Train wagon graphics - you are not stretching it properly

[RnMss]

Stretching the graphics is not the problem. The problem is that you didn't do it properly.


Looks like your have stretched the length in the 3D model. (Like the length is a linear function of the rotation, that's why it looks like rubber.)

That's not correct. You should leave length in the 3D model as it is, and instead stretch the generated 2D graphics.

[sckuzzle]

I'm not sure if you are just trolling, but is there a difference? Can you provide an example of the 2d-stretching such that it doesn't look like rubber?

[DaveMcW]

Yes, the bounce effect is due to a bad animation. The function should be: scaled_length = sqrt( length^2 + (length * sin(angle))^2 ) .

But the other point is still valid, a difference of 41% is not hideable.

[RnMss]

I'm not sure if you are just trolling, but is there a difference?

Why? To be honest I didn't get it. What words should I have chosen? (This is a real question)

Can you provide an example of the 2d-stretching such that it doesn't look like rubber?

I'll try. (But I'm not an expert at 3D animations, so it may take some time...)

[kovarex]

The example was just an example, we will use different function obviously. Also we don't need to stretch by 41% anymore, more like 15% and also the model is made in a way to make it not noticable. From the examples I saw so far, it looked very good.

[Sean Mirrsen]

I'm sure it's a "has been discussed since the beginning and could maybe have worked back then" thing, but why not make the visual grid of the game non-square? Fitting isometric graphics into a square grid creates more visual problems than just the trains.

[DaveMcW]

The function should be: scaled_length = sqrt( length^2 + (length * sin(angle))^2 ) .

Here is the result of doing that. Blender does not make it easy, I had to to compute everything in a spreadsheet and enter the differences in 20 keyframes.