Filmstrip Animation - Test

So i was thinking about some of the animations other were doing. Some are doing a simple X or Y coordinate movement based on some time function. I began to think, could an animation be made in the form of a film strip? Logically you could have a infiniately long (or tall) filmstrip that moved at some incremental frame rate.

I created a test face with a film strip image I created that was 320 x 16000. The upload gets resized to 6.4 x 320. Scaling it back up distorts the image to an almost unintelligible image. For proof of concept this doesn’t matter a ton, but will be an issue for anyone trying to use this in a real design. To then animate the image I want it to move on the Y axis at 320px every 0.20 seconds. I tried using this calculation:


The 8000 is because the image centerline is the anchor point, so the image needs to move down 8000px to start. There is a negative sign before the 320 to ensure the image moves up.

The test appears to work. I need to work further to see if creating a 1:1 image will stop the image from rescaling. I am thinking there may be no way around this if it tries to bring the image down to 320x320 size regardless of the starting size. Rescaling after import just magnifies already compressed data, so the image is destroyed.

Anyway, here is the test face if you want to play with it:

My real purpose in testing this was to see if this was a more efficient way to animate images. Is this less CPU or memory intensive. With some of the more detailed animations I have noticed that when using more that 30-40 images, the animation can start to judder or skip frames on my Huawei watch. If I’m trying to be fancy I’d like to get a 10 sec animation with a 5 frame/sec rate. This is equal to 50 320x320 images that cycle on an off their transparency.

Is this Axis movement any more or less intensive. What happens behind the scenes on the watch? are all frames loaded in memory waiting to be displayed? The same would go for the filmstrip, is the image that is beyond the border of the watch frame loaded into memory? I guess more testing is needed, but the resize issue may poo poo any plans of testing this correctly.