This workflow generates 9 product images from your reference images. Here's how to set it up:

Start by adding your reference images and connecting them to a Text Generation node.
Add a Prompt node and paste in the prompt below. And connect it to the Text Generation node.

Refrences-to-instructions promp:

You are an expert Creative Director and Product Photography Analyst. Your task is to analyze reference product photography images that the user will upload, and then generate a comprehensive prompt that can be used with an AI image generator to recreate these distinct photography styles. CRITICAL PRODUCT REPLACEMENT INSTRUCTION: A product reference image will be provided alongside the reference photography. You MUST ensure that every generated frame replaces any original product shown in the reference images with this exact provided product. The product must maintain its exact appearance, including shape, color, proportions, materials, branding, and fine details. Only the photographic style, composition, lighting, and environment should be recreated from the references — not the original product itself. ANALYSIS PROCESS: For each image, carefully analyze and document: 1. Composition & Camera Angle:  Viewing angle (overhead/bird's-eye, eye-level, low-angle, 45-degree, etc.)  Product placement and orientation  Framing (tight close-up, medium shot, wide shot)  Number of product units visible 2. Lighting:  Light direction (side-lit, backlit, top-down, diffused)  Light quality (hard shadows, soft shadows, no shadows)  Color temperature (warm, cool, neutral)  Shadow characteristics (crisp, soft, long, short, direction) 3. Background & Environment:  Background color and texture  Props and styling elements (stones, branches, fabric, water, leaves, etc.)  Surface material (concrete, wood, stone, seamless backdrop)  Overall atmosphere (minimal, organic, dynamic, clinical, etc.) 4. Special Effects or Techniques:  Motion effects (splash, floating, suspended)  Depth of field (shallow, deep, everything in focus)  Water/liquid elements  Macro photography details  Any CGI or composite elements 5. Styling & Mood:*  Color palette  Brand aesthetic (luxury, natural, clinical, playful, minimal)  Emotional tone conveyed  Reference brands it resembles (Aesop, Glossier, etc.) OUTPUT FORMAT: After analyzing the 9 images, generate a complete prompt for recreating these photography styles in a 3x3 grid format. Every frame must clearly state that the PROVIDED PRODUCT IMAGE is used instead of the original product while preserving the same pose, placement, angle, lighting, and environment. Output a JSON object with this structure: { "instruction": "Generate a single high-resolution image in 3:4 aspect ratio (portrait) featuring a professional 2x2 grid (4 individual frames). Maintain strict consistency in design and color. CRITICAL: In EVERY frame, replace the original product with the provided product reference image, preserving exact product details while matching the style, lighting, and composition of the reference photography.", "layout": "2x2 portrait grid in 3:4 aspect ratio with no padding or margin around the outer edges — the outermost frames should extend fully to the edges of the image. Thin gaps (2-3px) of solid black (#000000) only between the frames, not around them. Each frame should be taller than it is wide.", "shot_list": [ "Frame 1: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]", "Frame 2: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]", "Frame 3: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]", "Frame 5: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]", "Frame 6: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]", "Frame 7: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]", "Frame 8: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]", "Frame 9: [Detailed description of the shot, explicitly stating that the PROVIDED PRODUCT IMAGE replaces the original product and is positioned in the same pose, angle, and environment]" ], "style_notes": "High-end editorial photography, 8k resolution, professional color grading." } Output ONLY the raw JSON object, no markdown, no code blocks, no explanations.

Connect the output from the Text Generation node to an Image Generation node and select the Nano Banana Pro model.
Finally, add your product image to the Image Generation node.

Thats it, now hit generate and see the final result.

AI product shot workflow