This review represents what I think of Topaz Labs Video Enhance AI, after using the product over a few weeks in quarantine with various video types.
I am working with the trial version currently. My thoughts based on my tests – so far:
For upscaling high quality HD video (shot on a Lumix GH-2 originally) to 4K, this works very well with a noticeable detail improvement. Looks more like 3K video than 4K, but it is a very nice improvement.
For upscaling old 640×480 video, the enhancements are very limited. Depending on the upscale model used, the enhancements could look a bit like over done noise reduction with sort of a cartoon quality to the final result. I did use it successfully on some old archival footage (from poor quality B&W film) and it did a subtle bit of noise reduction which was useful, but not super valuable for the effort it required. Better noise reduction would have also looked frame to frame to identify dust and scratch marks and eliminate them – but I did not see that happening.
Some reviewers (like the one reviewer above) say they have ripped high quality SD DVD content and gotten relatively good upscale results to an HD-like level.
My view is that Video Enhance AI has some value in noise reduction of old films, but didn’t remove dust and scratches. At $199, its fairly pricey and time consuming to process old films with only a small benefit. If instead, the goal is to upscale some old HD video – those results are quite noticeable and good. If you need that, it might be a handy tool to have available.
Another product Topaz Gigapixel AI provides upscaling of still images. Some people have output video to JPG images and the used Gigapixel to upscale all of the images. I found Gigapixel only works on relatively high quality images. In other words, don’t try to scale up a 640×480 image – its barely usable. However, you can upscale an HD still (1920×1080) to a pseudo 4K (3840×2160) image. The more detail you start with, the better the upscaling.
I took some 16 megapixel images and did a 2x upscale – and this did indeed produce generally nice, some what more detailed, high resolution photos.
My camera (Lumix G9) has a built in high resolution mode that uses the in body image stabilization to move the sensor in tiny increments, taking 8 photos in high speed succession, and then integrating them into a super high resolution photos. This works very well for stationary subjects. Some have instead shot a conventional photo (20 MP) and then used Gigapixel to create a pseudo high resolution photo – with surprisingly decent results.
I view Gigapixel as another tool to consider – but not a panacea.
I highly recommend Topaz Denoise AI, and Topaz Mask AI. I use both of those all the time. I also use Sharpen AI some of the time – but Mask AI has its own sharpening capabilities and much of the time, I prefer those results to those of Sharpen AI.
Sharpen AI does have two unique features – the ability to rescue a blurred photo (from camera shake) or one that is slightly out of focus. I have tested the blurred rescue and it works quite well!
Anyway, I definitely recommend Denoise AI and Mask AI. Hands down excellent products.
Video Enhance AI is fine if used primarily to upscale HD to higher resolution – and time is not a big deal as this can take quite a while.
Gigapixel AI works very well on clean input images of good resolution. But don’t expect it to upscale poor resolution images.
Sharpen AI works well too – but sharpening is perhaps not my thing. However, I have found it did an amazing job on some blurred photos taken slightly telephoto from a moving boat. That was pretty cool.