For those who don’t want to read a long post:

Overview of photo processing and management

Digital photo collections grow, often at an alarming rate. Ease of management and processing is required otherwise you’ll lose images or never get around to processing them.

You need:

• A photo management tool, or a strategy if you don’t use a tool.

• A RAW converter, if you shoot RAW.

• Tools to enhance, de-noise, resize and sharpen your images.

Photo management

At its simplest, photo management means storing your photos logically on your computer, so you can find them again. Usually, you set up a folder structure that meets your needs. I use: Location->Year->Date. Another option is Location->Sublocation->Year->Date. Another is Subject->Year->Location->Date. It’s up to you to figure out what works for you. If you only shoot one type of photo, then a simple folder structure might be enough. But if you are into several areas, then it all gets complicated very quickly. This is where digital asset management software, known as DAM, comes into the picture.

Of the tools discussed here, DXO Photo Lab has decent DAM capabilities. Luminar NEO has limited DAM capabilities. DXO Pure RAW and Topaz Photo AI have none.

DXO’s Unique Abilities

DXO, both Photo Lab and Pure Raw, have access to DXO’s legendary database of camera and lens analysis. These tools know the strength and weaknesses of each combination of lens + camera and remove vignetting, chromatic aberrations, distortion and correct lens softness in a targeted way. It is not a generic algorithm but is tailored to the image’s needs. The result is the best possible version of your image. It also has class leading DeNoise technology that few can match and none can beat. However, their best de-noise tech only works with RAW images.

DXO Pure Raw is limited to these corrections and can save the image as a RAW file (DNG), which you then edit in whatever RAW converter you use.

DXO Photo Lab or Luminar Neo?

I think that’s the wrong question. DXO has things, mentioned above, that Luminar Neo does not. But Luminar has functions that DXO lacks. The two products complement each other, and I would not choose to be without either of them.

Here is a small comparison of the two:

RAW De-Noising

Digital cameras produce noise at high ISO settings. The higher the ISO, the more noise. Noise removal is not easy. Software has to distinguish between fine detail and noise. It doesn’t always get it right and that corrupts the image. Software used to have algorithms that tried to detect noise and replace it with something that blended into the image. Modern AI tools offer deeper analysis of the image and the ability to better distinguish noise from detail and replace the noise with natural looking pixels.

I decided to do three tests. A night-time image of a road, houses and a signpost with words; an astro-photo of Jupiter and its moons; a daytime photo with less noise. The first two tests push noise reduction to its limits.

The night-time scene

Taken at ISO 6400 with an APS-C camera, the noise is high but not as extreme as the astro photo. All three tools have done a good job with the noise reduction, with Topaz AI and DXO having an edge over Luminar. But all the results are more than acceptable and totally usable.

Here I did two conversions with DXO—one with all its lens and camera corrections enabled and the other without them. The difference is noticeable. With them enabled, DXO’s image has more ‘pop’ and the lettering is sharper. It is for this reason that I prefer DXO for initial RAW development, including noise reduction. As it can save the image as a DNG, further RAW processing in Luminar, or another RAW converter is possible.

Comparisons – with DXO doing just noise reduction

Comparisons with DXO’s Noise Reduction plus camera/lens corrections

The astro scene

I am not an astro photographer. I don’t have a camera or lens that is really suitable for it, so please forgive this image. Please look at the white dot with smaller dots on either side. This is the planet Jupiter and three of its moons. The challenge for the software is to remove the noise, and there’s plenty of that, without adversely affecting the celestial bodies.

Here is a comparison, at 200%:

Here, the newness of Luminar NEO’s denoise AI is evident. The noise has gone, but so have the moons! DXO has removed the noise but also one moon is missing. Photo AI has removed the noise but has retained all three moons.

Admittedly, it’s a terrible picture, and the third moon is barely visible in the unprocessed image. But Photo AI did the best job here.

The daytime scene

I took this image in the UK lake District on a dull and rainy day. I used a full frame camera, but am at ISO 3200. So, there is noise to remove and the fine detail in the photo makes it a challenge. Viewed at 200%, the noise is a fine speckling.

I used DXO’s DeepPRIME XD noise reduction, Photo AI’s Normal RAW denoiser and Luminar NEO’s Middle denoise setting. I left the other settings at those suggested by the tool.

Viewing the results at 200%, you can see that all three apps have removed the noise. DXO, which also does its lens corrections, had given the image a punch of sharpness as well. Luminar’s image has a little less detail than DXO and Topaz. It’s not much and wouldn’t show up when viewed normally, but it does show that DXO and Topaz Photo AI have a small quality edge when it comes to developing RAW images.

Non-RAW denoising

This is a tougher test-removing noise from a noisy JPEG. JPEGs have less detail than RAW photos, so the tools have less to work with. Interestingly, DXO’s highest quality noise reduction does not work on non-RAW images and its ‘regular’ noise reduction is inferior. So, this is a test between Luminar and Topaz Photo AI. As Photo AI is based on Topaz’s DeNoise AI product, it has the advantage of being the maturer product, and it shows here.

This comparison is based on JPEG versions of the above images.

The night-time scene

Topaz Photo AI suggested using its strongest setting to remove the noise, but I found it to be too strong for my tastes. It made some writing on the sign illegible. I was happiest with the result from the Normal setting, recovering some image detail as well.

Luminar was not able to remove all the noise from the image. DXO could not either. A year or two ago, their results would have been regarded as excellent. But Topaz Photo AI’s noise reduction is in a class of its own when used on non-RAW images.

The astro scene

The results speak for themselves. Topaz is the clear winner, although Jupiter is down to two moons. Looking at the original JPEG, this is not surprising. I’d say that Topaz Photo AI’s result here is remarkable, given what it was dealing with.

