Mesh and sizes

Astrid Kaufmat · November 25, 2012

Hello I have a question about the size of meshes.

I am building some stuff where I'd love to add some little belts.The whole structure worked for me and for resizing but some parts of it , if to small get deformed.

I'll be specific. What is bugging me is a little fibula for a belt.

The size should be not bigger than 1,5 or 2cm but if I do this in maya, in blender and then upload it it gets deformed, say "fatter" on the X axis. To let it show as it should I have to make it bigger.

To try to fix this problem I did some tries:

1) I was surer to lock the model before exporting it using layers too.

2)I tried with more polygons and less (nothing changed)

3)I've tried to resize it in world with a script but nothing changed as well

4) I checked again all the polygons to see if there was some bad polygons or edges shared by more than two polygons and there was none.

After all this I thought ok maybe it's the shape that might be not rendered good. So I changed shape and procedure to make that belt. This one had as start a cylinder I made others from a torus and from a cube too but the result wouldn't change I get the same iussue when the size gets small (2cm ) like the one I'd need.

I'll attach here a link to see the file I am talk about if you need so you can have more infos and see better what I am talking about. Here is the link

http://depositfiles.com/files/fk94b4p5k

Chosen Few · November 25, 2012

As far as size goes, mesh objects are no different from any other object in SL, in that they can't be smaller than 1cm in any dimension. If your little object is smaller than that, it's going to get upsized. To solve the problem, structure the mesh differently, so that it's not smaller than 1cm on any axis.

I did not download your file, by the way. The hosting site you chose doesn't say what it is. It doesn't even provide the file name. It's too much of a risk to download blind. I'm sure you can understand that.

Kwakkelde Kwak · November 25, 2012

Top of the page Chosen, it says belt12.dae.

@OP What should be the size of your belt? The object you posted is at least 100 times too big. If it is exactly 100 times too big, the size is 11.5 x 12.2 x 2.4 cm. That's a very big buckle, but if that's the size you wanted, for me it seems to upload just fine. I had to set the scale in the uploader to 0.01.

If you want your buckle to be smaller, as Chosen said, the minimum size is 1 cm. So you can scale down to 4.8 x 5.1 x 1 cm. If you want it smaller, fake the z value by adding some geometry or attach the buckle to another object before uploading.

Chosen Few · November 25, 2012

Kwakkelde Kwak wrote:

Top of the page Chosen, it says belt12.dae.

Thanks, Kwak. Actually, it didn't say it, the first time I went to the site. Apparently NoScript blocked it. It didn't show up until I disabled script blocking for the site. Interesting. I don't usually need to do that, just to see file names on file sharing sites. This particular site is not one I've ever used before. I don't like to let unfamiliar sites run scripts on my machine, whenever I can avoid it.

Kwakkelde Kwak · November 25, 2012

Fair enough, it never hurts to be careful...

Aquila Kytori · November 25, 2012

hi

While i was away making pretty pictures chosen and Quakkedle were answering , never mind i'll post anyways :

Kwakkelde Kwak · November 25, 2012

That's a good way to do it.

I do wonder if you really need it to be as detailed as you have it now. The smallest faces are smaller than 1 mm and the triangle count is 484. For an object that small that seems a bit over the top if you ask me.

Astrid Kaufmat · November 25, 2012

Great guys :catvery-happy: thanks for the answers I tried both ways linking it before uploding to the whole structure and it worked good.

The faking Z axis did it's job too hehehe that's neat trick,thanks.:catvery-happy: :catvery-happy: :catvery-happy: :catvery-happy:

Chosen Few · November 25, 2012

Kwakkelde Kwak wrote:

I do wonder if you really need it to be as detailed as you have it now. The smallest faces are smaller than 1 mm and the triangle count is 484. For an object that small that seems a bit over the top if you ask me.

I couldn't agree more.

For one thing, the back of the model doesn't need to exist. If all the buckle is going to do is sit flat against the belt, no one will ever see the back. Along those same lines, the cylindrical portion doesn't need to be a full cylinder. A half cylinder would work just fine.

Also, the cylinder doesn't need to have nearly as many sections as it has, just to read as round on-screen. Right now, it's got 16 sections. That could be lowered to 8, or even to just 6, and it would still seem round from most viewing angles, as long as the normals are soft.

There's a little indentation, where the cylindrical part meets the rectangular part, on both ends. That would be more efficiently suggested with texturing than with actual geometry.

