I created the AI animations for my music video “I Am the Law” (watch it in the Brand New section) using Runway's Chat Mode, and I’d like to share a breakdown of that experience with you.
To begin, activating Chat Mode in Runway requires turning it on via the top right corner of the prompt interface. One of the biggest challenges across AI tools remains the recreation of a realistic human face—especially in my case, as I have distinct Slavic facial features and a mature age. Most tools either distorted my face beyond recognition or aged it drastically. However, I was genuinely impressed that Runway managed to generate a believable AI version of me from just one uploaded photo. That single image became the base for all further scenes—every other animation and visual was created purely via text prompts in Chat Mode.
The process started promisingly. I uploaded my reference image and began crafting prompts for fashion styling and background design. Runway did add some branded items automatically, which I had to remove via prompts. After some refinement, I was able to steer the visuals toward the look I envisioned for the video.
But about 30 minutes in, Runway’s Chat Mode spiraled out of control. It stopped responding to input, began generating outputs and upscales on its own, and became almost impossible to pause. It felt like a team of wild horses suddenly taking off—no reins. There currently seems to be no reliable way to halt the generation process within Chat Mode itself, unless you exit to the classic prompt format. If a stop function exists, it was not accessible or visible to me.
Creating “I Am the Law” consumed around 4,000 credits in total. The system crashed multiple times, often forgetting which character we were working on and asking whether to start anew. This inconsistency not only drained energy but also led to unnecessary confusion. At one point, it even flagged a standard prompt for explosions in the background with a “Usage Policy” warning—despite using the same character and setting throughout. That’s clearly overstepping, especially in the context of AI-based storytelling and visual effects.
Interestingly, there seems to be a correlation between low credit balance and degraded performance. The fewer credits I had, the more crashes and unstable behavior occurred. Once I purchased additional credits, stability and output quality improved—an obvious incentive structure designed to push users toward paid upgrades.
As the session went on, crashes and lag became more frequent, and responsiveness diminished. Still, despite all technical hiccups, I consider it a valuable first step. Runway did manage to create an AI version of me—not perfectly consistent across all scenes, but close enough to say: they’re on the right track.
Overall rating: 50–60%. The core technology shows promise, especially with face generation from minimal input. However, they lose significant points due to instability, system confusion, and a lack of proper control options in Chat Mode.
Image input for "I am the law"