Generative AI Artist

Automatik Vfx; Berlin; Vor Ort; EUR 50.000 - 70.000; Vollzeit

Zusammenfassung

Automatik Vfx in Berlin is seeking a skilled Generative AI Artist to create and deliver high-quality AI-assisted visual work in a professional VFX environment. Candidates should have strong experience with ComfyUI and node-based workflows, alongside a robust knowledge of local image and video models.

Experience with tools like Photoshop and Nuke is preferred. The role involves refining models, documenting workflows, and maintaining high production standards. Strong visual taste and familiarity with AI models are essential for success.

Qualifikationen

• Strong hands-on experience with ComfyUI and building workflows.
• Knowledge of modern control and reference methods.
• Experience with AI model training and LoRAs.

Aufgaben

• Create high-quality AI-assisted visual work.
• Troubleshoot and refine AI models and workflows.
• Document workflows and maintain reproducibility.

Kenntnisse

ComfyUI

Node-based diffusion workflows

Local image and video models

Closed-source commercial tools

Visual taste and cinematic composition

Tools

Photoshop

Nuke

Ostris AI Toolkit

Jobbeschreibung

We are looking for Generative AI Artists who can create, control, refine, troubleshoot, and deliver high-quality AI-assisted visual work in a professional VFX production environment.

Candidates should have strong hands-on experience with ComfyUI and be comfortable building workflows from scratch rather than only using downloaded templates. You should be comfortable with node-based diffusion workflows, model loading, conditioning, masking, latent workflows, samplers, schedulers, denoise strength, seeds, upscaling, inpainting, outpainting, image-to-image, image-to-video, and multi-pass refinement.

Strong knowledge of local image and video models is essential. Relevant model families and tools include FLUX.2 / FLUX Klein, Qwen Image / Qwen Image Edit, Z-Image, Ideogram 4, WAN2.2, WAN Animate, LTX2.3, SCAIL and comparable open or locally deployable models and workflows. Candidates should be interested in actively researching, testing, and evaluating newly released models, custom nodes, and community-provided workflows, then adapting the most useful approaches for professional production use. Experience with closed-source commercial tools such as Runway, Luma, Kling, Veo, Midjourney, or API-based tools is a plus, but our main focus is controllable, reproducible local workflows.

Candidates should understand modern control and reference methods, including ControlNet, depth, pose, Canny/lineart, segmentation, reference conditioning, IP-Adapter-style workflows, regional prompting, masks, LoRAs, start-frame / end-frame workflows, and methods for maintaining character, environment, wardrobe, style, and shot continuity.

Candidates should also be able to use LLMs such as Gemma, GPT/ChatGPT, Claude, or comparable local and hosted language models to structure, enhance, and adapt prompts for different image and video models.

Experience with Ostris AI Toolkit or comparable LoRA training workflows is a plus, especially dataset preparation, captioning, validation, model testing, and bringing self-trained LoRAs back into ComfyUI production workflows.

We value visual taste, cinematic composition, lighting, camera language, anatomy, texture, realism, continuity, and the ability to finish AI outputs using tools such as Photoshop, Nuke, or similar. VFX experience is a plus, especially compositing, cleanup, plate integration, grain, lens effects, color matching, and shot-based iteration.
All candidates should be able to explain their process clearly, document workflows, work with feedback from supervisors and artists, keep outputs versioned and reproducible, and understand that production work requires control, consistency, and finish.