Key takeaways
- Artificial Intelligence is a transformative technology, significantly impacting various fields, including video production.
- AI is being utilised to create high-quality videos that rival real-world visuals, enhancing the creative process.
- Notable AI-generated videos in 2019 include a Nike commercial crafted by a neural network trained on previous ads.
- Google's Video Inbetweening technology exemplifies AI's capability to generate intermediate frames, creating coherent video sequences from just two frames.
Artificial Intelligence is ‘the’ technology of the 21st century, and as if one doesn’t already know, it is everywhere. From simple speech-to-text service, to converting a given text into speech with your own voice, AI has come a long way and now has many more applications than you expected. With more and more applications and devices integrating with AI, it has made its way to the creative fields as well. And in this particular case, we are talking about video creation.
Artificial intelligence is now also helping video production companies and the gaming industry to create high-quality videos and enhance the quality which can be compared to the real world.
Here are some of the videos in 2019 that were created by AI. We have also included videos where AI was used extensively while creating:
Legend That Thing
When you look at this video, the sportsman inside you might feel motivated. But when you think about the fact that this Nike commercial was written by a neural network trained on seven years of Nike ad videos, the AI enthusiast inside you might want to hit that play button again.
Google’s AI creates videos
In 2019, Google introduced what they called Video Inbetweening as a possible way to tackle the problems of generating plausible and diverse video sequences. Google’s AI was given only the start, and end frame of a video and the AI was supposed to sketch out the frames in-between them. Might look impossible to you, but Google’s AI research program has devised a way to create ‘plausible’ Video sequences from only two frames, the first and the last.
First AI Interactive AI Rendered World
This is an AI breakthrough by researches at NVIDIA was the first time where machine learning and computer graphics were combined to do image generation using deep networks. Meaning they trained the AI model to draw a new world based on the videos fed to it from the real world. The breakthrough will allow the developers and artists to new interactive virtual worlds for gaming, virtual reality, videos and automotive.
Project Sol
Project Sol is a display for the NVIDIA RTX development platform which makes use of ray-tracing in cinematic scenes which run on real-time. NVIDIA used ray-tracing on Unreal Engine 4. This is part 2 of the project.
AI doesn’t create the whole video but, the real-time ray tracing here applies AI and specialised processor cores to render highly realistic graphics in real-time.
AI writes Scripts for videos:
Along with creating the videos, AI has helped the creative field to generate scripts by training the model on different writing materials as data sets.
This Short Film Is Written Entirely By AI
Austin McConnell says he plugged in several well-known stories along with his own book into an AI open source program. The machine used these data inputs to learn these stories how to write, and the system was asked to produce its own script. The results using animation software are as shown in the video:
An AI Made This Video
Strictly speaking, the video wasn’t made by AI, but Austin Evans here is reading is text generated by an AI modern neural network called Transformer that completes your writing when you type in a custom snippet. Adam King built this Transformer with the OpenAI’s new machine learning model called GPT-2.
