A/B Testing Visuals

Generate multiple visual variants of the same product using different prompts, styles, or AI models — then compare results side by side to find the most effective imagery for your marketing.

Goal

Create 2+ distinct visual variants of the same product to test which performs better in ads, product pages, or social media.

What You'll Need

A product image (from Shopify Import or Image node)
At least two different Text Prompt nodes with different style directions
Sufficient credits for multiple generations

Build the Workflow

1. Add your product source

Add a Shopify Import or Image node with your product photo.

2. Create variant prompts

Add two (or more) Text Prompt nodes, each describing a different visual style:

Variant A — Clean studio:

Product centered on pure white background, professional studio lighting, clean commercial photography, minimal shadows

Variant B — Lifestyle context:

Product on a rustic wooden table in a cozy kitchen, warm natural window light, lifestyle photography, homey atmosphere

3. Add Image Generation nodes

Add an Image Generation node for each variant. Connect:

Each Text Prompt → its Image Generation node's prompt input
The product image → both Image Generation nodes' reference images input

Use the same model and aspect ratio across both to ensure a fair comparison.

4. Collect and compare

Add a List node and connect both Image Generation outputs to it. This lets you compare all variants side by side.

5. Run and evaluate

Press Cmd/Ctrl + Enter to generate all variants. Review the results in the List node and select the winning variant.

Node Chain

                  Text Prompt A → Image Gen A ─┐
Product Image ────                              ├─── List
                  Text Prompt B → Image Gen B ─┘

Ideas for Variants

Variable	Variant A	Variant B
Background	White studio	Lifestyle setting
Lighting	Bright, even	Moody, dramatic
Angle	Front-facing, centered	45-degree, dynamic
Context	Product only	Product with props
Style	Minimalist	Rich, detailed
Season	Summer, light tones	Winter, warm tones

Advanced: Multi-Model Testing

Test the same prompt across different AI models to compare output styles:

Use a single Text Prompt connected to multiple Image Generation nodes
Set each Image Generation node to a different model (e.g., Imagen vs. Flux)
Collect all outputs in a List node

This helps you discover which model works best for your specific product type.

Pro Tips

Keep one variable different between variants for a clean test (e.g., only change the background, not the lighting AND angle)
Use batch generation (3–5 images per variant) since AI output varies — pick the best from each batch
Label your groups with Group nodes: "Variant A — Studio", "Variant B — Lifestyle"
Use the same model and aspect ratio across all variants for a fair comparison
Download winning variants and test them in your actual ad platform or product page

On this page