A few weeks ago we have emailed all of our customers who are using the /image/optimize robot with a custom value for its png_tool parameter, informing them about the upcoming removal of said parameter and how it might break backwards compatibility for them.

Our image optimization robot

We have decided to push for this change because our customers frequently report that they are unsure whether they should use optipng or pngquant for their PNG file optimizations. Additionally, more often than not, they are even unclear about the implications of using these tools at all. Choosing the best tool for the job generally depended on the image at hand and it was not efficient to just use both and then select the best result in your app. Nor should you have to - after all, that is what you are paying us for.

We are fixing this now by completely removing the png_tool parameter from the robot. We have launched this change in sync with affected customers last week and so far all systems are green.

Instead of the customer manually supplying the 'best' tool, the robot now runs several underlying optimization tools on your files in parallel. It then picks the result of the tool that was able to deliver the smallest (non-zero bytes) optimized version to be the Assembly Step's result file.

We expect smaller outputs across the board as a result of this change, leading to a better experience for your users and lower bandwidth consumption on your CDN. It will also result in a lower Transloadit invoice, because regardless of how many optimizers the robot runs on your input file, we only charge for what ultimately goes in and out of the robot.

In addition to this improvement, we have also added support for GIF and SVG files to our /image/optimize robot. :sparkles:

In the course of the next few days we will automatically change your templates to remove the - by then deprecated - parameter. There is nothing that you need to do on your end, except to monitor if all your optimized images are still displaying fine. Our test suite gives us confidence that they will be, but please report any problems at first sight.

We hope you will enjoy these improvements to our /image/optimize robot as much as we do. :)