Intro to Runway

In this blog post I am testing out three different models, provided in RunwayML.

Model 1 — Anime Filter

I started out with this model because it seemed fun. I also tried to find the photo of my face since I was curious whether the model would be able to portray my face in the anime style. It turned out quite odd.

Model 2 — Art Line

The second model I decided to try called ArtLine. Basically, what it does is creates line art portraits out of images. As a digital artist myself I could not ignore this. Most of my art works are quite rendered so I was curious what kind of line work the model would generate out of them. The result was quite fascinating! Maybe I should use it for my future artworks?

Model 3 — Text Generator

For this example, I used the GPT-2 model, I decided to feed one of the poems I wrote back in my undergrad when I was heartbroken (oof). This is what it generated. Although, only two words here are actually existing words in Russian, I should note the ones that actually mean something are: человек — human and миловение — mercy. So interesting the model learned exactly those two words out of multiple lines of words.

Бли то боничных не резз гово человек,
ичю влинта не поцежения.
Короданд сомудог митовартить недно
Миловение