Tips'n'Tricks

The Tips in detail

"Propeller apears to rotate backwards""

by Ron Gidseg and Joey Ponthieux

Last Modified 01/20/2001 19:00:00


<Ron Gidseg>

Problem: You set up an animation of a propeller-driven plane. You double checked everything and

rendering was successfull, but as you watch the final, the propellers apear to rotate backwards.


Solution A (the short one):

Reverse the rotational direction of the propeller.


Problem: In certain situations the rotation will reveal it's reversed direction and mess it up all again.


Solution B (the longer one):

Take 360 deg divided by the number of blades to get blade separation. Divide this value by 5

to get the angular rotation per frame. (This will give the appearance of rotating in the correct

direction.)


An example: 3 bladed prop:

360 / 3 = 120 deg separation

120 deg / 5 = 24 deg/frame


Multiply this by 15 to get deg per second. In this case it works out to 360 deg prop rotation (1

revolution) per second (15 * 24).


For a 4 bladed prop, the prop will rotate 270 deg per second.


What you were seeing was a strobe effect from the low frame rate. The above formula will not be

physically accurate but will look right. It will also loop correctly if you have the blades identical and

if the loop time is a multiple of seconds.


I also suggest you add a disk to the prop area that is mostly transparent (95% transparent) and the

same color as the blades to give the impression of motion blur.

from: Ron Gidseg <ron.gidseg@lmco.com>



<Joey Ponthieux>

Ron's advice will work fairly well, however, I'll take the time to explain why. A propeller spins so

fast, that you can physically recreate the actual spin rate at the RPMs of an actual propeller and the

propeller can appear to be moving backwards if the RPMs and blade count have the correct

parameters. It is not neccessarily an optical illusion. It is a limitation of animation physics. In

animation we break down motion into 30 rames a second for full frame video and/or 60 frames per

second if you are field rendering. Even if the rotation value of your propeller is completely accurate

to the loating point, you can only produce 30 rotation positions for your propeller per second.


Assume you have a two blade propeller, Each blade is 180 degrees apart. Your propeller is turning

clockwise with positive rotation, and at frame one, one blade is at 0 degrees(or 12oclock). Assume

now that your RPM rate is 600rpms, meaning every 1800 frames the proppeller will turn a full

rotation 600 times. For this example we will use 30 frame per second video as our standard. If you

multiply 600(rpms) x 360 (degrees per one full rotation) you get a sum of 216.000 degrees rotation

over 1800 frames. By keyframing 0 degrees rotaion on frame 1 and 216,000 at frame 1800(1 minute)

you will yield accurate 600RPM rotaion. Now divide 216,000 degrees by 1800 frames and you will

yield your Rotation-Degree-Per-Frame which in this case a one frame rotation would be 120 degrees.


While this is accurate for 600 RPM rotation, what it means is that at every frame propeller 1 will

rotate a full 120 degrees. The problem you are dealing with is that propeller two on frame 1 was

sitting on 180 degrees. Propeller 1 , at frame 2, will be closer to the resting postion of propeller 2

when it was at frame 1, than itself(prop 1) at frame 1. So in essence you end up with a negative

rotation of Neg60 degrees per frame. You calculate that by taking the breadth from prop to prop of

180 degrees and subtracting the Rotation-degree-per-frame of 120. To simply here is what happens;


Frame1 Prop2 180 deg

Frame2 Prop1 120 deg

Frame3 Prop2 60 deg

Frame4 Prop1 0 deg

etc.


Because the propellers sharing near distance, alternate within those 60 deg, you end up with negative

rotaion within a 30 frame per second time standard. The solution to the problem is as simple as Ron's

example, however you are not required to divide by 5. 5 will always yield acceptable direction results

in Ron's example, however it may not yield the speed you want.


The solution to the problem can be resolved with any calculation which yields a rotation-degree-perframe

of less than one-half of the degree between the blades. Meaning that if you are using Ron's

example of a 3 blade prop, as he noted the seperation between each blade is 120 degrees. Divide that

number in half, this equals 60, and any rotation per frame of less than 60 will yield a positive rotaion

on the prop. Closer to 0 will make the prop move slower, closer to 60 will make the prop move faster,

if you produce exactly 60 degrees per frame rotation for the 3 blade prop, the prop will almost appear

to stand still since the blades will simply appear to alternate between 0 and 60. Any degree above 60

and below 120 will produce a negative direction to the blade rotation.


Ron's example works all the time because he is dividing by a value which practically insures a value

of less than half per frame in almost all situations. If you were to divide by four with the right number

of props you could end up with a standing prop. 3 will work better, for that matter most odd numbers

above 2 will more than likely work in Ron's example, however I am uncertian of Ron's use of 15

frames per second to get a revolution per second since, at least in NTSC video anyway, there are 30

frames in a second, not 15. Though I am not sure what format he may be working with.


I would note that I have rarely ever been able to get motion blur to work well on a propeller.

Admittedly though, Ive never done a propeller in Max so I dont know if Max might be able to do a

better job with it. You really do need some kind of motion blur to get the right appearance. Rons

suggestion of a circular transparent blade will work, though I would do it in the following manner.


Generate a texture map of the blades blurred along their path of rotation. Then generate a

transparency map which goes from opaque to complete transparent with in the span from blade to

blade and apply this to the circular poly. Attach the circular poly to the actual blade and align the

opaque edges of the map with the blade. Dissolve the poly face in when the blade comes up to speed,

and dissolve the actual blades out. I have also applied this kind of map to a circular extrude, extruded

to the same depth as the blades thickness. This comes in helpful when looking at the prop from the

side. A flat poly will dissappear from the side and the illusion will be revealed. A model with

extruded depth will provide something to see from the side. The opaque to transparent transparency

map will provide more of a sense of motion blur with a low cost on render times.


The phenomena you are experiencing is very similar to that seen on the lugnuts of a wheel as a car

goes down the road. There is an appearance of reverese rotation because the eye's refresh rate is too

slow. At just the right speed it picks up each successive image just as a lug is getting closer to another

lug but just behind it. The affect is an animated reverse of direction.



from: Joey Ponthieux