Complex Reference Prompting

Merging visual inputs with a detailed text prompt

Complex Reference Prompting media is blocked

Allow external media to connect to the provider and play this content.

Description

The user combines the two uploaded images into a single complex prompt that specifies character consistency, pose, and environmental details. The prompt focuses on maintaining identity while adapting the subject to a new perspective.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: Highlights a sophisticated prompting technique that references specific visual inputs for high-fidelity merging.
Built with: Mercury 2.5 GPT Image 2.5 Flare Image 2.0 Muse Image Seedance 2.5 MAI Image 2.5 HiDream O1 Image GPT Image 2 GPT Images 2.0 Qwen Image 2512 GPT Image 1.5 Wan 2.5
Tags: #gpt-image-2-5-on-dzine-flare-vs-sunburst-which-o #experiment

Discovered: YouTube
First posted: YouTube · Sep 12, 2026
Listed: Added 3d ago

Prompt

composition, environment, background, lighting direction, shadows, color palette, atmosphere, depth of field, visual style, and level of detail. Adapt the character from [ref_1] naturally to the pose and perspective of [ref_2] without changing their identity. Ensure anatomically correct hands, limbs, facial structure, object interactions, lighting, contact shadows, and perspective. The final result should look like the original [ref_1] was created with the character from [ref_2] from the beginning, with seamless integration and consistent lighting, proportions, and visual style.

Source

All projects

Apps & Tools