Hi Felix - unlike a circle the ellipse from center point takes points for where to place both its horizontal (corresponds to width entered) and vertical (corresponds to height entered) axis points.

So until you are at the stage of placing the second axis point, only a line will be drawn from the center to the first axis point.

The command does not yet have all its information that it wants to make an ellipse until you are at the stage of picking the 2nd axis point.

The click for setting the 2nd axis is kind of redundant when width and height are given when you are only working in a 2D viewport like only drawing in the Top view, but you can see a bit more clearly how it works if you place the last point somewhere in 3D - that orients the ellipse. For example try drawing an ellipse in the 3D viewport and for the last axis pick a point using z snap upwards, you'll see the ellipse turns upwards then.


That is correct that it is different than drawing a circle. Basically the difference comes from the process of how things work when you do not enter in any numeric values. When clicking points only, a circle picks a center point and then a radius point, while the ellipse picks a center point, an x axis point, and then a y axis point for the ellipse.

When you enter in numeric values they behave a bit differently - the circle finishes immediately when you enter in a numeric value, but the ellipse command still asks you to pick the same points that you would use if you had not entered numeric values - that's because a circle has a sort of uniform symmetry about it so that it doesn't make much difference which direction its x axis is pointing inside of its plane, so it just makes its x axis point in a default direction when you enter a radius value numerically.

- Michael

Michael,

ok, I understand why there is a difference in behaviour especially for the 3D view but shouldn't this behaviour be adapted to recognize that we are in a 2D view and behave like the circle command. By the way, the circle command could benefit from a similar behaviour (ellipse) while in 3D view and wait until we choose the desired axis especially when we specify the radius manually? I know it's possible to control the circle axis while in 3D but if you enter a radius value manually it jumps right back flat on the current cplane.

I don't know if this is pertinent or not but you talk about something like a property page that you wanted to implement. I learn recently that we can click on the line below the style that gives us info on the selected object and edit some "properties". Maybe, you could expand from there instead of adding a buttons or maybe just use something like mouse_over to give a clue to what's possible. In some program I've used, I've seen a property page like thing that allow us to enter values to translate, rotate and scale on X,Y and or Z maybe something like that would be sufficent, just a thought.

Thanks,
Felix

Hi Felix,

> but shouldn't this behaviour be adapted to recognize that we
> are in a 2D view and behave like the circle command.

The problem is that I can't tell in advance if you're going to be doing all the picks in a 2D view - it's totally possible to do something like pick the first 2 points in one view and then switch to a different view for picking the third point.


> By the way, the circle command could benefit from a similar
> behaviour (ellipse) while in 3D view and wait until we choose
> the desired axis especially when we specify the radius manually?

If you need to control the orientation of the circle, it's possible to control that fully by setting up the cplane in the orientation you want before drawing the circle. The tool for that is under View > CPlane - when you enter in a numeric value for a circle the orientation of the cplane where you clicked the origin point in will be used. In some cases snapping the point on to some geometry while in the 3D view can use that geometry as a temporary cplane, like if you snap on to a center point of a box face for example. So if you have some planes in your model you can actually draw on to them directly in the 3D view without setting a cplane first in that case.

Several other commands are similar - for example if you want to control drawing a rectangle on a particular plane, set the cplane to the plane you wish and then 2D type drawing commands will go in that particular plane.


> I learn recently that we can click on the line below the style
> that gives us info on the selected object and edit some
> "properties". Maybe, you could expand from there instead
> of adding a buttons or maybe just use something like
> mouse_over to give a clue to what's possible.

That particular line in the UI that you're talking about there gives a size readout of the current selection - so it seems pretty logical that when you click on it you will get some controls to alter size properties, but it doesn't seem quite as logical to think to go there for other kinds of properties totally unrelated to size though...


> In some program I've used, I've seen a property page like
> thing that allow us to enter values to translate, rotate and
> scale on X,Y and or Z maybe something like that would be
> sufficent, just a thought.

Probably these other programs have the concept of a kind of local axis point that is associated with every object, and those controls that you are talking about deal with manipulating that local axis origin and orientation.

MoI doesn't currently have that same kind of additional local axis point associated with every object - it's something that is kind of weird in a certain sense with CAD programs because stuff like having little axis lines associated with every little 2D line in your drawing is kind of a strange thing.

That could be something to add in the future but I'm not entirely certain about that since like I mentioned it's kind of odd with lines and other construction geometry. It's something that's more prevalent in software that is more focused on generating animations of all the various stuff in the model.

One thing that I'm worried about is that people seem to get a kind of false sense of accuracy with that local origin system where they feel that setting its x y and z position means that they are positioning "the object" at some specific location, but what's actually happening is that they're positioning a kind of arbitrary local origin point that's just associated with the object and may not even be touching or inside the object geometry at all.

In MoI if you want to position a point of an object at a specific x,y,z location you can use the Transform > Move command to do that.

- Michael