JA EN
shiichan

CloudFront's Image Transformation Just Got 4 New Features!

AWS What's New ★★★★★ View Markdown
#aws#web#serverless

Hi, it's me, Shiichan! Today's news will excite anyone serving images through CloudFront.

AWS What's New aws.amazon.com

What was announced?

AWS What's New announced four new features for Dynamic Image Transformation (DIT) for Amazon CloudFront — the service that transforms and optimizes images in real time as they're requested.

The story so far

Until now, DIT made it hard to combine multiple detection methods for advanced cropping, and some devices simply didn't receive optimized images at all. The automatic optimization side was especially uneven across browsers — roughly 30% of all traffic was actually being served unoptimized. There was no dedicated tool to test transformation results, and the feature set differed depending on whether you ran DIT on ECS or Lambda.

What changes

Here are the four new features:

  • Enhanced smart cropping: Combine multiple detection methods — faces, labels, text, logos, and custom Amazon Rekognition models — in a single request. Aspect ratio, padding, and gravity constraints are configurable and can be prioritized based on business needs
  • Enhanced automatic image optimization: A new tiered detection approach uses CloudFront's multi-tier device detection, layering device classification headers and configurable fallbacks behind Client Hints. This closes the browser-support gap that previously left images unoptimized and extends right-sized delivery to every device type, from phones and tablets to smart TVs
  • Interactive image transformation playground: A testing tool that shows transformation results on the spot, along with detailed metrics like original dimensions, output format, file size, compression ratio, and processing time
  • ECS/Lambda feature parity: The same full feature set is now available regardless of which architecture you choose

Dive Deep

The key part of enhanced smart cropping is that it no longer relies on a single detection method like faces or labels alone — it combines multiple detection results so that products, text, logos, and specified objects stay inside the crop. Because you can prioritize elements and control the composition, it's a great fit for images where certain elements absolutely can't be cropped out, like product photos on an e-commerce site. It's available in all commercial Regions, plus the opt-in Regions of Hong Kong, Bahrain, Cape Town, and Milan.

Wrap-up

  • CloudFront's Dynamic Image Transformation gained four new features: enhanced smart cropping, improved automatic optimization, a playground, and ECS/Lambda feature parity
  • Combining multiple detection methods for cropping keeps products and logos intact in the composition
  • Multi-tier device detection now reaches the roughly 30% of traffic that previously missed out on optimization
  • The playground lets you check transformation results and metrics on the spot
  • Worth trying right away if you're building out image delivery on CloudFront

Related posts