Daniel Isnard

I’m a detail-oriented AI visual evaluator with a film and post-production foundation and hands-on experience building generative image and video workflows. I apply structured visual QA methodology—covering instruction adherence, realism, anatomy, lighting consistency, artifact detection, and region preservation—to assess AI-edited and AI-generated imagery against production standards.

Backed by nearly two decades of luxury quality-control experience, I spot subtle failures in visual credibility and deliver clear, actionable, evidence-based feedback for AI training and model improvement teams. I evaluate edit accuracy, reference consistency, and production usability against project-specific rubrics, while documenting failure modes like warped anatomy, synthetic skin, distorted geometry, and hallucinated text.

I also bring creative fluency from high-fashion and editorial sensibilities, building and pressure-testing multi-model workflows across platforms such as Google Flow and Higgsfield. My experience spans text-to-image, image-to-image, text-to-video, and image-to-video, with practical control over prompt iteration, character consistency, and start/end-frame delivery.