The following are extracted shifts as shown on one
animated element, rather than the multiplied ones
as before. This is shown in parallel to the actual
movement of structure as one walks past, to show
exactly what is happening in the videos: 3 series of
3 videos each, foreach stage i took while improving
As I have been working on these animations, testing various options and parameters, which is still an ongoing process, I realized how difficult it is to give the movement I want without any distortions. It's like making a map: there are classes of maps but each of them contains a certain distortion that comes from bringing a three dimensional space onto a paper. There is a whole set of rules which form and technique of making maps is appropriate for which scale and scope of the map to minimize the distortion and make is comprehensible to the reader.
I feel as if the same happens here, with my animations. Because I work with a screen-based (flat) depiction of a three dimensional space, and on top of that I deal with two sets of motion: human movement through thespace, and the movement of the space itself.. The way I was applying the structural shifts to my animations before, was to keyframe the shift in position every one or two seconds for all the layers in the relevant 1 or 2 seconds.
This resulted in the positioning of layers arriving at one point where the keyframes were and giving the 'ribbon result', which was rather annoying and distracting
I then came with a new method where I apply the shift to every new frame where is begins, so rather than moving from an already established point towards a keyframe that's shifted (in the future so to say), it shifts at the spot, moving instead to an already established point later in the animation. This gives a more smooth motion of all the layers together, and so works better visually, but also results in a much faster shift of the elements, and so only really works in reduced frame rate; the motion applied of 1 pixel per frame using the previous method was comparable to 1pixel per 5 frames with this technique. Another problem here is that, because the layers are only shifted at the present point, and move towards a pre-set final point (unlike previous way where the keyframes WERE placed at a future point in timeline), the closed they get to that final point the more obviously they steer towards it, even if the shift has moved them beyond that...
Following my last animations that showed the spaces
closing in on thevisitor, I went back to try ones that
shift wide open again, changingthe parameters slightly.
In this video, with speed reduced by half, I applied 1
pixel shift every 5 frames.
First thing I realised with this work is that it does not
matter so much what shift I apply, as long as the video
is played initially at the correct speed, it will work.
The main difference will be the intensity of the shifts
and how perceptible they are to the viewers' eyes.
Another thing that this animation proves is that the
spaces that close in around a person look much more
powerful in the film than these that open wide.Therefore
I will focus mainly on that from now, at least for a while.
The following two animations are variations on the one
above, first with only half of the layers left, the figures
shifting 2pixels per 10 frames;
and another version when I kept the opacity of all
elements at 100% throughout to see what the visual
result of that will be: