Stella Particula

Description

I remember Damascus on a snowy day. The narrator navigates a storm of emotions and memories, torn between extremes, heat and cold, chaos and calm. A snowy day in Damascus becomes a fleeting moment of peace amid inner turmoil. Through poetic fragments, he searches for balance in a world of constant motion and shifting identity.

Process

I was born in France, from a French father and a Costa Rican mother, but as soon as I got 2 months, my parents went in Syria for work. We lived in Damascus. I lived there for 6 years : and from those years I barely remember some images one of which is this snowy morning in our small garden, covered in snow. Some years ago, I went through a severe depression that lasted 7 years and one day, I remembered this image, calm and peaceful. I was going through lots of emotions, trying to find a balance between happiness and sadness, kind of a mental stability, equilibrium. This life abroad, this image of our garden in snow in Damascus (where you don't usually expect snow) became the symbol of the text I wrote. A text that represent the torments and inner reflexions of the narrator trying to find balance. Something that is blurry, with lots of contradictions but at the same time clear : I was thinking that everything makes sense if you link the pieces of your life. That finally, the scenario is not written, but the paths you took in life were steps that are always going forward. The struggle to find balance is something inherent to all human beings, and for artists sometimes it's something even more complicated. Juggling between our creations, the things we want to share and the vision that the viewer has of what we make. AI was very important here to express those generated and distorted memories and to have this coherence between concept and image.

Tools

Full AI video - with some real photographies (as inputs). Voice-over recorded at home. Music made on Ableton Live, with VST and modular gear. One cut is made with 3D and analog glitch gear. AI tools to clean the voice-over. AI tools for slow-motion and upscaling. Workflow: I first wrote the text and then made the music matching the mood I was going for. The voice-over was recorded very soon in the process, to have a guide for the video production phase. A moodboard was then created with reference images, and then an early storyboard was sketched. Then, on my computer, using SDXL with Loras (Fooocus as UI), a large number of reference images were generated to visually interpret different parts of the text. Experimental renders were created locally in ComfyUI using different models, prompts, and occasionally input images to explore hallucinated and dreamlike aesthetics. This stage served to feed the initial moodboard and the storyboard. Runway was used to generate most of the final AI-driven video content (with SDXL generated images or Frames generated images as inputs). It was also usefull to refine certain sketches done in ComfyUI. Real photographs were woven into the process, used as bases for generation or integrated directly. The final production process was made with the adobe suite. Colors were changed to match the concept and message of the different sequences. Lots of editing was done.