It seems like everybody is talking about AI tools nowadays, and unquestionably, AI is transforming many industries that impact our daily lives. Video and audio editing are no exception, and the list of tools available seems to be expanding by the day. Some of them are outstanding, and others… not so much. But the one thing they all have in common is that they are making sophisticated video and audio editing available to more people than ever before. However, that doesn’t mean that someone with no experience can pick up one of these tools and get professional results without any training or knowledge of how video works.
Before going any further, we must first be clear about what AI is. To begin with, I really hate the term "AI." It's a term that has permeated nearly every aspect of our lives, from voice assistants to self-driving cars and even recommendations for what to watch on streaming platforms. It’s often hailed as the future of technology, but in modern parlance, it’s really just become a lazy catch-all term for anything that seems too complex or mysterious to understand, akin to magic in some ways. It’s similar to “the cloud,” another vague buzzword that has become meaningless through misuse and abuse in marketing. A more precise term for what we generally call AI would be "pattern matching."
Many software tools and platforms claim to have AI capabilities when, in reality, they are simply advanced software that performs specific complex tasks. For example, an email marketing platform that uses data analytics to personalize email content is not AI but just sophisticated software utilizing data-driven algorithms. These systems demonstrate none of the characteristics that define true AI: Adaptability, learning over time, and human-like intelligence. It’s essential to be discerning when encountering the term “AI” in marketing. Understanding the distinction between genuine AI and sophisticated software is vital for making informed decisions as consumers and ensuring that the technology we embrace lives up to its promises. Clear communication and transparency are key in promoting a more accurate and honest representation of what technology can truly deliver. A good example of true AI in video editing would be Nuke's Copycat neural network.
One of the most significant consequences of the misuse and abuse of the term is the widespread perception that AI delivers instant results without any human skill or knowledge required. Just tick a few boxes, and it just works. In reality, those models are developed and fine-tuned by teams of experts who spend countless hours collecting and preparing data, training models, and refining algorithms. Software is only as good as the models its developers use to train it, and users are not somehow excused from having to understand how it works or what it actually does under the hood.
One tool that has garnered a lot of attention in recent years is Topaz Video Enhance AI (TVAI) from Topaz Labs. Despite its name, it falls into the "sophisticated algorithm" category, because, while it does depend on carefully trained models, it lacks the ability to adapt itself to the content or for the user to train the models. The models are simply downloaded and applied as they are, leaving the user at the mercy of the developers to create and train good models that may or may not be appropriate. A quick YouTube search will yield dozens of claims that it “works like magic,” that it “will blow your mind,” or that it’s “scary good.” Remember that many YouTubers who give these glowing reviews are either paid influencers or are otherwise compensated by Topaz Labs. That’s not in any way to suggest that TVAI isn’t good software – because it is – but it’s important to be realistic about what it can and can’t do well. TVAI is good at what it does, but what it does is not what most people think.
We use it here at ReelMagic as a cost-effective alternative to expensive hardware like Black Magic Teranex converters because we are a small specialty shop with limited budget, and I have a good bit of personal experience using it. I can say with some confidence that TVAI is really good at two things: upscaling and frame interpolation. More or less, everything else that it claims to do can be done better with other tools - many of them free - like AVISynth, Resolve, or Nuke. Topaz Labs markets it as a veritable Swiss army knife of video restoration that will stabilize, denoise, sharpen, deinterlace, deblur, dehalo, enhance faces, and deblock your video by stacking a variety of ever-expanding enhancement processes. And like most extraordinary marketing claims, the evidence to support them is mixed at best.
I would never attempt to do all of these things with one tool. Yes, it will denoise your video, but it won’t do it better than RemoveDirtMC or DustBuster. It will deinterlace your video, but it won’t be better quality than QTGMC or NNEDI3. It will sharpen your video but won’t do it better than LSFMod. It will stabilize your video but won’t do it better than Fusion or WarpStabilizer. The list goes on and on. Now, that's not to say that Topaz couldn't at some point down the road become a full-featured professional editing suite like Nuke or Resolve that will actually do all of these things on a professional level. But the development path they are on at this stage doesn't suggest that to be on the cards any time soon. They seem more interested in appealing to amateurs who either can't or won't make the investment to learn the intricacies of doing these things the right way and are looking for a shortcut. Even sticking to just upscaling with the standard Proteus model, a single video often requires different settings for different shots because the amount of grain and other artifacts can vary as the camera’s position and lighting change. Moreover, no complex process like upscaling will be perfect and will always leave behind artifacts that must be cleaned up manually in another application like Resolve or Photoshop. It can be a lot of work, and good results require a working understanding of digital video basics.
If you try to use TVAI as a one-stop shop for video restoration, you’re much more likely to end up with an over-sharpened, over-smoothed video than any sort of professional result. If you post something like that on YouTube, you’ll have no shortage of detractors to point it out for you. While Topaz Video Enhance AI is undeniably good in the areas of upscaling and frame interpolation, it's essential to understand its limitations. Despite the desire of some to use it for video restoration, it's really not a good choice for this. While it can somewhat improve the visual quality of older footage, it just tries to do too many things to do all of them well. Compared to the specialized features and algorithms of other tools, it's the jack of all trades and master of none. Using Topaz for this type of work is like the old saying, "When all you have is a hammer, every problem looks like a nail." For high quality restoration work, taking the time to learn professional software is a much better choice.
Topaz Labs is falling into the trap that sinks good companies faster than any other – trying to be all things to all people. Subway got into the pizza market. Sears got into insurance and banking. McDonald’s took on Starbucks in the coffee market. They all failed at these because they forgot what they were good at and lost focus. At some point, you have to choose your values and decide what you’re going to be good at. If Topaz focuses on its upscaling and frame rate conversion models and markets itself as a semi-pro alternative to hardware conversion, they can carve out a niche in the serious video editing market and take a bite out of companies like Black Magic. But as long as they try to be everything under one roof to people looking for shortcuts, they will be forever limited to the home enthusiast market.