lemec's exercises...

[lemec]

After the great debate on silhouettes (in the Contents Sharing forum) I decided to try a few simple animations to test my understanding of silhouettes in perspective. It turned out to be a great exercise and although I didn't learn any new perspective fundamentals, I found that it really drilled some good habits into me, and I learned a bit more on an "order of operations" that makes things a lot easier with fewer revisions.

I think after this, I'll probably start trying to move that square off-axis, and then do two at a time, and just try and find ways to make solid objects do the same thing.

This took me THREE tries to get it this good! I learned a lot about how to plan out my spacing.

square_rot1.gif (14.46 KiB) Viewed 25619 times

This was my FOURTH attempt!!!!

square_rot2c.gif (18.13 KiB) Viewed 25630 times

[lemec]

Okay!!! Getting the hang of it!!! This was totally worthwhile.

I did this in two passes. First, I made an oval fly around, edgewise. Then I could use the oval as a perspective guide for the rectangle.

square_rot3.gif (25.28 KiB) Viewed 25629 times

[lemec]

Okay... I figured I'd try a very simple movement, but with a complex object.

Again, this took multiple passes.

First, I used my "black rectangle" pass. I wound up having to stretch out my Instances and play it back at 30FPS to ensure that the car would stop (and bounce forward on its shocks) the way I wanted it to. I invested a LOT of time on this step.

Lesson learned:

Animation isn't illustration. The pretty drawing can happen after the motion is pretty.
You need to indicate where the transition of movement begins and ends in space.
You need to indicate the path through space taken to get there.
You need to indicate the acceleration and deceleration (via spacing) of the object as it travels this path.
You need to decide at the beginning which sections will be done on threes, twos or ones.

After I had a very smoothly moving "black rectangle", I animated two dots to approximate the position of the central pillars (where the division of front and back seat). Using dots was a smart move because I could effectively use the Light-Table, which is great for dealing with spacing for simple objects. Had I tried to use the Light-Table on a complex drawing of a car, I would have had a hopeless mess of lines. I also found that the Light-table is effective as a spacing tool on silhouette outlines (not filled silhouettes).

I chose ONE frame to depict with full details. I chose the one where the car was closest to the camera because it was largest in frame and I could go as crazy with detail as I felt.

Soon afterwards, I found that animating all that detail would have been ridiculous, although possible. But the best thing to do for a rough animation pass was to animate the silhouette outline, and to try and register each frame of the silhouette against my black rectangle. I could easily tell if my lines were lagging or boosting out of position, relative to the black rectangle and the two pillar dots. In retrospect, I should have added a couple more dots to mark the extents of the front and back to make the preservation of volume easier.

I'm not tired of this yet! I'm glad to see that I can really feel the mass of the car getting closer - the animation is working nicely. Now that the largest shapes (and the hardest part) is over, I have a great framework against which to register the smaller details.

[masterchief]

Mark,

You are doing this utilizing a function of your RocknRoll plugin??

regards,
William

[lemec]

I'm using RockNRoll's rolling/flipping functions, fullscreen crosshair and Paper Trail. Sometimes I freeze the Paper Trail. I've got that mapped to a foot pedal.

[lemec]

carstop.gif (52.98 KiB) Viewed 25486 times

Since the car was travelling on only ONE axis, it made sense that the motion paths of all of its verticies would be parallel, and therefore share a single fixed vanishing point.

So, I turned off my fullscreen crosshair, paper trail, and then turned on Protractor! I only use about 1-3 points to align various shapes and completely redrew the last four frames. I'm really happy with the solidity of the volumes - and now I can impose a slight lens distortion to the animation.

Learning a LOT about tweening without a light-table. the drawing has reached a stage of complexity where everything dissolves into a morass of stray lines with the LT on. So, I've found that I can "rock forwards" to the next frame, hover my stylus cursor over a vertex, then "rock back" to the previous frame, and gauge the amount of travel of the vertex I'm trying to tween. I can then make a much better judgement of where the vertex should go. I stopped trying to tween individual lines - it's too difficult and unreliable. However, by tweening just the verticies, or specific areas where the surface form experiences a significant change of direction, it is easy to interpolate the connections between verticies.

It's a very painstaking process, but it's becoming gradually more efficient. This has been a great exercise so far. It almost looks like I rotoscoped it.

[lemec]

DONE!

Learned something new...

Since I'm using Protractor (which obeys the laws of curvilinear perspective) the perspective of my old sliding rectangle placeholder is incongruous.

So, I switched to using a bounding box to indicate the change in position of the car - much easier to register the volume of the car against the bounding box, and the bounding box gives me a proper sense of volume under curvilinear perspective distortion.

