Marketing AI OS 'ENSOR' Adds 'Element Swap' Feature for Ad Banners — Upload Two Images to Precisely Replace Only Selected Areas While Preserving Background, Composition, and Lighting

REHATCH adds an ad banner element swap feature to its marketing AI OS 'ENSOR'.
NQ 56/100

📋 Article Processing Timeline

  • 📰 Published: March 30, 2026 at 00:26
REHATCH Co., Ltd. (Headquarters: Chiyoda-ku, Tokyo; CEO: Ryota Sakoda) announces the addition of a new "Element Swap" feature to its marketing AI OS "ENSOR," designed to support ad banner production.

■ Background

In display advertising, there are many cases where only the product image, logo, or key visual needs to be swapped while keeping the same layout. However, when an image generation or editing tool rebuilds the entire image, the cohesion of the banner — background, composition, and lighting — tends to break down, leading to revisions and rework.

- Want to swap only photos or logos to match the message or placement, without changing the background or lighting
- Want to turn around swap tasks quickly while preserving design consistency
- Too many steps make it hard to integrate into the production workflow

With this new feature, ENSOR lets users upload just two images — a pre-masked banner and the replacement asset — to obtain an image with only the specified element swapped.

■ Feature Details

This feature precisely replaces a specific element within an ad banner (product image, logo, portrait photo, etc.) in a user-selected area with a different image. Users can focus purely on swapping the element while the background, composition, and overall lighting mood outside the selected area are preserved.

[Image: Example of property asset swap for real estate (custom-built homes)]
[Image: Example of product asset swap for e-commerce (health drink)]

▼ How to Use (2 Steps)

Step 1: Launch the "Element Swap" feature and prepare your images following the on-screen instructions.
Step 2: Upload two images together — the ad banner with the selection area already marked, and the replacement image.

Processing begins as soon as both images are provided, generating the result without unnecessary confirmation steps. Fine-tuning the output or swapping a different part can be done by repeating the same steps.