Posted on 28 Comments

The Story Behind DeNoise AI

Why did we introduce Topaz DeNoise AI?

As photographers, we all have situations where we end up with noisy photos, like when we’re shooting in low lighting or shooting fast actions.  Even for more traditional photography, we sometimes find noise in shadow areas when adjusted in post editing. Over the last decade, Topaz Labs has been dedicated to developing the best possible tools to address this problem, along with other common editing hardships.  Topaz Denoise AI is our latest effort.

As many of our users know, our existing products “Topaz DeNoise” and “AI Clear” are commonly considered the best in photo noise reduction.  So why did we produce another one? The short answer is that the new DeNoise AI is even better, both in image quality and ease of integration with your photo editing workflow. Since we’re constantly looking to provide a better customer experience, this was the logical next step in noise reduction software.

Topaz DeNoise is one of our old plugins. This program used traditional image processing techniques and has many fail cases when the noise amount and types of images vary. Later in AI Clear, we saw that AI technology based in machine learning was very effective at removing noise and  blur, while also giving us crisp and detailed images for different types of noises. AI Clear exists as an adjustment within the framework of our editing software Topaz Studio.After a time, we found we were not quite happy that we could not put our best AI network in AI Clear because of this limitation. This was for two reasons. One, the AI technology demands a lot of resources and we were limiting the resources available by using the software within Topaz Studio. Two, AI Clear had to be interactive and work seamlessly with other studio adjustments. These reasons prevented us from using the most powerful AI networks at our disposal. Therefore, we decided that we would come up with a standalone and plugin version of this software where we would not compromise on the AI itself.

In DeNoise AI, we utilized a new software architecture to accommodate our best AI network. Albert, our CTO, dedicated all of our training machines to train different network variations. Our developer Bowen implemented all network variations in the Topaz AI inference engine and validated the quality and performance of the networks. After months of training and testing, we are now able to use a bigger network that handles more varieties of noise. DeNoise AI is trained to remove high ISO noise, sensor noise, thermal noise, banding noise, and noise from some scanned images. You can also provide hints about the type of noise through two sliders if not satisfied with the default output to improve the processing. Let’s look at some examples below,

The original image on the left has very strong noise and banding artifacts. The DeNoise output is clean and some details are generated in the very dark noise area.

This image has very strong banding artifacts in different levels. The AI removed those banding artifacts and generated some convincing details.  

Finally, check out two more examples of high ISO noise and the output from DeNoise AI in default settings.

Now the question is, in terms of noise and artifact removal, how can you get the most out of DeNoise AI? To do that, I’ll explain what is going on behind the scenes. In order to train DeNoise AI effectively, we used a lot of clean and noisy image pairs as examples to train an AI network. This is so that later when the AI sees a noisy image, it can predict the clean version of it. Mathematically, we mimic the way that real-world noise is added to the image, during capture or digitization. Our noise model generated millions of noisy-clean image pairs, which were used to train the network. The DeNoise AI network is also able to receive “hints” about the noise level and type from outside as a help.

There are three sliders you can also control from the interface if you don’t like the automated detection. Pushing the sliders to the extreme level will not give you the best output all the time. If your noise level is low then keeping the slider in the lower range should give the best output. With DeNoise AI, you now you have more fine control on the output through the slider rather than just a few steps.

So, is DeNoise AI better than AI Clear? What we can tell you is that they are totally different types of AI network. Personally, I like to play with the Remove Noise and Enhance Sharpness sliders and see the outcome in different images. In our experience, DeNoise AI will typically provide a far better output than AI Clear. But you never know! We encourage you to give both a try.

If you have already played with the trial version or purchased your own copy, let us know what you think. We’re eager to hear how DeNoise AI works for you!

Posted on 105 Comments

Let AI Sharpen Your Photos

As photographers, we want to shoot sharp photos, period!  Our goal of getting sharp images is one of the major factors in the thousands of dollars we spend on good cameras and lenses with features like optical stabilization.  We’ve also invested time in learning techniques, tips, and tools for shooting sharp photos, like lugging around a sturdy but heavy tripod, setting proper shutter speed, learning about depths of field, proper focus, and so on and so forth.

Even with the best equipment and expert techniques, we sometimes still end up with images that could be sharper. That is why almost every photo editing software provides a “Sharpen” feature.  For example, Photoshop has no less than 6 different “sharpen” filters on my last count.

While we hope that getting a sharp picture will not be a problem when using a sharpening software, we know that’s not always the case.  Unless your image has a very small blur, existing software “Sharpen” functions do not work well.

