Video-based generative AI

- We know that image-based generative AI is about creating new imagery from existing images or from text prompts. Well, what is video except for many images one after another? Take 30 frame-per-second video. That's 30 images every second. So in terms of AI technology, we're not only talking about the two-dimensionality of the video frame, but we're also talking about the added dimension of time. That means that video-based AI isn't just about generating new frame information, but also about predicting and generating new frames over time. Let's take a look at some of the possibilities, starting with some basic video enhancements and restoration practices. First, AI can be used for upscaling, deinterlacing, motion interpolation, and more. These are highly effective ways to reduce artifacting and noise, significantly up-res video, and inject crisp detail, all within AI models that were trained on large video data sets. AI can also enhance, repair, and add color to black and white video. It can also be an effective way to add or remove elements within the video frame. And speaking of changing elements within a video frame, another powerful form of video-based AI is deepfake technology. Deepfake, which is a combination of the phrases deep learning and fake, is synthetic media where a person in a video is replaced with someone else's likeness. It's not simply a face or voice swap. It's a highly complex form of artificial intelligence that learns unique visual and oral characteristics about a subject and then maps that onto another subject. This means that people can be made to say something they never actually said. - Our enemies can make it look like anyone is saying anything at any point in time, even if they would never say those things. Moving forward, we need to be more vigilant with what we trust from the internet. It's a time when we need to rely on trusted news sources. - I have an entire course that discusses all aspects of deepfake video, and I encourage you to check that out for a deep dive on this powerful technology. Now not all deepfakes are meant to deceive. There are certainly also practical and beneficial uses to it. Deepfake technology can be used in the entertainment or e-learning industries to let you record a line or performance correction rather than requiring a full reshoot. And as with image-based AI techniques, it can also be used to age or de-age actors. It can also be used to assist in dubbing foreign languages into videos. The possibilities go on, but rest assured there are many ways this technology can be used to enhance processes and workflows. The digital video avatar industry is also a huge part of video-based AI. You can create realistic looking digital avatars that you can make from a photo and either a voice recording or text. This can be a low-budget way of creating a basic explainer video or other marketing content. - Hi. I'm Peter. I'm joining Ashley in this LinkedIn Learning course to let you know that using digital avatars can help personalize your customer experience. - In sales and advertising, video AI can be used for customer personalization. Rephrase.ai creates digital avatars that address people by name during the customer experience. Take a look at this example of a company who created a personalized message of their Beco brand ambassador, speaking directly to a customer named Shavam who abandoned his digital cart. - Hi, Shavam. I'm sure you love our Beco products, but I noticed that you added some products to your cart but didn't buy them. I totally get it. Sometimes you're about to click buy, and you remember something and leave, but let me know if you need any assistance with the product information or at the checkout. Take care. - Text to video generative AI is also possible beyond basic talking head footage. We are at the cusp of this type of technology being used to create more intricate video subjects. You'll simply be able to type in a text prompt, and the AI model creates the corresponding video. This type of tech has so many exciting possibilities, including text to video, text and image to video, image to video, stylization, and more. The list goes on, and video-based AI continues to evolve at rapid speed, but certainly, this type of tech can continue to empower creators and enhance workflows in many different creative industries.