The Mesh Holiday Home Test

Today, having a little time on my hands between projects I sat down and decided to look at a side element of mesh that I hadn’t really explored – On sim mesh, in particular for housing.

Setting up the scenario. I have always loved building my own home in SL – its a nice little break from the day to day operations, and right at this moment I have my 3^rd home in SL that I built myself.

Its construction is a hybrid of prim and sculpt.

Some time back I managed to split off the sculpt sections of the house and export the prim elements of it in to Blender. Today, I sat down and started to look at how I could recreate it all in mesh.

This is of course a different game from my usual work. Rather than detail, houses tend to be more about keeping things as simple as you can.

The fewer vertices you have, the better.

And now an intermission where I will play some jaunty music...

Four hours later, I am sitting down, with a part built system. It looks good and its totally optimised.

First issue I can see immediately is the wonderful physics elements. This is a new area of the mesh upload for me – don’t usually use it however inworld objects need to be physics objects so collisions can occur as expected.

First problem. Without physics, a basic upload cost is 75L$ however with physics on and set reasonably, its jumped to a staggering 350L$ to upload...

Can I say ouch...

Lesson 1 – Uploading physics enabled mesh is vastly more difficult (as the viewer packed in during early trails), and seems to cost more.

Next thing I found is partly an explanation as to why mesh houses that are on the market at the moment all look so similar and I will come back to that in a moment.

The thing I spotted first up, was the strange in world cost of the objects. In fact, they seemed to be just as heavy as if I had used regular prims. This actually surprised me, as they shouldn’t have been – I would have guessed they would have been lighter on the prim cost.

During my design process I was frugal over what I was using – if a wall need 4 vertices, it got 4 vertices. Nothing more.

It took a little while to figure out why and where all these costs were coming from, after I stripped back everything to see what was heavy, and what wasn’t.

A wall is a simple object really. 4 vertices, 1 face. Even a complicated wall with 5 angles isn’t that complicated – it is just 16 vertices, and 5 faces.

Obviously at heart a “house” is nothing more than a box with a roof... unless you are me. At which point the house suddenly becomes a slightly organic design.

I put in lots of nice curves, elements that give depth and detail. Except right there, is the issue.

Curves are not vertices friendly.

In my house, I have a nice little arch system – its a bit 1970’s but I like it. This is what it looks like

Now, lets talk about this.

In world, it is exactly 8 primitive objects.

Conversion to mesh took the 8 objects, cleaned them totally removing all unnecessary vertices, merged the objects together in to 1 object. Even when cleaned up, its not exactly massive really but the curves do add to the mesh costs. Total vertices is 380.

There is though a setting you can use before upload – either solid, or smoothed.

For it to look nice and chiselled, its best to use solid – total mesh costs on sim – 12.

Now, you can get better results by setting smooth, however when you do set smooth, it makes it loosened and less defined. Even with smooth on, the mesh cost on sim is 8.

So... same design, same look, and apparently the same cost – 8 with prims, and 8 with mesh (or 12 if you don’t smooth which makes it more costly).

And god forbid you try to link it to the house. I linked the mesh part to the house.

152 prims unlinked.

420 prims when you link an 8 prim mesh object to it

And yes, I checked that four times to make sure I wasn’t having problems reading it.

But, is this 100% right?

Well, I checked another area of plain walls in my hall.

What these are is a 5 prim element (specifically the back wall).

The mesh object does look nicer and smoother but once again, there is a problem. It is, with physics (which is required for that as its requires a collision area), 6 prims on sim.

So even a 20 vertices object, with physics, is slightly worse. I am sure if I smoothed it, it would be much less, but the fact of the matter is I am seeing no distinct advantage to using mesh over using prims here.

In fact all I am seeing from the various areas of the test is this – slightly improved design, slightly more in world cost all for the 4 hours I put in to remaking this.

This brings me back to the current available mesh houses that are out there today and I don’t mean this as a comment on their quality (which is very good).

The simple fact is the designers of these houses as I now understand, have had their hand tied behind their back. These houses have to be simple, slab sided object. Boxes effectively and those boxes are what they are. They do look good, but the real creativity behind designing houses in mesh is limited by the fact they are far too costly to produce for customers.

Their in world costs tend to be far more than if you were using prim or sculpties.

So yet again, we have mesh failing epically to live up to expectations from a designer and customers point of view.

Yes, mesh does again have benefits when it comes to design, but the drawbacks seem to be very problematic and it does seem to be 100% down to how mesh was implemented in world.

I am going to perceiver on that front and do some more tests but the purpose of this test was to establish a baseline of what mesh can do for structures and buildings in Second Life, and the result seem to be somewhat underwhelming.

At some point, I may get tired of say that...