Dear friends, glad to meet you again for one more post of the “How Do I Do This?” series! Despite many out there saying “sizes don’t matter”, when it comes to videos this is not as true. For example, you can’t read something that is just too small. Today we’ll learn how you can resize the frames of a video and zoom with Amped Replay. Also, we’ll get some ideas on why this could be extremely important on some occasions.
Resizing, or “scaling”, is one of the oldest applications in image processing. The idea is pretty simple. We have an image of size 50 x 100. We want to enlarge it, perhaps by a factor of 3, so that it becomes 150 x 300 (we normally want to preserve the original aspect ratio, which means, scaling the height and width by the same amount).
So we need to “create” a lot of new pixel values! This is not done by magic, but with an interpolation process. The computer will use a formula to compute new pixel values from existing ones. When your zoom just “makes pixels larger”, it means you are using the minimal “nearest neighbor” interpolation. This is useful to get an idea of the available information: you preserve the original pixels, just making them larger. But if you want your enlarged image to look better to the eye, then more advanced approaches exist, such as the “bicubic interpolation”. We’re now going to see that both of these interpolations have a place in Replay.
Let’s work with an example. For the sake of bringing things to an end, let’s pick up last week’s example and continue from it. We had started with a video of a car, which originally looked like this (the magnified rectangle was added by us, with Replay of course):
“Hey, that’s SMALL!”, I can hear you say. Luckily, there’s a promising Resize filter just waiting for us. We activate it, and we’re presented with this:
As you can see, we can either choose from a dropdown list, or we can manually write the target Width. Also, we can let the filter compute the other value for us (or the same for the Height). The checked “Ratio” box ensures that we’re preserving the original proportions so that the image will not look stretched in any direction. If you uncheck it, then you’re free to enter both the Height and Width, do it at your own risk ;-). So here’s how the image looks after choosing 400%:
By resizing the image we’ve obtained two important results:
- It looks visually better: thanks to bicubic interpolation, the obtained image looks less “pixellated”, therefore more natural and easier to read or understand to the human eye;
- It has a decent size (676 x 484), so everyone can look at it at its natural resolution (no need to make additional zoom) after you export it. Instead, exporting an image which is just too small will force the receiver to zoom it at viewing time. You don’t know which software they will use, so you don’t know which interpolation algorithm will be applied; in other words, you’re sending out images knowing that they will look different to the receiver’s eyes – not the best option in forensics!
One last note: if you’ve ever used Amped Replay, you surely have noticed that you can just scroll the mouse wheel (or click on the magnifier button at the bottom) to see an enlarged version of pixels. This is very different than using the Resize filter. It is a viewer functionality, so it will not affect the exported result. Also, it will only “make pixels larger” using the nearest neighbor interpolation mentioned above, since its only goal is letting you evaluate how “much information” you have to start with.
When you’re viewing the image at a zoom level different than its actual size (which means, different than 100%), the Zoom writing on the bottom becomes red to warn you that what you’re seeing on the screen will NOT match the exported image.
That’s all for today! We hope you’ve found this issue of the “Amped Replay Tutorials: How do I do this?” 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 tutorial so you won’t miss any!