Land impact? Are meshes just hoax with no use?

[Yan Hoxley]

Hey there, I've just tried to play with meshes .dae export, all works fine, mapping works and so on, but land impact value is way over what I thought it could be... See the example on picture. Upload setting was average with lowest phys. Source for .dae export was sculpted hut from past + UVW unwrap, not even vertice changed, so its direct mesh export from sculpt objects (I expected same land impact or doubled, not 1800% higher).

Every mesh i tried to import had acceptable land impact, only when it had small size... so its kinda useless, worth only for simple mini objects like chair, bucket or anything small with detailed texture mapping.. thats it.

Simply you cant have any mesh but small, to stay below 15 - 20 land impact.
Well for me... I would be more happy, if I could upload or save custom uvw map for sculptie, rather than all land-expensive meshes...

[Imnotgoing Sideways]

You're doing it wrong. ()

 

 

[Dilbert Dilweg]

I Kind of wonder the same thing.

I made a pergola and first off I made the collums/pillars as low poly as possible (6 total). Standalone they are a land impact of 4 set as a prim. The beams overhead are a land impact of 1.5 (2 set as prim) 14 total.  Uploading the pergola as a mesh and settings the physics shape to high. The best I can do is get it to 73 prims/land imnpact. . that is without a material. With materials 163 lol wtf

If I upload the parts by them selves i get a land impact of 44 for the whole thing assemebled in world. And when linked together the land impact does not change if I resize it..

It just seems like a lot of work for such a borked system. What is the use of having to build and assemelbe in a mesh program when it is better to just upload parts and assemble in world

[Drongle McMahon]

i'm afraid that is the effect if you use sculpty meshes as regular mesh. It reflects the true inefficiency of sculpties. Nine sculpties is more than 18000 triangles, which is way too much for this sort of structure. For an object this size, the low LOD mesh will be the most important in determining the download weight. It needs to be made with alpha planes for the railings which will hugelu reduce the poly count. Without seeing the lowest LOD you made the physics from, and knowing whether you clicked "Analyse", I can only guess whether the physics weight or the download weight is the cause of the high LI. What does it say if you use the More Info link? If it's the physics weight, which seems likely, then a very simple shape needs to be made for this. That is best for sparing the physics engine, anyway.

ETA - the answer to you question is "No".

[Yan Hoxley]

Object has 8900 triangles (even floor is from planks)... in more info it says display weight is highest...  Hut is very simple, I expected this object with this simplicity level, will have land impact about 20

anyway so you say alpha railings as solutions... so why I use meshes, if I need make railings as alpha, that hut is all about 3d railings and dropped shadows duh...how can I get some realism with alpha ralings...

it looks I will stay stick with sculpties, coz when I will create Hut with 9 prims it will forever stay 9 prims, whatever will happen not depends on size, scripts inside and so on.... and with care it gives good texture bake results

Anyway I did small simple wooden boat and it has 25 LI, funny, I can say Im able to do it from one sculpt and it will look almost same... I did nice mesh palm, again nothing, result after making it RL size of palm = 177LI for 2 palms with coconuts (7 LI as small object)

In short.... LI stays "normal" and acceptable, until its small... gazebo not works... result is still.. meshes are quite useless coz of land cost and I dont do clothes or wearable objects where nobody cares of LI

If this gazebo hut would be made from regular prims it would has around 120 prims (counted it b4 while), if gazebo is transformed into double size than on picture, result is 270 LI

still cant see other advance of meshes, than uvw mapping... my builds are based on lowprim from my first created object in SL, I cant just put up with mesh behaving like this and double support with quiestions about prim count, primcount after unlink, primcount after resize, for me meshes looks as quick trick job after V2 fail (after V2 release I stopped to log into SL for quarter of year, then I installed Phoenix/Firestorm to start SL-live again)

[Drongle McMahon]

The alpha railings would only be used for the low LOD. That achieves the low LI while keeping the full detail at distances less than the second LOD switch, which should be about 100m for this if I guess the size about right. Here is an example. The top two are closeups; the bottom two are the LODs where first the crossbars become alpha, and then the whole frame becomes alpha, at the distances they switch in low graphics setting (RenderVolumeLODFactor = 1.25). At higher settings the changes are when it is much smaller. The LI of this is 6, at 7.5 x 7.5 x 6m.

It is certainly true that it takes more work to make optimised mesh with low LI, but it is possible. The same is true for making LOD-proof sculpties. What does the sculpty version of the gazebo looks like after two LOD steps? Of course it's entirely your choice whether you think it's worth the effort. Mesh is not compulsory, but it isn't a hoax either.

You can see more examples in the windows and skylights of the gallery in MeshHQ 3 (Aditi). The building as a whole is not very well optimised though (250 LI). It was made before the LI accounting was finalised. It should be possible to get it substantially lower. The monolithic roof is the worst part.