Now let’s go into a technical explanation about how a given software sharpens your images. Up until now, the photo sharpening tools in software typically use one of two type methods: image filtering or deconvolution.

The first type of sharpening is Image Filtering. The basic idea is simple: blurry images lack spatial high-frequency components.  Therefore, applying a filter that boosts high-frequency will make the photo look sharper. Generally, this technique will works for photos with just a little blur.  Since this method is fast, most software sharpen tools use this method. In Photoshop, “Sharpen”, “Sharpen More”, “Sharpen Edge” and “Unsharp Mask” all use this type.  

Figure 1 (click to view details at 100%)

The Image Filtering method works well if you just need to sharpen the photo slightly, but if your photo has a bigger blur, it does not work well.  The problem is that when boosting high-frequency photo components, the tool will also also boost noise and artifacts in photos. Over using a sharpening filter that uses this method will create images with artifacts such as “fat edges”, “halo”, and noise amplification.

The second type of sharpening method is called Deconvolution.   Imagine you are looking through a camera at a star at night.  You will see, instead of a single bright point, a small blurry disk.  This disk represents the so-called point spread function. This function summarizes the blurring process as a mathematical operation called Convolution. Thus, removing blur is modeled mathematically as Deconvolution. This type of method was first developed primarily for astronomy and later for spy satellites. At its creation, normal photographers did not have access to this tech due to the computer speed and the stability of the algorithms.

The situation changed around 2010. There was a break-through in the deconvolution algorithm to make it feasible to use on a PC.  We were the first company to bring this tech to photographers when we released a Photoshop plugin, Topaz InFocus. This method works much better for small and moderate blurs.  It also works, to a certain degree, on blur due to camera shake or from a moving object. To this day, Topaz InFocus is still be considered to be one of the best sharpening tools on the market. Later, Photoshop also added “Smart Sharpen” and “Shake Reduction” based on deconvolution.

Figure 2. Large motion blur (click to view at 100%)

In addition to being much slower than the image filtering method, deconvolution needs to know the point spread function, which is very hard to estimate. It is also very sensitive to image noise and jpeg compression.  Therefore, it fails most of the time for large and complex blurs, such as one in Figure 2 (a). Since Hollywood’s crime dramas had (incorrectly) shown that they could usually turn very blurry photos into sharp ones to catch the bad guys, Topaz InFocus fell short of some of our customers expectations.  I took this quite personally and have been on the lookout for a silver-bullet ever since.

For eight years we did not find one…. not until recently. I felt we might finally have a shot to redeem ourselves thanks to the rapid development of Deep Learning in Artificial Intelligence (AI).

AI approaches the problem very differently. Instead of studying the mathematical model of the blur process and how to solve the inverse problem, we train an Artificial Neural Network with millions of blur-sharp image pairs.  The neural network will eventually “remember” what the sharp image should look like if it sees a blurry image. After months of training, the neural network could produce a sharper when image given an image it had not seen before. Figure 1 (c) and Figure 2 (c) are examples from our latest product, Topaz Sharpen AI.   Here is another example to sharpen an out-of-focus photo.

Figure 3. Out-of-focus blur (click for 100%)

Looking at Figures 1, 2, and 3 more carefully, you will find that Topaz Sharpen AI seems be able to create images with very fine details rather than simply sharpening the edges.  This is what makes Topaz Sharpen AI truly unique – it actually synthesizes convincing details even if the blurred image does not have any through the power of AI.

There is a reason that nobody has released an deep-learning sharpen AI tool for photographers so far.  It is quite an engineering challenge. For example, the neural network requires extremely high computation that makes regular PC pretty much unusable.  I am so proud that our team was able to overcome many challenges to bring Topaz Sharpen AI to you, from Dr. Acharjee’s  ingenious neural network architecture, to Image Processing Lead Bowen Wang’s efficient GPU neural engine development, to our whole product team’s application design.

You can try it yourself by downloading the 30-day free trial.  For anyone who owns Topaz InFocus or the complete bundle, it’s a free upgrade!

Does AI finally make blurry images a thing of the past? Not at all.  The example images I used are pretty much the best case scenarios to impress you with the AI.  You will find it is still a hit or miss situation sometimes. However, it is a huge step forward when comparing with any existing solution, including our Topaz InFocus. And best of all, we are just at the beginning of the Artificial Intelligence (AI) revolution.  You can be assured that there will be many more amazing things to come.

About the Author

Albert Yang founded Topaz Labs over 10 years ago, to form a company that adopts and implements the latest technology to introduce cutting-edge tools to the Photo market. With over 30 years of programming experience, he’s proud to offer his technical expertise to our users as the primary developer of our latest tools.