After early AI classes, a fashion creative built a full campaign image for

Project Description For DELADA SS25, I created an AI-generated editorial campaign image featuring a model in a full DELADA look. I designed the concept, prompted the model, styling, mood, setting, and composition, then iterated across multiple AI models and post-production workflows to achieve a cohesive, fashion-editorial result. The final image is intended as a campaign visual that reflects DELADA’s distinctive aesthetic: modern, slightly cinematic, and styled with a controlled, intentional color palette. I generated this image a few months ago after our first classes with Domenic, at a time when AI image models were not as advanced as they are today. While I could regenerate the model now with higher technical realism, I intentionally chose not to “perfect” it. The image’s tension—porcelain, polished beauty against raw, worn elements (like the rough boots and industrial setting)—captures the refined-punk contrast that is central to DELADA’s identity. This balance of control and abrasion felt creatively aligned with the brand, so I preserved it as part of the final campaign expression. Achievements - Produced a campaign-ready editorial image that reads as a coherent fashion photograph (styling, pose, lighting, and environment aligned). - Successfully translated a clear creative direction (vibe, silhouette, attitude, setting) into prompt language and repeatable iterations. - Achieved brand consistency (tone, mood, typography placement, and overall visual identity) suitable for SS25 campaign use. - Combined multiple AI outputs and manual post-production into a single refined final composition. Tools - AI image generation: Higgsfield; “Nana Banana” model/workflow (used for variations and refinement). - Post-production & compositing: Adobe Photoshop (cleanup, compositing, retouching, typography, final color work). Methodology Creative direction & reference - Defined the campaign intent (editorial mood, styling attitude, color palette, environment). - Specified key requirements: full outfit visibility, fashion pose, coherent lighting, and a cinematic editorial finish. Prompt design - Built prompts covering: model description, wardrobe styling details, editorial mood, camera/lighting cues, composition, and background environment. - Included constraints to keep the image “campaign clean” (avoid unwanted props, distortions, inconsistent fabrics, or off-brand styling). Iterative generation Ran multiple generations across different models to explore variations in: - model look + pose - garment silhouette accuracy - lighting consistency - background alignment and editorial realism - Selected the strongest base image and generated supporting variations where needed. Refinement & post-production - Composited and refined details in Photoshop - Unified the final look with color grading (tone, contrast, highlight control) to ensure a consistent editorial finish. - Added campaign typography (“DELADA SS25”) with final layout balancing for a clean campaign presentation. Final delivery - Exported in high resolution for digital campaign use, with a composition that can be adapted for web, social, and lookbook formats. Learnings - Strong results come from specific creative constraints (lighting direction, lens feel, pose intent, garment priorities) rather than overly long prompts. - Mixing models is effective: one model can excel at mood/lighting while another performs better on silhouette and garment fidelity. - AI outputs often need a professional finishing pass—retouching and color unification are what make it feel campaign-ready. - Keeping a prompt log and versioning outputs dramatically improves repeatability for a consistent brand campaign series. Relevant URLs - Brand website: www.delada.net - Instagram @delada_