I recently had a chance to experiment with Colossyan.com, a tool that lets you create AI avatar videos from text scripts—and even in different languages. To test it, I used the Italian translation of a script on the Lense-Thirring effect and created a short video of “myself” speaking fluent Italian.
My avatar was generated from a simple 15-second scan using my webcam. Interestingly, this was my third attempt, and it’s the one I used in the first scene of the video (wearing glasses). I paired it with the voice generated from my very first try—a 1-minute voice recording. The result was surprisingly decent, considering the simplicity of the input.
The second scene shows my first try, where I made a couple of mistakes that taught me a lot about how Colossyan handles scanning. I did the recommended 1-minute face scan, but I moved around too much and accidentally touched my face. The avatar ended up mimicking these actions in an almost too-realistic way—at one point, the avatar touches its own face, reflecting my fidgeting during the scan. This raised a question for me: Why does the app encourage longer scans and ask you to focus just on your face if it’s capturing so much of your body language?
In the third scene, which was my second attempt, I smiled mid-way through the scan because I thought I looked too serious. Unfortunately, the avatar decided to “freeze” that smile on me throughout the video, giving me what I can only describe as a creepy, fixed grin.
Overall, my takeaway is that Colossyan isn’t necessarily revolutionary compared to other avatar tools I’ve tested, but it’s an incremental improvement in realism—especially in the face rendering and voice sync. I’ll continue exploring it, especially its features for building interactive learning modules, which look promising. I’m curious to see if it can move beyond basic talking head videos and support richer educational experiences.
If you’re considering testing AI avatar tools for multilingual videos or e-learning, Colossyan is worth a look—but I’d advise experimenting carefully with your scans to avoid the “creepy smile” effect.
It’s scary! I can see the use of AI for translation purposes, but it can be very dangerous if used for other causes…