A little over a week ago, Martin Haerlin created a video using RunwayML‘s Gen-1 Video to Video tool that went viral. While I had played with Gen-2 after it was released to the public recently, I wasn’t super happy with the Text to Video output…but I knew the tools had potential to create something more coherent and useful. I decided I’d take a crack at a similar technique to the viral video, but with travel in mind.
The Transformative Power Of Travel
Here’s the finished product:
If you’re interested in how I did it, you’ll find step by step instructions below.
Tutorial: How I Created The Video
The video took a couple of hours of work but the overall process was remarkably simple and straightforward.
Step 1: The Script
This has got to be the most important step…starting with a clear vision helps tremendously. I worked with ChatGPT on this step, outlining exactly what I wanted to create and filling in the blanks during the co-creation process, till I felt I had something I really wanted. The main tip here would be to lay out a framework the LLM can use and then refining from there.
Step 2: The Visuals
I didn’t have time to film a brand new video, so I dug through my photo archives to find a video where I’d done a selfie with some nice natural background scenery. I used this as my base video. I also worked through the settings in my script to create new images I could use to inspire the video-to-video model using Midjourney.
The next step was to use RunwayML’s Gen-1 video to video tool to transform my own video using the themed images I created with Midjourney.
This step does take a bit of experimentation and time, plus you need to be using a paid version of the tool to create video output that is long enough, isn’t watermarked and can be upscaled.
Step 3: The Audio
Enter Elevenlabs…I used a deep American male voice to do the narration for the video – this step was probably the most straightforward.
For the background audio, I simply downloaded a track I liked from YouTube Studio (something that could be used for any purposes, without attribution).
Step 4: Bringing it all together
Once I had all the raw ingredients ready to go, I simply put them together on my favourite video editor, Final Cut Pro.
There you have it…easy as 1, 2, 3. Provided you have the inclination and can start off with a solid idea + script, the rest is just grunt work. The video to video conversion output from Runway Gen-1 is decent, but I suspect with future versions it will likely get a LOT better and more realistic (plus hopefully high resolution).
What do you think? Anything I could have done differently or better? Have you spotted any other takes on the original viral video? If so, please share or leave feedback via a comment or DM to let me know!
0 comments on “Transformative Travel – A Travel Take on the Viral Runway Gen1 Video”