Pivot Point Location & Rotation

[EnCore Mayne]

is there a learned/experienced way of establishing a precise point on a mesh object (made in Blender) so that it rotates around where the pivot point was placed? let me explain further. (refer to illustration below) i have a mesh design consisting of 3 objects. the center object (part 3) must remain stationary while the two outside parts (1&2) should be able to rotate individually around where part 1 & 2's cylindical pieces connect to the main static object's edges.

the problem comes in when i try to rotate the outside parts inworld (as a link). the center is anywhere BUT where i needs it to be. i can do this with inworld prims, i just add an equal and opposite (yellow) part with the green cylindrical prim being the root. make the extended part transparent and nobody knows how it was done. not so easy with meshes out of Blender. so..., could anyone advise me as to how to accomplish this herculean task, if indeed it can actually be done?

[Wulfie Reanimator]

Just extend the model with some extra geometry, so that the "geometric center" in Blender is where you want the "center of rotation" to be in SL.

My old comment has a video example:

https://community.secondlife.com/forums/topic/454526-smooth-rotation-of-a-linked-prim-about-an-axis-that-does-not-intersect-the-prims-origin/#comment-2115268

[EnCore Mayne]

well i never. who woulda thought it?

actually, it makes perfect sense. all i have to do is get an exact copy of the rotating object, rotate and invert it, et voila. that works a charm. yet, with that part being so complex i tried something similar to your video making the extention a simple cube. problem with that was, i had to drag the farthest extent WAY out with gross movements barely moving the origin. i had to make that extension part an editable part whose dimensions would be reasonable for the user. i tried again with the duplicated, inverted, rotated part but this time reduced the number of vertices to a manageable level. aussi, un phenomenon incroyable!

once again Wulfie, thank you for your continued genius in support. bounding boxes for everyone!

[EnCore Mayne]

Wulfie, or anyone knowing a soupcon more than myself, i've run into what (i think) is a big problem and has grinded progress to a screeching halt.

in my illustration it shows axes square to the 2D page. in reality (Blender) that cyclindrical object (pivot point) is off the standard 90 degree axes. it's akilter on all axes. not sure exactly what numbers it's off from square but nonetheless, inlies my problem. i've used Object>Apply > Location & Rotation to the whole object after i got the extension to work and when it's inworld the Local rotation manipulator is aligned to the Global axes.

i would like it to be aligned to the odd angled moveable rotational part. now that i've Applied locrot to the 3D model everything's squared to the noble directions. must i start again and NOT apply location rotation before i bring it in or is there a way to realign the axes to the off kilter part???

argh.

[arton Rotaru]

11 hours ago, EnCore Mayne said:

Wulfie, or anyone knowing a soupcon more than myself, i've run into what (i think) is a big problem and has grinded progress to a screeching halt.

in my illustration it shows axes square to the 2D page. in reality (Blender) that cyclindrical object (pivot point) is off the standard 90 degree axes. it's akilter on all axes. not sure exactly what numbers it's off from square but nonetheless, inlies my problem. i've used Object>Apply > Location & Rotation to the whole object after i got the extension to work and when it's inworld the Local rotation manipulator is aligned to the Global axes.

i would like it to be aligned to the odd angled moveable rotational part. now that i've Applied locrot to the 3D model everything's squared to the noble directions. must i start again and NOT apply location rotation before i bring it in or is there a way to realign the axes to the off kilter part???

argh.

Eyeball it in orthographic view to rotate it until it's X axis aligned. Apply rot & scale again.

[EnCore Mayne]

woohoo! thanks arton. out of the dark now kinda. (i have too much to learn...).

[Rick Nightingale]

May I add one little trick to the 'extend the geometry' that I've used for years to make doors etc.?

When you put that extra vertex at the opposite corner to where you want the pivot (to extend the bounding box), instead of making it a triangle, do this:

When you've finished making the object, select that vertex you added, and go to the menu: Mesh/Sort Elements/Selected. Make sure Vertex is selected in the tool pop-up.

