Don’t Turn It into an Eye Test! Learn How to Properly Resize Your Images with Amped Replay and Why It’s Important

Hello, dear Amped blog fellows! This week’s tip is dedicated to Amped Replay‘s Resize filter. Don’t be disappointed: although it’s one of the simplest tools, we believe it deserves a tip for a reason. Keep reading to find out!

If you’ve ever used Amped Replay, then you know how easy and fast it is to go through its iconic workflow.

Everything is just in the right place, and filters in the Enhance panel are already sorted in the proper order, so to speed up your work and prevent mistakes as much as possible.

You may have noticed that the last filter in the Enhance panel is Resize. When you click on that filter, you’ll be shown the current image resolution, and you’ll be free to choose a resize factor (using the dropdown menu) or manually entering a target size.

There are two main reasons why the Resize filter stays at the end of the workflow:

  1. The Resize filter uses bicubic interpolation to enlarge your image, as it provides better visual output compared to simpler algorithms. Since we’re introducing new (interpolated) pixels, we believe it’s better to do so after you’ve already compensated for other defects affecting your image.
  2. After you’ve applied all necessary enhancements, we believe it’s important that you produce a final image having a decent pixel size. Let’s think of a case where you’re focusing on a license plate, so you cropped your frame to keep that detail only as shown below.

The Resize filter we’ve just enabled informs us that the current image resolution is just 65 x 24 pixels. In the image above, it seems much bigger just because Replay’s Viewer is showing it at 400% its real size (as confirmed by the red writing below the pixels). That’s why we recommend that you set Replay’s Viewer zoom to 100% to have an idea of what your image will look like before exporting your result. To do so, just click on the “converging arrows” button shown below, or just scroll your mouse wheel until you reach “Zoom 100%”. This is what the cropped license plate looks like at its native size in our example:

That’s tiny, isn’t it? Now, what do you think will happen if you export this image and send it to a less-trained colleague or to your boss? They will likely open the image with their default computer/smartphone image viewer, and they’ll zoom to make it “human-friendly”. And we don’t like this, for two reasons:

  1. We don’t know which software our addressee will be using to view the image, so we don’t even know how the image will be rendered if they zoom it. Will it be a bare nearest neighbor zoom, a bicubic zoom, or a fancy-and-undocumented-proprietary-algorithm zoom? In other words, we don’t know how the image will look like on our addressee’s device if we force them to zoom on it.
  2. Worse yet: if we’re sending the image to different people, they’ll probably be using different software, so they’ll likely get different pixels once they zoom. Believe us, in some cases this can make a difference!

That’s why we recommend that you produce a final picture that does not need to be zoomed to be looked at. Considering that both computers and smartphones today have at least 1280×720 resolution, we can safely resize our license plate by a factor of 400% and get something usable at its natural resolution, as shown below.

Notice that, thanks to bicubic interpolation, the license plate above looks better than when we viewed it at 400% its size using Replay’s Viewer (go up two images to convince yourself). Indeed, when you zoom using the Viewer, Replay simply “enlarges” pixels but does not interpolate them, because we want to let you assess your actual image quality. Interpolation is only applied by the Resize filter.

This week’s takeaway is: don’t turn it into an eye test! Help your colleagues and other people by providing images at a good resolution, so they don’t have to apply the processing you omitted!

One final note before we say goodbye: while in this tip we concentrated on the problems of sharing pictures/videos that are too small, the same reasoning applies to the case of “too big” images. If you send out an outrageously magnified picture or video, chances are it will be down-scaled when viewed on a standard monitor.

That’s all for today! We hope you’ve found this issue of the Video Evidence Pitfalls series interesting and useful! Stay tuned and don’t miss the next ones. You can also follow us on LinkedIn, YouTube, Twitter, and Facebook: we’ll post a link to every new tip so you won’t miss any!