Anton L.

I direct and produce series and films through AI pipelines - episodic vertical drama, feature-film sequences, short films, music videos and branded films - and deliver them finished: scripted, generated, composited, graded, mixed, mastered. I've been doing this since generative video could barely hold a face together - early 2023, the Will Smith spaghetti era. Before AI, I directed, shot and edited commercials and branded films across Europe (Germany, Belgium, Poland, Italy, France, the Netherlands, Latvia, Georgia, Croatia) and on US productions. That's the part that matters: I work as a director, not a prompt operator. Treatment, shot list, blocking, performance, cut. WHAT I DELIVER Episodic series - pilots and full seasons for vertical-drama platforms and streaming; consistent characters, world and tone across dozens of episodes. Cartoons (anime, 3D, and 2D styles; the primary goal is to create something unusual and unique rather than replicating an existing style). Feature and short films - full scenes and sequences rebuilt or originated in generative pipelines for independent features Music videos and artist films Branded films and commercials Rebuilds - hand me a shot list, script or animatic; I return it at cinema quality HOW I WORK Fixed price per project, quoted per finished minute, with milestones and revision rounds agreed before generation starts. I own the pipeline end to end: script - boards - character and world locks - generation - compositing and VFX - edit, grade, sound - masters in any format (16:9, 9:16, 1:1 etc.) Locked scripts stay locked; if a line trips a model filter, I flag it - I don't rewrite it. PIPELINE (current – ​​constantly updated) • Video generation: Seedance 2.5 / 2.0, Kling 3.0, Veo 3.1, Wan 2.6 / 2.7 / 3.0, LTX-2.3, Flux 3, Hunyuan Video, Runway Gen-4.5, Hailuo - MiniMax, Luma Ray 3, Vidu - most of it orchestrated inside ComfyUI on my own hardware, with custom workflows for character lock, style lock and batch runs. • Image and look development: GPT-2 /2.5, Midjourney, Flux, Qwen-Image, Z-Image, Nano Banana Pro / 2, Seedream 5, Stable Diffusion; LoRA training for characters and looks. 3D, VFX, virtual production: Blender, Unreal Engine 5, RealityCapture photogrammetry, Gaea, Quill VR. When generation alone won't hold a set, I build the set in 3D and generate on top of it. • Post: DaVinci Resolve (edit, color, Fusion), Photoshop / Illustrator / Lightroom. Voice, lip-sync, upscaling and cleanup handled in-pipeline. • Agents: I use Claude (Fable 5.1, Claude Code) and GPT-6 Astra to script and automate the repetitive layers - Blender scene assembly, Resolve scripting, ComfyUI batch runs, shot logs. I'm not waiting to see whether agents change this work; I'm building with them. I test every major model release the week it ships and move the pipeline only when a model earns its place. WHY CLIENTS COME BACK One person from treatment to master - no handoffs, nothing lost between concept and frame Series-grade consistency - the same faces, sets and light in episode 40 as in episode 1 Real production discipline - milestones, revision rounds and delivery specs set up front Fixed pricing per minute / per second - of finished video; you know the total cost before the project starts