Yevhenii P.

I make finished AI video ads and films, and I build the ComfyUI pipelines that keep them on model: same product, same character, shot after shot. What I deliver • Product and brand ads (vertical and 16:9). Recent example: a 12-second promo for a nail salon in Poland, styled as its landing page. Six of the salon's real manicure designs became 3D "products". MiniMax H3 generated the page, and I animated the cursor in Remotion so every click lands exactly on the scene change. • 3D→AI cinematics. Unreal Engine 5 or Blender supplies the geometry, camera and motion; AI supplies the photoreal look. I led a 60-second promo for a defence-industry client this way (UE5 → Nano Banana Pro → Seedance → DaVinci) and made a full-AI defence film solo, from 3D through to sound. • Consistent characters. Custom LoRA training (ai-toolkit) plus reference-driven generation, so one face holds across stills, carousels and video. • Fan and concept cinematics, product previs and look tests (real estate, superyachts). How I work • ComfyUI graphs that I build, debug and optimise myself: local MiniMax H3 (Ref2VA, sound included), Wan, LTX, upscaling and quality passes; cloud GPUs (RunPod / Vast.ai / RunningHub) when a job needs scale. • Structured prompts with timed shots, so the model follows the edit instead of improvising. • Post that makes the output feel finished: editing, grading and sound in DaVinci Resolve; Remotion for exact UI/motion overlays; audio clean-up for generated sound (resampling, rebuilt top end, loudness to platform spec). • Automation when it pays off: Python, API-driven batch generation, Telegram bots. What you get — Output that reads as shot, not "plastic AI". — Consistency across shots and revisions, plus an honest view of what AI can and cannot hold. — End-to-end delivery: concept → generation → animation → edit → sound → final file. Tools: ComfyUI, MiniMax H3, Seedance, Wan, Veo, Kling, Nano Banana Pro, GPT Image, Flux, Krea, Seedream · Unreal Engine 5, Blender · DaVinci Resolve, Remotion, ffmpeg · Python. Background: BA in 3D (Vistula University, Warsaw) and motion-design training. Languages: Ukrainian, Russian, Polish, English. If AI is not the right tool for your job, I will tell you before we start.