2YK/ High Fashion photoshoot Prompts for Z-Image Base (default template, no loras)

An AI image generation practitioner known as BerlinBaer has open-sourced a practical toolkit for the community: a curated collection of 40 high-fashion photography prompts specifically optimized for the Z-Image Base model. The prompts were created by feeding Pinterest images into a QwenVL multimodal AI to generate descriptive text, which was then cleaned and polished. The primary purpose is not for creating a single perfect image, but for systematic batch testing—allowing users to generate 40 different scenarios to evaluate how a new LoRA (Low-Rank Adaptation) or workflow performs across diverse lighting, poses, and compositions. The prompts are intentionally designed to be neutral regarding gender and race, instead using a dynamic prompting system that randomly selects these attributes, ensuring broad usability and testing consistency.

The technical specifics reveal a focus on cinematic vibes, extreme camera angles (like fisheye or low-angle shots), and challenging colored lighting setups. BerlinBaer notes that these prompts were tested against other models like ZIT and Klein 4B, but Z-Image Base consistently delivered superior results in maintaining the intended dramatic lighting and dynamic poses. A standard negative prompt is provided to avoid cartoonish or low-quality artifacts. This release is a significant contribution to the practical, iterative side of AI image generation, providing a standardized test suite that moves beyond single-prompt artistry to methodical model and tool evaluation. It empowers users to quickly stress-test their customizations in a controlled, reproducible way.