Tegobe

• GIF

  TEGOBE

Inspiration came from blending my "signature" MLIBTY Shadows series with Croatian and Chinese traditional music, using AI.

MLIBTY Shadows I started MLIBTY Shadows sculpture series at the beginning of 2013, after I had had this idea in my head for some time. Well, the whole story goes even further back in the past, almost ten years before that. In 2004, during my studies at the Facultad de Bellas Artes de Barcelona, I started working on a small series of works, combining photoemulsion and etching. Due to the complexity of the process and a lot of experimenting, the series was finished in 2005. One of the works was a little bit different. It's called Vecinos (The neighbors). Living in such a big and populated city like Barcelona created a feeling that there was me on one side and "them", all the others I don't know or I don't see, on the other. In this work, they are neighbors posing in front of the building I was living in. This is the work that started it all, the root of MLIBTY Shadows. The shadows behind the curtains, invisible inhabitants you never meet, but you feel their presence. I continued with a series of small paintings, combining photography, painting, digital art and AI. All the photos are mine, taken during my travels, skate trips, or in my environment, and I intervene on them with acrylic and ink. Characters are always in black & white, they have no hands and no legs, they are shadows with faces.

What it does

This music video defies the classic concept of music videos, where dynamics seem more important than the art itself, especially in this AI hype, where it's extremely easy to create a large number of shots and different scenes. It's static, apparently nothing happens, which is the philosophy behind the whole MLIBTY Shadows concept, where almost nothing happens, or very little happens, but the atmosphere is threatening.

How we built it

I built it with my own model trained in Alias, exclusively on my works from MLIBTY Shadows series. I also did the music using Pinokio and Suno and mixed in Audacity. The whole project was then edited in Davinci Resolve

Challenges we ran into

The challenge was to make a static AI video that looks like one take and runs in a loop, so it's not all just AI, I did some manual animation and editing. Also, I mixed the various fragments of music and vocals that were generated separately, using different AIs.

Accomplishments that we're proud of

This is the first full Mlibty Shadows music video, after almost thirteen years of the existence of this project, where I managed to blend traditional Chinese sounds with Croatian folk music, with the mood of early hours drunkenness in a Balkan bar, but without aggressiveness. The video is tokenized on SuperRare.

What we learned

Up to date, I have used all kinds of platforms, trained my own models locally and online, but AI still hasn't passed MLIBTY Shadows test (a test that I invented), which is not adding ears, arms, and legs to the MLIBTY Shadows in their generation. Alias has the best score, but still getting an original MLIBTY Shadows feels like a happy error. Negative prompts, in all cases, only motivated the models to add what I wanted to exclude. It's easier to create complicated monsters without ears and extremities than a simple entity that just has the body and the head (without ears).

What's next for Tegobe

I already have the next song ready, slightly in the Tegobe style, but this time I also added a bit of Spanish and Bosnian influence, so the next step is to decide whether to continue with the MLIBTY Shadows visualization or do something totally kitch.

Special note

DWYDLYLA is my alias for music projects

Built With

• alias
• pinokio
• suno