The 16 mm film format mentioned is usually a motion-picture, rather than still, camera format, generally shot at 24 frames per second (FPS), though values from 8 to 48 FPS were fairly common.

Presumably the stacking was of 100s of frames would involve combining several seconds (8.33 seconds would be 200 frames of film at 24 FPS) to generate a higher-resolution image. Success would depend on how much camera and subject movement occurred over that time, though image stabilisation should help somewhat with the former.

I suspect a clearer explanation of the process was omitted from the article, perhaps due to poor editing.

To take many exposures of the same scene and average the pixel values with the corresponding pixels from all the exposures. This increases the SNR and dynamic range, but naturally doesn't work very well if your subject isn't static. It doesn't increase the resolution of the image though.

For the 16mm movie cameras in this case, they probably selected frames from rigidly mounted cameras with little subject motion to get a good result out of it. Glenn strapped in tight in his tiny cockpit probably provided them with a decent number of frames they could stack without introducing much motion blur. In fact, you can see a bit of blurring at the end of one of the white straps center frame in that shot: https://cdn.arstechnica.net/wp-content/uploads/2025/09/03b-G...

It's a pretty common technique in astrophotography. Here's a page that goes into some more detail in that context: https://clarkvision.com/articles/image-stacking-methods/

There are algorithms in computational photography that effectively "stack" multiple images of the same thing in order to achieve a kind of 'super-resolution'. It's been a bit since I've done any of that, so I do not recall, but my hazy memory says its pretty straight forward linear algebra (and has the concomitant drawbacks that algorithms of this type have).