Finally, there are quite a few places where there are unecessary quads could either be reduced to tris, or eliminated altogether.

With the above changes, I was able to get it down to 110 tris instead of 484, in just a few seconds, and it looks pretty much the same. If I wanted to spend a few more minutes, I could likely get it even lower. From the distance at which people typically view a belt buckle no one would ever know the difference.

ETA: I forgot about the added cube. It doesn't need to be a whole cube. It could be just a single triangle. So, make that 111 tris.

Astrid Kaufmat · November 25, 2012

Chosen Few wrote:

Kwakkelde Kwak wrote:

I do wonder if you really need it to be as detailed as you have it now. The smallest faces are smaller than 1 mm and the triangle count is 484. For an object that small that seems a bit over the top if you ask me.

I couldn't agree more.

For one thing, the back of the model doesn't need to exist. If all the buckle is going to do is sit flat against the belt, no one will ever see the back. Along those same lines, the cylindrical portion doesn't need to be a full cylinder. A half cylinder would work just fine.

Also, the cylinder doesn't need to have nearly as many sections as it has, just to read as round on-screen. Right now, it's got 16 sections. That could be lowered to 8, or even to just 6, and it would still seem round from most viewing angles, as long as the normals are soft.

There's a little indentation, where the cylindrical part meets the rectangular part, on both ends. That would be more efficiently suggested with texturing than with actual geometry.

Finally, there are quite a few places where there are unecessary quads could either be reduced to tris, or eliminated altogether.

With the above changes, I was able to get it down to 110 tris instead of 484, in just a few seconds, and it looks pretty much the same. If I wanted to spend a few more minutes, I could likely get it even lower. From the distance at which people typically view a belt buckle no one would ever know the difference.

ETA: I forgot about the added cube. It doesn't need to be a whole cube. It could be just a single triangle. So, make that 111 tris.

Well Thanks for the help Chosen few .

BTW if you would use a bit more time reading the iussue you could see why I used that big subdivision. As I said before I used many tries some with lowest subdivisions and some with higest . The prim I put there was just one with the highest number of polygons.

My question was pertinent to size not about the weight and numbers if polygons.

The meaning was easy translated into other words was: why a mesh even if so much subdivided could get deformed?

I leave you with my daily thought.

Mankind is a beautiful thing, the only living beings who have a spoken language, while animals haven't got a spoken language. Though even if animals haven't a spoken language, but a semantic language; this wouldn't mean that they are inferior than human beings .Animals would never kill other living beings for the sake of it and would never kill their similar just for the sake of it. PEACE and LOVE.

Chosen Few · November 26, 2012

Astrid Kaufmat wrote:

BTW if you would use a bit more time reading the iussue you could see why I used that big subdivision. As I said before I used

many tries

some with

lowest subdivisions

and some with

higest

. The prim I put there was just one with the highest number of polygons.

I assume you're talking about your brief mention of "tried with more polygons and less." I'm not sure how you expected that that would automatically inform me that the model you posted was the "more" version. It could just as easily have been the "less". You didn't specify.

In any case, I'm not sure why you might have thought that simply adding more subdivisions would in any way affect SL's sizing requirements for models. No matte hor many times you subdivide a model, it remains exactly the same size.

Also, the model you posted did not appear to have been subdivided at all. All appearances were that every vertex had been placed deliberately. None appeared to have been generated as the mere result of splitting larger quads into collections of smaller ones.

Perhaps your definition of "subdivision" in this context is something different from the standard terminology?

Astrid Kaufmat wrote:

My question was pertinent to size not about the weight and numbers if polygons.

Well, to be fair, even though the answer was about size, the question itself was not. There was nothing in your original post to indicate that you already knew that size was in fact the issue. All signs were you'd been throwing everything you could think of at the model, to try to get it to behave the way you wanted, with size being just one possibility among many. Of the four things you listed as potential solutions you'd tried, only one of them had anything to do with size, after all.

Obviously, if you had already known that the issue was not about the number of polygons, you would not have included "tried with more polygons and with less" in your list. You can't open the box on something like that, and then expect people not to talk about it.

That said, since I knew the answer had everything to do with size, and nothing directly to do with poly counts, I deliberately waited until AFTER the size question had been fully answered, and until AFTER you'd acknowledged your understanding of the answer, before I posted anything about reducing the number of polygons. There's nothing wrong with taking a discussion in a different direction after the original question has been answered.

If you recall, I was the first to inform you that size was the actual problem. Others then added to that, by providing more pointed detail on ways to solve the problem. You thanked us all, and said you understood.