Once I'd worked out the issues of keeping perspective and inbetweening (without popping) it simply became a lot of work - not a bad thing because it helped to drill some good habits into me and by comparision, the animation of living things seems a whole lot easier now.

My back aches, just a bit...

[lemec]

...by comparision, the animation of living things seems a whole lot easier now...

Tomorrow I shall put my money where my mouth is. Next exercise:

Inbetween this:

Click to view

[D.T. Nethery]

Learning a LOT about tweening without a light-table. the drawing has reached a stage of complexity where everything dissolves into a morass of stray lines with the LT on.

So, I've found that I can "rock forwards" to the next frame, hover my stylus cursor over a vertex, then "rock back" to the previous frame, and gauge the amount of travel of the vertex I'm trying to tween.

I can then make a much better judgement of where the vertex should go. I stopped trying to tween individual lines - it's too difficult and unreliable. However, by tweening just the verticies, or specific areas where the surface form experiences a significant change of direction, it is easy to interpolate the connections between verticies.

Mark,

This has been a very interesting. Thanks for posting your work in progress.

I agree with you that learning to do accurate inbetweening without the light table is a valuable skill. My teachers at Sheridan College insisted that we learn to do it first without the backlight , using only the method of rolling the drawings on the pegs to establish our major shapes in blocking in the inbetween drawings . (then use the backlight to check certain details and for doing very close , mechanical line-between-line slow-ins or trace-backs , that sort of thing)

Being able to do this method of inbetweening , "eyeballing" it as I'm flicking back and forth with the drawings held lightly between my fingers as I check the shapes and volumes is what I miss most when I'm working on the Cintiq .

m03_Flipping.gif (10.13 KiB) Viewed 25310 times

How are you handling the part you describe as :

"rock forwards" to the next frame, hover my stylus cursor over a vertex, then "rock back" to the previous frame"

What are you using to rock forwards and rock back ? The best I can do right now is to use the forward arrow key and back arrow key shortcuts programmed into the Griffin Powermate knob . I rock it back and forth to simulate rolling the drawings. Have you found some other way to do it ? Just curious.

[ZigOtto]

isn't that exactly the TVPA [Flip] button purpose ?

[lemec]

How are you handling the part you describe as :

"rock forwards" to the next frame, hover my stylus cursor over a vertex, then "rock back" to the previous frame"

What are you using to rock forwards and rock back ? The best I can do right now is to use the forward arrow key and back arrow key shortcuts programmed into the Griffin Powermate knob . I rock it back and forth to simulate rolling the drawings. Have you found some other way to do it ? Just curious.

When I tried animating traditionally, I used the same method you've shown - bottom-pegs flipping, although when I did my inbetweening, I would put my tweens in the middle of the stack that I was flipping, rather than in the front of the stack. As I made adjustments to the inbetween, I'd just pull the front pages forward, out of the way, but I'd "roll" the pages back and forth to ensure that I had a good continuity of movement.

So my RockNRoll plugin does essentially the same thing. I have it configured, with GlovePIE (a keyboard/gamepad/input device mapping software), to operate in conjunction with a gamepad. When I press and hold down on the D-pad, RockNRoll will play forward from the "current frame" until it hits a bookmark or the end of the animation. Then it will pause until I release the D-pad, and it will play backwards until it returns to the current frame. Up on the D-pad does much the same thing, except in reverse. So I can "rock back" and "roll forwards" by holding up and down on the D-pad. The advantage to using RockNRoll over something like a mousewheel or powermate knob is that the flipping obeys bookmarks, which I strategically place at the extremes of a transition of motion, and the flipping remembers what frame I was working on, and returns to it when I'm done previewing.

This "rolling" behaviour may be switched off, so that it just snaps forwards and back between bookmarks. The rolling behaviour is good when I want to ensure smooth and continuous movement without popping. Snapping behaviour is good for comparing an inbetween with the extremes, or with certain "reference" frames. Reference frames are what I call any frame where something is in good view, and is undistorted by squash or stretch and can be used as a good reference for proportions and construction. I snap between by inbetween and my reference frames to make sure my inbetweens don't start to wander and deliquesce and lose their form (which happens a lot with straight-ahead animation).

Rocking back and rolling forwards also gives me the advantage of being able to work as if I'm working "straight-ahead" and "straight-reverse", attacking the motion in both directions and avoiding "wander".

- Sorry I haven't posted any updates in a while - I've had a number of commissions! Gotta pay the rent, somehow...

[D.T. Nethery]

isn't that exactly the TVPA [Flip] button purpose ?

For checking a completed action, yes, but there is no precision in it for doing inbetweening .