[Dilbert Dilweg]

Mesh just doesnt seem to prim friendly. Nice mesh comes at a high prim cost

[Kwakkelde Kwak]

You are simply approaching your builds the wrong way, Drongle explained quite a bit on that already.

You cannot turn a sculpty or prim build into mesh and expect good results.

I suspect you made it a single mesh, which might or might not be the most efficient way in terms of land impact. There are some things to take in mind.

Size matters, if you make a big mesh, like your gazebo, the lowest LoD will not pop for a very long distance (Drongle made the calculations on it, I'm sure he can link the threads that cover that). This means you can use a very very simple model for the lowest LoD, possibly even no model at all. I've built some things like that and I can honestly say the LI is a lot lower than when it would be built with normal prims, as much as a factor 10. For the models for LoD med and low you could use the alpha planes Drongle mentioned, you shouldn't have any problems keeping the number of triangles under 100 doing that without screwing up the looks of the build.

The other way to do it is by splitting your model into pieces. This can have a small advantage since duplicate meshes are streamed only once. The LI will go down when you use smaller pieces, but the LoD switches will be a lot sooner (closer), with two disadvantages. First of all, not all LoD switches will happen at the same time, making the switches more obvious. Secondly you will need to have more detailed models for the lower LoDs. There is a limit in number of pieces used aswell, since every piece will have a landimpact of at least 0.5.

All in all you need to figure out a good workflow. A lot of people including myself have rebuilt sculpty models in mesh for illustration purposes on these forums. The original creators had the exact same complaints about the landimpact but were convinced about the benefits of mesh when they saw the meshes were not only better looking, but also had a lower landimpact in the end.

Finally, like Drongle, I'm curious which weight is responsible for the high landimpact of your gazebo, download or physics? And what exactly are the numbers for those two?

[Drongle McMahon]

"(Drongle made the calculations on it, I'm sure he can link the threads that cover that)"

Details are in this thread. Too mathematical for some. Others have given more easily digested explanations.

[Masami Kuramoto]

---

Imnotgoing Sideways wrote:

You're doing it wrong. (

)

---

Linden Lab is doing it wrong. And the OP has the numbers to prove it.

As long as the land impact formula exists side by side with traditional prim accounting, people will look at those numbers and say: "Screw mesh, I'm doing it with sculpties!"

Mesh creators have to bend over backwards optimizing their geometry and still find their land impact in the same ballpark with a prim build. This is bad policy because it fails to reward the efficiency that is needed to reduce lag in SL.

[Dilbert Dilweg]

---

Masami Kuramoto wrote:

---

Imnotgoing Sideways wrote:

You're doing it wrong. (

)

---

Linden Lab is doing it wrong. And the OP has the numbers to prove it.

As long as the land impact formula exists side by side with traditional prim accounting, people will look at those numbers and say: "Screw mesh, I'm doing it with sculpties!"

Mesh creators have to bend over backwards optimizing their geometry and still find their land impact in the same ballpark with a prim build. This is bad policy because it fails to reward the efficiency that is needed to reduce lag in SL.

---

Exactly my thoughts. It is not an easy simple task to optimize efficiently and completely useless to model a model and have to upload parts and assemble in world.  I am at the point of saying screw mesh for building purposes. Why would I want to give up double the amount of prims just for something to look slightly better,,, If mesh is supposed to be more efficient then the limits should be raised.

[Kwakkelde Kwak]

---

Masami Kuramoto wrote:

---

Imnotgoing Sideways wrote:

You're doing it wrong. (

)

---

Linden Lab is doing it wrong. And the OP has the numbers to prove it.

As long as the land impact formula exists side by side with traditional prim accounting, people will look at those numbers and say: "Screw mesh, I'm doing it with sculpties!"

Mesh creators have to bend over backwards optimizing their geometry and still find their land impact in the same ballpark with a prim build. This is bad policy because it fails to reward the efficiency that is needed to reduce lag in SL.

---

I won't argue where it comes to the backbending, but the numbers you address mean squat. These are numbers of someone who doesn't know how to build in mesh, you could blame LL only for not providing sufficient information or tutorials. Nothing is wrong with the way mesh works and nothing is wrong with keeping the old way of counting intact for sculpties.

Same ballpark as primbuilds? If those are the results you get, I wonder how you build your meshes. As I said earlier, my experience is that meshes result in a lower landimpact than sculpties most of the time and lower than normal prims up to a factor 10. That would have to be one hell of a ballpark.

A problem is that most people in SL didn't know a thing about 3d modelling until they started building on the grid. Then they converted to sculpties. These two ways are substantially different from building in mesh, I'm sure you can second that since I'm sure you modelled before you came to SL.

