top of page
Digital Media Productions
UKAHT Project: Who are your audience? (continued research + video planning)
Research into AI:
This week I have decided to research a lot more about AI image and video creation, with the help of Tina; I was able to recieve access to the runway AI university account that had credits on it to enable me to not only test out but also create videos for my project.
First, I had to really look at how Runway AI and their video creation tool works. At the same time I was also able to sign up to Google's AI tool Gemini for a year due to being a student which will help me as I will also have access to Google Veo 3, which is a new Video AI. Runway and Gemini will help me to complete my tasks.
Researching I was able to come across two very good videos that helped me understand the AI a lot more. Below I have attached the two videos that helped me understand how they work more.
This video taught me how to train the AI model and explained RUNWAY - Act 2 a lot more in detail.
The second video helped me understand a lot more about the new features and things I can do or use to help my videos look better.
AI trials & work:
Learning all this new information, I also realized that I will have to create my own videos of myself to then have AI learn or replicate using my video and the prompt for Runway to properly understand what I want it to create.
At first I wanted to test out a few prompts and I wanted Runway to create a video of a granny sitting on her armchair drinking her tea, all in claymation style to give it a vintage and a "Christmas" vibe, told AI to put a christmas tree and decorations in the background. Unfortunately it failed massively. First, I did not even realize I actually needed to generate the image then use the image as reference so that was bad, but after creating a good image, when given the prompt to create the video/animation, it did horribly wrong.
Below I have attached the initial picture created for the animation.
After it created the image I decided to record myself doing a few basic actions which were supposed to create the animation. I have added a couple of the videos below
After creating such failures, I decided to keep the idea of a old lady drinking tea but make it realistic which I have decided to use on Gemini using Veo 3.
Below I have attached the videos Veo 3 created, which I actually ended up using for the final advertisement:
After recording them I have put them on Runway and inserted a "Character Model" that being the old lady and I have told it to just follow my motion.
Unfortunately, every single time I have tried it did worse and worse and I was not only using credits on all these failed tries but also time and patience. It seemed like nothing I wrote or told it to do helped and only made it worse. Unfortunately I did not end up saving/downloading them as I was so ashamed and upset with the way the ended up looking that I felt there was no need for them and I unfortunately showcase them. Trust me, they were bad
I have then edited the clips together to make sure they have smooth transitions in between them and have a good end screen with a Call to action that will have viewers actually go and look at the website and buy the product instead of skipping past.
Although the videos are not the best, I unfortunately ran out of credits and time and they were my only option and I had to work with them.
bottom of page