CanadaMark
Offline
• • • • •
Upload & Sell: Off
|
 p.1 #17 · Denoising in DxO PL8 vs Topaz Photo AI | |
FWIW I've still been spending a ton of time comparing the various RAW converters as their versions are changing quite frequently. I feel this is the single most important step in post processing as it has the largest effect on image quality, plus it is the single biggest time saver, so it's critical to me that I'm always using the best one possible, for my work anyway.
I haven't bothered to put together any more comparisons to post on forums, only because they are quite time consuming to make and the conclusions/results haven't changed materially from the last ones I posted. For context, my comparisons focus primarily on real world wildlife images, rather than studio setups or anything like that, but I don't notice any significant performance differences when changing subject matter to, for instance, landscapes or macro.
DXO PureRAW 4.x in my opinion is still the best of the bunch by quite a large margin, and they are still doing many things that the other programs do not in terms of how the NR is handled, how the sharpening is done, near pixel perfect edge detection, and how the lens profiles/corrections are applied. It is also still the only program that builds an all-new RAW file from the ground up. Until the others catch up in those areas, it's hard to imagine them competing with DXO without some major innovation. I am still on their BETA development team because I believe in the software, I have no financial affiliation with them. I had to sign an NDA, but I can tell you one of my suggestions did make it into the public release of Pure RAW 4.0 though so that was nice to see. Batch processing is still the fastest, assuming you have a fairly modern discrete GPU as PureRAW 4 leverages the GPU much more so than earlier versions of the software. It will also use all the CPU cores/threads that you can throw at it, but the GPU is doing the heavy lifting. The primary downside of DXO remains the same - profiles aren't available immediately when a new lens/camera is released, and the reason for that is due to how the profiles themselves are created. Less popular or hard to get camera gear, drones, and smartphones can take a very long time to get support. At times, the waits can be quite long if the equipment is hard to get into their lab. Popular gear gets support more quickly but it still is not immediate.
Topaz AI still has all the same problems it has always had with edge detection, obviously fake detail, artificing, still no great way to batch process images, and as a company they are starting to lean into the "credits" business model with is almost universally hated by customers. I don't like the results, and I don't like the direction the company seems to be headed. Every time they advertise an improvement, I run my tests and never seem to come to any different conclusion. Maybe one day they will surprise me. YMMV.
Adobe is pretty good, but overall I still have the same complaints as it applies the NR more globally, and cannot compete with the sharpening and distortion correction modules DXO uses because of how those profiles are made. You can get the Adobe result pretty close to DXO, it just takes several extra steps which means spending more time. The main advantage here is you only need one piece of software and most people already own Lightroom/ACR. It also works on any image and you don't have to wait for specific profiles. This remains my #2 choice, and what I use for camera/lens combinations that do not yet have a DXO profile. I also think this is going to get a lot better over time with Adobe's vast resources.
Luminar, Capture One and ON1 aren't even really in the conversation as they aren't competitive, in my opinion anyway.
The only caveat I would add is that occasionally, for certain individual photos, one method might work slightly better than another, but it's not realistic when processing hundreds or thousands of photos at a time to try every method combination with every photo. What matters to me is what gives the best results most of the time and in the majority of scenarios. After I cull my images, I want to be able to dump all the RAWs into a single batch process and be able to trust that the results will be ideal or at least very close to it.
DXO releases new versions of PureRAW around March every year (Photo Lab is refreshed every Fall), so it won't be long until we see what they have in store for the next iteration 
|
