Warning!!! Don't Scale an Object to 0

Hu,

i just wastet 2 hours of my life, by trying to find an awsome error.:mad:

I find out, that if you add an Object and scale it like (10,10,0), it will not be recognised by a Mouse-Over-Any Sensor, aslong any Value is set to Zero…

Sorry about this thread, its no Question, but i had to give this away to free my mind…

To the Admins: feel free to delete this, if you thinks this post is pointless…i had to write this somehow…

Waist hours on errors its pretty usual believe me, somethimes stupid things takes half of the day to be solved well its a good information aniway, thanks sevi

This shouldn’t be. I mean a plane has a zero dimension as well.

Edit:

ran a test with 2.49b
If set to zero in Blender it is fine - when reading the scale it is not zero but a very small number 9.99999974738e-05.
If set to zero via Python mouse over does not work as you wrote.

Maybe that has something to do with the bounding box elimination. I do know.

But good finding :wink:

Yes, it has to do with collision Box or so. Here is a screenshot, that shows the difference between two planes, one is scaled with phyton, the other is original. Strange is, that actually both have the same scaling, but the one scaled with phyton looses something. Have a look on the green diagonal line missing in the right plane.

Attachments


What’s the meaning of this?
I can’t have mouse over and selection upon no collision objects?
There got to be a way! I know that the logic bricks rely on bullet physics when it comes to mouse over! But I sure there’s a python way to bypass that! Right?

Bypass: make the scale a very small value rather than 0.0 :wink:

I mean, when one wants to use no collisions in objects or not use bullets at all!
Since objects are accessible via python, I believe a hitObjectList or position wont be required to know which object is the mouse over… thus bypassing collisions requirements for the mouse sensor!
In other words not use the mouse sensor at all!

Rasterizer has a screen ray to function and a scren vec function. From there it should be easy to go with python.

Edit: Sorry - it’s camera that have those functions…