Primbuilding or sculptbuilding doesn't translate to meshbuilding, people getting the horrible results are most likely trying exactly that.

[Dilbert Dilweg]

I have to disagree but then again I am, not a modeling theologian either. I have never messed with Sculpties and learning to build mesh low poly. And find it pain stakingly hard to upload a decent optimized mesh,

Anyone building mesh is mostly better off making avatar attachments where LI is not a big issue

[Kwakkelde Kwak]

It does take a lot of practice and some "feeling".

When building a "larger" mesh, one could use Drongles numbers to predict things, but that's not what I do and I suspect it's not even what he does himself. I could be wrong on that ofcourse.

Anyway, I usually build my highest LoD, the way I want it to look, without holding back on detail, but while trying to keep the geometry to a minimum. Some parts of that require a good sense of 3d space, some parts are just common sense. For example, never use hidden geometry. Another example, no vertices on straight edges. Both do not enhance the shape in any way, but they do increase polycount. Cylinders based on the SL prim? Often you can get away with 5 or 6 sides without it being noticable, this depends on the dimensions ofcourse. A rope is very convincing with 5 sides, a windmill supposed to be round needs more ofcourse.

That's a big advantage of mesh btw, you can set the number of sides to anything you need, not the fixed 24 for SL prims or the fixed 4, 8, 16, 32, 64, 128 or 256 sides of sculpts.

Once the highest LoD is done, I upload on the betagrid. Then I zoom out with the "rendervolumeLODfactor" set to default. It will become quite clear at what distance the LoD switches, or better put, it will be quite clear how small the object appears on screen. Using that view, one can see where geometry can be saved, for example, a small sphere seen from 50 meters away doesn't need to be 24x12 (If you take the prim sphere as reference) for a convincing look, something like 12x6 or 12x12 or even 4x3 will look just as good, depending on the size. If the object isn't as 3d as a sphere, but almost flat like a wall, window, gate, banister, walkway etc, you can probably get away with a flat plane with dedicated texture. Those are 2 triangles or 4 when two sided. With the object rezzed on the betagrid, you will also see when the lowest LoD becomes visible, or not. If that switches at let's say 200 meters, my personal opinion is you can probably do without that model altogether, ofcourse for some objects like landscaping ones, that probably isn't an option.

Then there's something similair to the above, but not quite the same. Interiors can only be seen from...the interior. So it usually doesn't matter if it can't be seen from distances as short as 10 meters. Scrap the lower LoDs and see the landimpact of your favorite chair or table go down by a factor 10. Maeve has made good use of this a while ago and regularly posted on the forums about it, I don't think she ever posted her endresults though.

btw, ever tried to make a hole in a sculpt? I know I have and it is very possible, but it's very counter-intuitive and doesn't result in something I prefer over a mesh.

I politely disagree on the avatar attachments

Most visual lag on the grid is because of avatar attachments, I am convinced it happens quite often that a single avatar is good for more geometry on screen than the surroundings combined, I even think that's an understatement. Months ago we (forumusers) did the math on it and I think the whopping theoretical number of triangles that can be attached to a single avatar was 1 or 2 billion. If there is any place where mesh can be nothing short of a disaster it's in avatar attachments. Open the "avatar render cost" option and start looking for numbers over 400,000, 10 times more than the recommended number by LL. I'm sure you won't have to look very long. 40,000 sounds low, but with mesh it's not that hard to stay aroiund 20,000.

[Yan Hoxley]

for the record, I was working in 3ds max years before I firstly logged into SL... so its not about noob 3D mesh modeler.

Some ppl here seem sticked with my expample picture.. yes its converted from sculpty... I know now that seem to be wrong way... but I tried to model even simple meshes starts from draft, lowpoly as possible, and results wasnt good either...

I will also try method to upload piece by piece form items and then reassemble in SL and link together, if it will help.. as someone said here...

Im thinking about to start make small 3D games for android or PC in some 3d engine to finally have some fun with 3D. Secondlife modelling fun is hardly limited... Imagine how much time took custom uvw mapping of railings and planks on floor and textures render of my gazebo just for nothing.

 

SL meshes are poor from aspect, that you can make complex sculpt + invisible prim boundary (if needed) and it will have 2 prims, even when is 1024x1024m big (or 64m when you resize)... meshes cant compete with sculpts until metric size of meshes will not affect LI

You can even build same item from regular prims, and at some metric size, it will start to beat the mesh as the mesh LI rapidly grows with size....

 

This all kicked my effort to finally do meshes as is native for me.. I still have to just bend and push'n'move vertices of prefabbed sculpt shape...

Imagine my 10prim sculpted waterfall... I did 40x20x8m sculpty waterfall wall: It has rock shape + plants + 4 prims with particle scripts + 2 prims with texture animation and sound control menu script + scripted animated water foam + offsim prim (in case someone wants to place it offworld) .. it looks kinda realistic now..