Now, that vertex is the very first vertex in the database. While it will not be rendered in SL because it is a single vertex and gets optimised out, it still gets used when setting the bounding box because the first vertex cannot just be gotten rid of. Or something like that. It works, anyway.

The benefits are there is not even a miniscule triangle to hopefully not see (or have to have a material for it so it can be made transparent) and it saves two vertices.

Extra: If actually making a door, you also need to extend the physics model to the bounding box size of course. For physics, you can't use the single vertex trick so my physics model does have a tiny triangle ending at the same location as that vertex above (just for simplicity - doesn't actually have to be in the same place, it's just that I copy that vertex into the physics model to get the position right for the physics triangle). The door's own physics is a simple cube stretched to fit the visible door size. When uploading, Use Analyze with method set to Solid. That will eliminate the tiny triangle from the physics completely but leave the door's physics intact as per your physics model. If my door has knobs that project and I don't need physics for those, I'll make that tiny, extra physics triangle long enough to extend the bounding box 'thickness' to account for the extent of the knobs too. again it gets optimized out of the physics, leaves the door's actual physics as I want while extending the physics bounding box appropriately.

Edited March 15, 2023 by Rick Nightingale

[EnCore Mayne]

a single pixel used for the extension to precisely place the pivot point (origin) does not work.
three vertices (without having a face) does not work.
3 verts with a face works.

and, for anyone beating their heads on their desks, if there's more than one pivot point, as my example has, don't reapply a rotation once any part has been set. each part requires reorientation in the 3D view window for the rotation to be applied to that specific part. live and learn....

[Rick Nightingale]

1 hour ago, EnCore Mayne said:

a single pixel used for the extension to precisely place the pivot point (origin) does not work.

Strange because I've been using that for years, and it most certainly works. I can show you the doors! Lots of them. Perhaps it needs to be uploaded with Firestorm... I never use the LL viewer so perhaps there is some different behaviour.

Which viewer do you upload mesh with?

Edited March 15, 2023 by Rick Nightingale

[EnCore Mayne]

maybe cause it's an attachment?

i'm using Firestorm. could be the uploader too. things were a bit wonky in the beta grid. or just my lack of concentration. i can try again but i'm fairly sure it wouldn't take without the triangle with a normal.

[Rick Nightingale]

Attachment should not matter.

Ignore all the following, I'm incompetent is all.

Interestingly, I was going to send you a proof .blend and .dae... but when I uploaded it, instead of putting the pivot on the hinge edge, it put it in the centre of the edge. That's... very odd. I've checked an original door made just the same and it pivots exactly on the hinge.

Just trying to figure out if I did anything different right now. Like I said though, I've been doing it just like this for years. I wonder if something changed, or I did something different this time.

The LL Viewer and FS both did the same, and they did at least put the pivot on the edge of the visible mesh. Just not right on the corner, but in the middle of that edge. Hmmm... watch this space.

 

Edited March 15, 2023 by Rick Nightingale

[Rick Nightingale]

An original door uploads and works with the pivot exactly on the door hinge, as I expected. That's with a single vertex at the opposite bounding box corner.

You can even see the pivot point in the mesh upload preview window if you rotate the preview. It rotates on the altered pivot point.

[Rick Nightingale]

LOL - Sorry, but blame the neighbours who are, once again, playing music next door so loud it's deafening in our house. I can't think straight.

I didn't put my vertex in the right place. Simple.

I'll PM you with a link to a blend file on my server in a minute. Just hide the phys model (because they overlap), edit the door, select Vertex selection mode and you'll see the extra, single vertex. Then you can check how I did the physics if you want although for an attachment that doesn't matter. Export it with SL presets and check it out. It works - pivot is right in the edge corner, where the hinge would be.

 

Edited March 15, 2023 by Rick Nightingale