Testing PureRaw with DeepPrime 3 by DxO
Over the years, I tested a lot of software for denoising or sharpening. Recently, I came across the new version 6 of PureRaw by DxO. At 130€, it looks reasonably priced for the claims they make. You get a RAW converter with denoising and some basic functions, and a Plugin for Lightroom. Moreover, it has a test version with full functionality for 14 days. So, why not give it a try?
There is a desktop version as shown below. Handling this simple interface is quite intuitive It offers two denoising algorithms, where the more complex XD3 takes considerably more time. On my system, I would stick with the more elementary DeepPrime 3 for a smooth workflow. I did not see too much difference anyway.
The software does also feature "spec removal" and "automatic lens sharpness optimization". I cannot tell you much about that. But it may help to improve the noise reduction in general.
You can also call the interface from Lightroom via a Plugin, either with or without a preview. I prefer to do that. The comparisons on this page have been done using Lightroom. I did later fix the exposure and other details there.
The plan is to compare the algorithms to Lightroom's built in denoising, the simple one and the AI tool. The shot at the start is only at ISO1600 which should not challenge any software. Since that is the most I usually need for my photography, my desire for an extra denoising step is limited. But birders and other wildlife photographers tick differently. They need that extra range of ISO to 32000 or more.
So let us take one of this extrem cases, a macro shot at ISO32000 (that is indeed 32 thousand), taken with the beloved 105mm on the Nikon Z50 II. Assume, you had to take it that way because of wind movement or the necessary high DOF. A bird or a bee will often require the same. The result is terrible when you zoom in or print just a little larger.
![]() |
| 100% crop, ISO 32000, no noise reduction |
In the browser, however, the image is still completely acceptable. The reason is that the process of showing the image in a reduced format kills details and thus also noise. Here is a version of the full image where I cropped only a little bit on the top and bottom and applied my usual sharpening.
![]() |
| Full image, ISO32000, no noise reduction |
We can now try to denoise this with Lightroom's AI and also with the simple built-in denoise sliders, drawing them as much to right as necessary to match the AI result for noise. The problem is that you either smear details or have some grain remaining when you restrict yourself to the quick, but simple method. Have a look at the white spikes on the berry in the front.
| ISO32000, Lightroom Denoise Sliders |
Next, I ran the file through PureRaw. The plugin can be started with a preview window or in the background. After 15 seconds with the DeepPrime 3 algorithm, the result comes back. It is automatically stored in a collection. I show you a 100% crop after sharpening with 40%, because it is not unsharpened enough by PureRaw I think.
| ISO32000, PureRaw, DeepPrime 3 |
For comparison, here is the same snippet with the best result I could come up with Lightroom AI Denoise. The amount of this denoising is set to 50%.
| ISO32000, Lightroom AI Denoise 50% |
I found this result just as good, if not a little bit better. It takes longer, however. Finally, let us compare to DeepPrime XD3. This takes 3 minutes on my system which is far too long to apply it regularly.
| ISO32000, DeepPrime XD3 |
That's a very clean image and all details are preserved. If it is worth the wait depends on your needs.
Im summary, I decided that will not buy this software. Lightroom AI can do the same, especially for my small printing demands. Probably, the only benefit would be to use their more complex algorithm. If you need the best results, it may be worth it. After all, I find the software reasonably priced. And it does not need a subscription like Lightroom.
![]() |
| ISO32000, Lightroom AI Denoise at 75% |





Comments
Post a Comment
High there! Looking forward to hear from you.