But what if I would like to add realistic boulders, better rock shape and so on and put all the scripts... Am I able to do the same from mesh, with same or almost same LI, with same metric size? NO I am not.

[Kwakkelde Kwak]

You seem to ignore every reference I make on meshes that are better in landimpact than the equivalent sculpt or prim build. As I wrote in my previous post, the trick is not in the high LoD, i never hold back on that myself.

The trick is to set your LoD models up in such a way that you don't need so much geometry. The auto-LoD used by SL is a nice shortcut but usually gives poor results in both looks and number of faces/vertices. The gazebo you built doesn't need more than 100 triangles at LoD low and no more than 50 at LoD lowest by my estimation.

Again, I'm curious what causes the high number, is it download or physics weight? If it is the physics, it's really simple to get the number down and shouldn't take you more than five minutes.

[Masami Kuramoto]

---

Kwakkelde Kwak wrote:

These are numbers of someone who doesn't know how to build in mesh

---

Sure, but why should he bother? Where is the incentive, as long as the metrics favour the large prim build by a factor of 32 and he doesn't even need to worry about LOD?

Try to explain that to him.

[Kwakkelde Kwak]

---

Masami Kuramoto wrote:

---

Kwakkelde Kwak wrote:

These are numbers of someone who doesn't know how to build in mesh

---

Sure, but why should he bother? Where is the incentive, as long as the metrics favour the large prim build by a factor of 32 and he doesn't even need to worry about LOD?

Try to explain that to him.

---

Explain to me why my 64 meter big building in mesh is ten times less in landimpact than the equivalent prim build. I have no clue how you come up with a factor 32 the other way. Building so big need only one model, the highest LoD. The rest can be 3x the number of materials used for vertice count, so LoD med, low and lowest have a maximum of 24 vertices and a maximum of 8 triangles.

[Drongle McMahon]

meshes cant compete with sculpts until metric size of meshes will not affect LI

Or until sculpty sizes do affect LI! (...which is perhaps what they should have done for sculpts > 10x10, unsupported before mesh). The problem is not that mesh are costed too high and according to size, but that sculpties (and spheres and toruses etc.) are costed too low.

The attachment problem is already a disater with sculpties. I have seen shoes with 60 sculpties in each. That is up to 240,000 triangles for the pair (when you are close). Jewellery can be even worse. My own hair used to have 200 twisted toruses, with more than 2k triangles each. Thats nearly 500,000 triangles, and many of them were flexi too! The failure to limit attachment complexity is surely a major problem, with or without mesh.

[Kwakkelde Kwak]

---

Drongle McMahon wrote:

I have seen shoes with 60 sculpties in each.

---

I've seen much worse. Both in mesh and sculpt btw... unfortunately.

[Pamela Galli]

I am saving LI on furniture, for the most part, but not always -- for houses, not so much.  I spend a lot of time "optimizing" to get the LI down, and the results are really not that satisfactory.  I often have to choose between good LOD or good LI.  The advantage for me is that I can put the door or window where I want it and make it any size. So on balance I am happy with mesh but it is sadled with disadvantages that prims and sculpts just do not have.

 

[Yan Hoxley]

Ive seen your reference'bout it, but everyone could say that... Post some landmark or give us picture + LI of your large model you talking about... I want to see that

[Masami Kuramoto]

---

Kwakkelde Kwak wrote:

I have no clue how you come up with a factor 32 the other way.

---

I can see that factor right there in the OP's screenshot. Please explain why that mesh build with 8900 triangles deserves to be 32 times more expensive than a multi-prim build of identical size and triangle count. This is so obviously wrong, I don't even know what there is to argue about.

The land impact formula is fine, but it needs to be applied equally to prims, even at the risk of breaking some content, otherwise people will not do what is necessary to bring lag down and increase framerates. People shouldn't be making sculpties in 2012, but they still do, because the trouble of building low-poly meshes with multiple LODs does not get rewarded the way it should.

[Masami Kuramoto]

---

Drongle McMahon wrote:

The problem is not that mesh are costed too high and according to size, but that sculpties (and spheres and toruses etc.) are costed too low.

---

Exactly my point.

[Dilbert Dilweg]

Okay I think I am starting to get it now. I rebuilt my pergola and used my high poly pillars. Not massivley high but double the verts and faces than the one I was uploading before. Then set up my LOds like instructed and really impressive.I think I impressed my self there for a second lol. I got the LI down to 20 where as it would have normally been about 120 lol. I set up a cheesy 6 cylandar physics file with not much to it but the basic shape needed.

One question tho. What causes the Land Impact to change when setting the object settings from "convexhull" to "prim"?

Some models I upload the LI stays them same when I change to prim, and some increase the LI.