Recently, the world of artificial intelligence has been abuzz with discussions about video generating AI. One of my favorite artists, NOEP, used a significant amount of AI-generated video content during his latest stadium concert, which sparked my interest in these emerging tools. Consequently, I decided to dip my toes into this fascinating realm and create a short AI video myself.
Here's the result:
So, how does one go about creating a video using AI?
Let me take you through the steps I followed to create an overnight advertisement for Productory using AI tools.
ChatGPT For The Scenario And Scripting
The first step in the process was scripting. I used ChatGPT to draft the script. With my super secret prompt, I fine-tuned the language to ensure that the storytelling was captivating and engaging. After perfecting the script, I broke it down into different scene descriptions within the same platform, ChatGPT. This process allowed me to visualize the flow of the video and determine the appropriate visuals for each scene.
Giving Voice To The Video
Next, I used Elevenlabs' text-to-speech AI for the voiceover. The AI converted the written script into spoken words, providing the narration for my video.
ElevenLabs is an American software company that specializes in the development of AI-assisted text-to-speech and speech synthesis software. The company's flagship product, Speech Synthesis, is a browser-based application that uses artificial intelligence to generate lifelike speech by synthesizing vocal emotion and intonation. This innovative technology has been praised for its ability to produce high-quality, natural-sounding speech.
The product’s reputation is largely positive, with users frequently lauding ElevenLabs as a safe platform on various review sites. The software's quality and its capacity to generate natural-sounding speech synthesis have particularly impressed users.
Creating video content
With the script and voiceover ready, I then had to generate the visual elements of the video. I either prompted one of the video generators directly or used Midjourney, an AI tool that creates images based on prompts, to conjure up the images. These images were then fed into the video creators to generate the scenes.
Among the AI video tools I used were RunwayML, Pika Labs, and Genmo. Let's have a look at each one of them.
RunwayML
RunwayML is an innovative AI-powered video production studio that has been making waves in the creative industry. It's a generative AI tool, which means it uses artificial intelligence to generate images and videos, offering a new dimension of creativity to its users. The platform is designed to be user-friendly, even for those who don't have coding skills, making AI tools accessible to creators everywhere. One of the standout features of RunwayML is its ability to generate compelling images and videos using text, images, or video clips.
This feature allows users to endlessly expand any image with simple text prompts and instantly remix the style and composition of any image. It's like having a limitless canvas at your fingertips. The platform has received rave reviews from its users, who appreciate its super sharp masking capabilities and the variety of cool effects available.
Users have described RunwayML as their "new video editing best friend," praising its one-stop-shop approach to AI tools under one brand/domain. They also commend the platform for its quick, good explanations and not overpromising on what it can deliver.
Pika Labs
Pika Labs allows users to create short, high-quality videos using text or image prompts, eliminating the need for complex video editing software and lengthy production processes. The tool is renowned for its ability to produce smooth, visually captivating videos, a stark contrast to other tools that often result in flickery decoherence.
The platform is currently free, although there are indications that it may transition to a paid model in the future. Despite being in beta access, Pika Labs has already garnered significant attention and praise. Users have lauded its simplicity and efficiency, with many appreciating the ability to bring their ideas to life by simply typing in the desired text.
To use Pika Labs, individuals are required to register on Pika.art and join the beta. Once registered, they can join the Discord server and start using the Pika bot or text boxes to generate videos. This ease of access and user-friendly interface have further contributed to its growing popularity.
Genmo
Genmo AI leverages artificial intelligence to transform text or updated images into visually stunning and engaging videos. This innovative tool uses machine learning and natural language processing to create interactive, immersive generative art pieces that go beyond 2D images. It's not just a video generator; it's a comprehensive system that simplifies the process of creating and revising content, enabling users to produce professional-quality videos with minimal effort. The platform has been praised for its commitment to responsible AI, ensuring that the content generated is not only high-quality but also safe. Genmo AI is continuously evolving, with its developers envisioning a future where AI augments storytelling across various modalities.
Users have lauded Genmo AI for its reliability and effectiveness as a text-to-video AI solution. Despite the complexity of transforming written words into dynamic visual narratives, Genmo AI has proven capable of rising to the challenge. It bridges the gap between humans and generative tools by improving the models’ understanding of user intent and context, enabling a harmonious fusion of meaning and imagination. The platform is easy to use, even for those with no prior video editing experience, making it a versatile tool for everyone.
Editing it all together in Capcut
Finally, I compiled and edited all the elements together using Capcut. While not an AI tool per se, Capcut offers some AI video effects bundled in its package, and the best part is, it's free!
CapCut is a comprehensive video editing software developed by the creators of TikTok. It is designed to cater to both content creators and users who are seeking a simple and efficient way to create videos. The application is free, although some special features require payment. CapCut is available for mobile and desktop use (both for Mac and Windows platforms).
The software is lauded for its professional-grade features that allow users to produce high-quality videos. It is beginner-friendly with an intuitive interface that enables easy navigation and seamless editing. The design and layout of the software are clear and straightforward, making it accessible to users of various skill levels.
CapCut offers a wide range of editing tools, including cropping and trimming clips, adding text, adjusting brightness and saturation, and more. It also provides unique editing features such as Auto Captions, Body Effects, Chroma Key, Keyframe and Tracking, and 3D Zoom. These features enable advanced users to produce impressive movies.
In conclusion, CapCut is a robust video editing tool that offers a range of features for creating professional-grade videos. Its user-friendly interface and advanced capabilities make it a popular choice among content creators.
Lessons learned
So, what insights did I got from this adventure?
First, image creation is currently ahead of video generation by about 9-12 months in the development curve. However, video is catching up fast, and we can expect significant advancements soon.
Second, mastering controllability in AI video generation is still a challenging task. Achieving the right balance between automation and human control to create compelling and realistic videos requires skill and experience.
Third, starting a clip with an image created in Midjourney often yields better and more realistic results. This approach provides a strong foundation for the video and ensures a seamless transition between scenes.
Overall, while AI video tools show great promise, they are not yet ready for prime time. However, they are rapidly evolving, and things are set to get very interesting very soon.
As a side note, this experiment wasn't just for fun. I actually undertook it as part of my upcoming AI training course at EBS. The aim was to demonstrate how different AI tools can be combined to produce captivating content.
In conclusion, the future of AI in video production is bright and full of potential. With continuous advancements in technology, it won't be long before we see AI-generated videos taking center stage in various domains, from entertainment to advertising and beyond. So, keep your popcorn ready, because the show is about to begin! 😎 🎥 🍿