I was not the first one to start talking about your model's poly count. I added to that discussion, by providing some detail on ways to reduce the number of polygons, in the same way that other people had added to the discussion of size. The progression of information happened the same way on both topics (as is usually the case), even though the order of participants was different.

One thing you should always keep in mind when coming to this, or any other, public forum to post a question is that the answers you receive will not always be directed at you personally. Because it's a public forum, the answers are intended for public viewing. Therefore, whenever any of us sees an opportunity to help inform the public on an important topic, we should always take it. I do that whenever I can, and so do most of the other people who frequently participate. It's how we all learn from each other, and it's a great thing.

That's exactly what happened here. When I suggested ways to reduce your model's poly count, I wasn't scolding you for having created it the way you did. I was trying to teach every reader how to examine a model, spot areas that can be reduced, and then do it. If you don't choose to change your belt buckle, it's no skin off my nose. The 300 or so extra polygons in that little buckle aren't about to make or break anyone's frame rate. But a certain percentage of the peole reading the post will end up applying the same principles to their own models, and the people who learn from them will do likewise, and that very likely will end up making a big difference, to the benefit of us all.

The rampant lack of optimization of poly counts is one of the biggest problems SL faces right now. Until and unless people learn to use as few polys as possible in every single model they create, SL will never run as smoothly as it could. For everyone's benefit, it's important that we continuously talk about the subject.

We've done exactly that with other, equally important topics, and it's made a tremendous difference. Texture size is a good example. While it continues to be a gigantic problem, it's nowhere near as bad as it was before we in the forum community began making a point of talking about it as often a we could.

Very often, the "common knowledge" that permeates SL begins right here on the forums. I'm not about to stop doing my part to hep fan those flames, and neither should you.

Astrid Kaufmat wrote:

The meaning was easy translated into other words was: why a mesh even if so much subdivided could get deformed?

Because the number of subdivisions is not the reason for the deformation. The deformation happened because your model was too small along one axis, exactly as I surmised in my first reply.

I'm not sure I understand what you're trying to say, with this rephrasing of the question. A moment ago, you said your question was about size, not about poly counts. Now you seem to be saying poly counts were in fact part of the question. You do understand that subdivision counts and polygon counts are ultimately the same thing, right?

Astrid Kaufmat wrote:

I leave you with my daily thought.

Mankind is a beautiful thing, the only living beings who have a spoken language, while animals haven't got a spoken language. Though even if animals haven't a spoken language, but a semantic language; this wouldn't mean that they are inferior than human beings .Animals would never kill other living beings for the sake of it and would never kill their similar just for the sake of it. PEACE and LOVE.

That's a nice thought, although I'm not sure how it's relevant to anything we've been talking about.

As much as I hate to have to burst your bubble by letting pesky things like facts get in the way, animals do have spoken language and animals do kill just for killing's sake. It's the epitome of human arrogance to assume that just because we don't understand the language an animal is speaking it must not be speaking at all. And it's only through romanticism, self loathing, and wishful thinking, that we assume we're the only ones that kill for no good reason.

It's been well acknowledged by the scientific community for many decades now that dolphins, whales, and other marine mammals have their own spoken languages. So do birds. (I have four birds in my house. They talk to each other all day long.) Insects have chemical languages, which are at least as complex, if not more complex, than our auditory language. The "speak" via chemical signals. Squids communicate with each other by flashing patterns of color and light across their bodies. Their "speech organs" are the chromataphores in their skin. Humboldts, in particular, are thought to have an incredibly complex language, consisting of thousands of different patterns. The list goes on and on and on.

As for animals that kill for no apparent reason, that's quite a lengthy list, as well. Lions kill other lions all the time, and don't eat them. Hyenas kill all kinds of other creatures, for no discernible reason, other than the apparent enjoyment of the kill. Ants routinely carry out entire wars, and even take slaves. Orcas will often kill other whale species, for no apparent reason. Dolphins brutally murder porpoises, and on occasion, other dolphins. The fossil record strongly indicates that mososaurs killed other mososaurs as a matter of course, but did not eat them. Heck, even the two dogs I grew up with used to kill squirrels and chipmunks all the time, just so they could play with the dead bodies as toys. (It was awful to witness.)

No, we humans really aren't very different from those with whom we share this planet. We are animals, just like any other, even if we sometimes prefer not to admit it.

Astrid Kaufmat · November 26, 2012