Gavr Garbel's Work | ContraWork by Gavr Garbel

Gavr Garbel

Senior AI Video Artist | Cinematic Consistency & QA Expert

New to Contra

Gavr is building their profile!

Aachen, Germany

Work Posts Services About

Aachen, Germany

Bridging the Uncanny Valley The biggest fear for brands using AI is the "melting" look. My mission is to kill that artifact-heavy aesthetic. This reel is a demonstration of how a structured pipeline can force any tool—even a budget-friendly one—to output premium-tier content. It’s about knowing the laws of physics and how to "break" them correctly inside the latent space of the neural network. No hallucinations. No floating limbs. Just pure, calculated visual power.

Cover image for The Goal:
To create a high-end

The Goal: To create a high-end visual concept for Clay that personifies their core philosophy: "Order from Chaos." The Execution: I designed a 3D sequence focusing on the transformation of raw data (abstract cubes) into a structured, laser-precise GTM engine. The project emphasizes premium lighting, fluid physics, and surgical precision to align with Clay’s brand identity. Tools Used: AI-Driven Visuals CapCut (Editing & Sound Design) HandBrake (Final Master Optimization) Services Provided: Concept Development 3D Motion Graphics Sound Design & Post-Production.

AI video is tired of being weightless. 🧊 Most generations feel like a fading dream—no mass, no gravity. I’m testing a way to fix this by locking the frame and mastering the Delta between texture and light. When you kill the "AI float," the viewer finally starts to feel the object. Is it a render or does it have weight? You tell me. 👇

Cover image for Why 99% of "Prompt Engineers"

Why 99% of "Prompt Engineers" are Amateurs (And Why This Profession Doesn't Exist) 1. The Illusion of Control People treat prompts like magic spells. In reality, a prompt is a primitive attempt to describe the multidimensional mathematics of 3D space using basic words. If you don't understand the physics of light, the laws of optics, and the latent space of the model — you aren't an engineer. You’re a fortune teller. 2. Why AI Writes Prompts Better Than Humans Eliminating logical contradictions and generation errors requires competencies that humans simply don't possess: Total grasp of 3D frame physics (The AI is natively trained on this). Instant detection of logical conflicts (The AI sees this in high-dimensional vectors). The model's native language. I don't write prompts. I build autonomous LLM-based systems that act as the engineers themselves, because they understand AI physics and logic from the inside out. 3. The End of the "Spellcaster" Era The future doesn't belong to those who pick synonyms; it belongs to those who build technical pipelines. My method is a rigid structure — a "Brick" — that transforms chaos into a predictable production process. Bottom line: Stop calling people with a dictionary "engineers." We need system architects. Everything else is just cargo culting. Tired of the neural network lottery? Stop hiring people who just look for synonyms for the word "masterpiece." I build autonomous AI technical control pipelines. My system (see the example in the image) analyzes frames at the level of 3D physics, optics, and anatomy. Want stable results for your video production? DM me. Left: Amateur "Spellcasting" (🔴 Red Marker) Center: AI Production Feed / Technical Brief (🟡 Yellow Marker) Right: Autonomous AI Fixer Analysis & QA Report (🟢 Green Marker)

129