (although I’m not sure that it is using them properly)
- Better Digital Asset Management facilities than V1, but I’ve not tested them enough to determine their usefulness.
- Clear View Plus – probably the best image clarity enhancer in the business. It is so easy to use, as well.
- Integrates with Lightroom
DXO Photo Lab V2 Cons
- Very slow ‘Prime’ Noise removal and not as good as Ninja’s
- Slow image processing
- No support for .DNG files from unsupported RAW formats, which means that Canon .CR3 files cannot be processed in the current release (October 2018)
- Noise removal not as good as Photo Ninja’s, although still very good in its own right
- Images have a ‘painterly’ look to them when viewed at 100% (this isn’t likely to be noticeable at usual viewing sizes)
Prices (26th October 2018)
DXO Photo Lab V2 Elite costs £159 GBP, although it’s on offer for a while at £119.99. The upgrade cost is currently £59 – a decent saving.
Photo Ninja costs $129 USD (£100 GBP using today’s exchange rate). The upgrade cost from previous versions is $59 USD (£46 GBP).
Which one should I get?
There’s no easy answer to that. This release of DXO Photo Lab isn’t making me want to reprocess all my images but there will be some where its local adjustments will be needed and its Clear View Plus tool is so useful.
Here’s one final image – the Photo Ninja result given the DXO Photo Lab Clear View plus treatment along with DXO’s lens distortion correction…
DxO has so much more to offer than Ninja. No way I’d take that over PL. However, I’ve always thought PL noise capabilities were a bit hyped and your comparison proved it. Nice job.
Thanks for article, I’ve been using Photo Ninja for years and have just tried PhotoLab 2. Reference your findings about the high ISO NR I have to disagree with your findings about PN doing a much better job. By default PL applies high lens sharpening, which at higher ISO settings causes unpleasant noise around the edges, which is what you have on your ISO 1600 image. Turn this lens sharpening down quite a bit or even off and the image looks much more like PN except away from the edges PL has a smoother, finer grain looking noise. Even at low ISO this lens sharpening is best turned way down.
I also tried both at a very high ISO setting of 25,000. At first I thought PN did a better job of colour NR as at 100% it removed all the colour noise and at 100% PL still had a few odd bits of colour when viewed at 100%. Then I noticed that PN had removed completely some small yellow rings from a £20 note in the image ( and had removed wanted colour from other areas ), yet PL had completely retained them. I had to set PN’s colour NR slider to 50% to bring the rings back to an acceptable level, but then I ended up with unacceptable colour spots over a lot of the photo.
Although PL lens sharpening is very aggressive in small amounts it can be used at any ISO to give more detail than PN – I used the detail slider in PN to match the detail in PL in an ISO 6400 image, but PN ended up looking more noisy overall.
So overall for NR I would give the slightly edge to PhotoLab 2.
Interesting points. I don’t often shoot at high ISO so it’s not really an issue for me anyway.
These days I’m tending to use Topaz AI Clear or Topaz Noise AI for noise reduction as I find them even better. But I haven’t done a proper side by side test of them – I just like the look of their results.
I had a go at an ISO 800 image with PN, PL and Topaz and no matter what I did with the sliders in Noise AI I couldn’t keep the detail and remove the noise like I could in PL. AI Clear looked better on this image, but even with the lowest sharpening setting it’s a applying quite a lot of sharpening which makes the noise look a little worse. If you could lower the sharpening it would probably look a little better than PL.
I then tried a ISO 6400 image and got the images looking fairly similar though Topaz AI Clear did look a little cleaner and a touch sharper than PL.But that’s just a couple of images and there isn’t really one NR program that will do the best on every image.
Yeah, noise is often image specific and there’s not a one size fits all solution.
Then again, with sensor resolution being so high on even modest cameras I sometimes wonder if we obsess to much about it. At 100% on screen the noise from these high resolution sensors can look awful. But, unless you’re intending to make a massive print then you’re going to downsize the image considerably for the final result.
Downsizing an image tends to reduce noise (in my experience), and so many images these days seem to be viewed on phones and tablets.
It’s much like lens sharpness – upgrading to a 10% ‘better’ lens may not yield visible results at regular viewing sizes…
Yes I’m just doing it to make sure I’m happy with PL before I buy it.
I also compared a ISO 200 image between PL and Topaz and although I got them looking pretty similar there was a couple of areas where PL removed detail as I guess it confused it with detail but was still there in Topaz. On further inspect I found a couple of ares where Topaz had removed detail and in PL it was still there.
Most modern NR seem fairly close, it’s whatever works best for you and you workflow.
Agreed. My last landscaping trip was 100% using a tripod and 100% ISO 100 using a full frame camera. Noise? What noise?
I also developed some of the images with DXO PL 2 and Photo Ninja and always preferred the PN results to DXO even though DXO had lens/camera specific corrections and Ninja didn’t.
I used a ColorChecker passport to take a series of images to teach Ninja about my camera’s colour performance and, when using this custom profile, its results are beyond pleasing.
I still own DXO and would use it when needed. Its chromatic aberration removal can handle things that Ninja struggles with – mainly because it has built in specific corrections for the camera and lens whereas Ninja doesn’t and needs to be taught about them – this can take a lot of time and trial and error.
I don’t think there’s any one size fits all solution. Sometimes I use Skylum Luminar, especially when I need to recover significant shadow details. Luminar does that better than the others, I think.
But Ninja works for me 80% of the time, with a bit of Topaz AI Clear.