The daytime scene

Both Topaz Photo AI and Luminar NEO have removed the noise. The Topaz version has more detail. A year ago I’d be extolling Luminar’s result, but Topaz is ahead of the curve. This reflects its maturity and Luminar’s relative inexperience.

Noise reduction – Conclusion

The quality of all the apps is amazing. For RAW images, I will stick with DXO (either Photo Lab or Pure Raw.) It produces clean results and has the big advantage of its camera+lens corrections. It usually produces the cleanest, sharpest denoise results. The only exception was in my astro photography test, where Topaz AI did better.

In truth, all three apps produced excellent results, but DXO is the best overall.

For non-RAW images, Topaz Photo AI is currently the pack leader. Unless DXO expands its best denoise algorithms into the non-RAW area, it won’t ever compete here. Luminar is chasing hard on Topaz’s heels, and I think it’s a great effort for a new release. Competition will ensure that both Topaz and Luminar work hard to improve their results. It will be interesting to see how they compare in six months. But for now, Topaz Photo AI is the best non-RAW image denoise tool.

The sharp image

This is the easiest type of image to sharpen. The source image is sharp and just requires some enhancement. The challenge is to provide just enough sharpening and to not introduce noise or unwanted artefacts.

Luminar

When opened in Luminar, the following options are available:

• Universal correction

• Motion Blur correction

Three strengths are available: Low, Middle and High.

Luminar can also detect faces and provide specialised sharpening for these.

Luminar also has advanced masking technology that enabled me to select the mountain, the foreground, the sky or any combination of these. The mask can also be edited or created manually, but for this image it correctly identified the elements in the photo. I was, therefore, easily able to select which parts of the image to sharpen and which to leave alone.

There’s nothing wrong with the options or the results, but the execution is painfully slow when generating previews of the settings. There’s time to go and make a cup of tea while it is generating a preview. Thankfully, it caches the previews, so comparing Low/Middle/High settings isn’t bad once the previews are created. But the time waiting for the initial previews is painful. Hopefully, they will improve performance in future releases.

Topaz Photo AI

Topaz offers more options that Luminar:

• Subject only: Photo AI detects the subject of the photo and offers to only sharpen it. In this image, it correctly identified the mountain and foreground as the subject. It usually doesn’t make sense to sharpen clouds, so Topaz made a good choice here. You can edit the mask it generates to identify the subject, and you can select to sharpen the whole image. The flexibility is good and being able to manually edit the mask is a good option. It’s not as fine-grained as Luminar, but it leaves you with control to correct errors.

• AI Model: Standard, Strong, Lens Blur, Motion Blur.

• Strength (0–100)

• Minor Denoise (0-100)

Where Topaz Photo AI excels is speed – the preview image is updated rapidly, allowing you to asses the options painlessly.

The result

There’s not much difference between the results. For a high quality image, I would be surprised if any sharpening tool struggled to produce a good result. I think Topaz’s result is very slightly cleaner, especially when viewed at 200%. But both results are excellent. The main difference is the speed – Topaz is fast but Luminar is sloth-like. Hopefully, Luminar will improve the preview generation speed in future releases.

The blurred image

Here I present the tools with a photo where I tried to handhold the camera with a shutter speed of 1/25th second for a 50MM lens with no image stabilisation. Also, I shot it at f/1.4 handheld, increasing the likelihood of it being out of focus. This has resulted in some motion blur. Even with today’s image stabilisation and high ISO settings, some pictures may still suffer from motion blur. Can these tools really save such an image? Let’s see.

Luminar NEO

I selected Supersharp… Motion Blur… Middle strength.

After another long wait, Luminar produced a very decent result.

Sharpening: Topaz Photo AI

Topaz offers both a Lens Blur and Motion Blur option. For this image, the Lens Blur option with strength of 32 (out of 100) offered the best results.

The results

There are definite differences between Luminar’s and Photo AI’s result, and this no doubt reflects the different AI models they use. Rescuing a blurred/out of focus image requires a degree of creativity by each app, as they are trying to recreate what a sharp image would look like. So, it’s not surprising that they have come up with different results. I think both are usable. Of course, this type of sharpening is only needed in an emergency, when an image is spoiled by bad focus or blur. Getting the picture sharp when you click the shutter is the best approach.

Upscaling comparison

Topaz Photo AI offers many settings, reflecting its maturity. It offers upscale by 2x, 4x and Max (usually 6x), as well as setting the desired size in pixels, inches and centimetres. It has four different AI models, and I chose High-Fidelity for this image.

Luminar offers far fewer options: 2x, 4x and 6x and two AI models.

Both offer face detection, which helps to keep faces looking natural, although that’s not required here.

Both Topaz and Luminar took about 90 seconds to process the image. But there are significant differences in the output. Firstly, Topaz offers to save the image as TIF, JPEG, PNG and DNG. Luminar saves the upscaled image as a JPEG, with no options to control its quality. Its JPEG output was 1/4 the file size to Topaz’s, with Topaz’s quality set to 95%. That means Luminar is heavily compressing the result, losing many pixels in the process. With upscaling, you need all the quality you can get, so saving as a medium quality JPEG is not a great idea.

In the comparisons below, the image labelled ‘Original’ is upscaled using the BiCubic algorithm.

Viewing the results at 200% shows that both the Topaz and Luminar results are vastly better than the BiCubic algorithm, but the Topaz result has finer details than Luminar’s. Luminar’s result is good, but Topaz’s is better. If I were making a large print, I would use the best quality image I could. Large prints are expensive and are usually for my customers. Quality matters.