Physics shapes less than 0.5m

[Apple Pumpkins]

Is this a 'feature' or a bug? If my mesh is less than 0.5m in any dimension, the custom physics shape reverts to convex hull. On anything I make, not isolated to this specific mesh. Notice on the attached image the size in the Y axis; image attached to post, any help would be appreciated and thank you.

[Aquila Kytori]

Hi

37 minutes ago, warehousefifteendesigns said:

Is this a 'feature' or a bug?

Its a feature.

The workaround is to add an extra triangle or vertex ( see my reply in this thread: 

  arton will be home shortly and can explain how to do the vertex thing in 3DMAX) to make the object at least 0.5m along the relevent axis.

 

[arton Rotaru]

2 hours ago, Aquila Kytori said:

arton will be home shortly and can explain how to do the vertex thing in 3DMAX)

Sure thing, I can do that if needed.

[animats]

Interesting. Can you have a zero-thickness physics shape, and does it work?  I've seen a bike which appears to have such a thing, a plane at the centerline which supports the bike. Murasaki's bikes work that way. I can't tell if the plane is very thin, or actually zero.

The usual collision detection algorithm, GJK, won't work right on a zero-thickness object, so I suspect that convex hulling is also being used to  inflate it a bit.

[arton Rotaru]

1 hour ago, animats said:

Interesting. Can you have a zero-thickness physics shape, and does it work?  I've seen a bike which appears to have such a thing, a plane at the centerline which supports the bike. Murasaki's bikes work that way. I can't tell if the plane is very thin, or actually zero.

The usual collision detection algorithm, GJK, won't work right on a zero-thickness object, so I suspect that convex hulling is also being used to  inflate it a bit.

If you use a triangle based collision mesh, then yes. However, if you set a triangle based collision mesh physical it will switch to convex hull secretly, and with that there is the so called collision radius of 5 cm like a cushion around the hull. That collision radius is a feature of the Havok engine as well.

[Apple Pumpkins]

3 hours ago, Aquila Kytori said:

The workaround is to add an extra triangle or vertex ( see my reply in this thread: 

  arton will be home shortly and can explain how to do the vertex thing in 3DMAX) to make the object at least 0.5m along the relevent axis.

 

 

1 hour ago, arton Rotaru said:

Sure thing, I can do that if needed.

Thank you guys, creating a sneaky vert has done it something I will keep in mind from now on. Thanks again

1 hour ago, animats said:

Interesting. Can you have a zero-thickness physics shape, and does it work?  I've seen a bike which appears to have such a thing, a plane at the centerline which supports the bike. Murasaki's bikes work that way. I can't tell if the plane is very thin, or actually zero.

The usual collision detection algorithm, GJK, won't work right on a zero-thickness object, so I suspect that convex hulling is also being used to  inflate it a bit.

I tried using planes for physics on floors but ran into the same issue where the floor in 3DS was zero on the z axis, the convex hull problem. I also noticed during this process that the uploader pulls verts very subtly on completely flat planes to be the SL minimum size for any object - something I'll have to add verts to get around too I think.

[Chic Aeon]

Just wanted to note that I understood the conversation AND I made a door day before yesterday that works perfectly WITHOUT that nasty triangle "hinge" (WOOT). I can't say the process is second nature yet but at least I got things working. 

Poco a Poco. 

 

Also using cube (analysed) physics also works to alleviate the issue, no?  I think that is why WAY BACK when -- I started using cubes -- Drongle, it's ALL Drongle's fault LOL.  

Edited March 8, 2018 by Chic Aeon

[ChinRey]

On 8.3.2018 at 7:54 PM, animats said:

Interesting. Can you have a zero-thickness physics shape, and does it work?  I've seen a bike which appears to have such a thing, a plane at the centerline which supports the bike. Murasaki's bikes work that way. I can't tell if the plane is very thin, or actually zero.

It only works if it's actually zero. Here is a three year old picture of me standing on such a sheet. It has perfect physics:

[animats]

I meant a plane edge-on. The vehicle control system keeps motorcycles upright; you don't need support against roll. The ideal "sled" for a bike is narrow, like a skate blade. Zero width turns out to work until it lines up with the joint between the left and right lanes of a road. Then it falls into the slot. So in practice there's a minimum width.

It turns out that Murasaki's bikes ride on a thin ellipsoid prim, not a flat plane.