Jump to content

You don't know what you're missing.


Nacy Nightfire
 Share

You are about to reply to a thread that has been inactive for 3605 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

Hehe, it's more than I would use if I would ever make rigged clothes.....

Ok, tried something and yes, you can wear objects over 255 prim equivalents, I forgot this was based on the link limit of normal prims..duh... So eerm, yes if people want to push it, they can go insane with meshes....

We don't have 30 attachment points anylonger, I think I got 80 things on before the wear option was greyed out. (max is 8 per single point, but there seems to be another limit)

Linkset is ofcourse still 255 max

Maximum vertice count is 65,536, if that's a closed cylinder that's 131,070 faces

A custom avatar could be 65,536 verts aswell, so another 131,070 faces

So in theory one avatar could be 80 x 255 x 131,070 + 131,070 faces = 2,673,959,070 faces

I'm not 100% sure about the number of attachment points...

Link to comment
Share on other sites

Actually I have seen it.  I check in once in awhile with a mesh viewer just to make sure I am not missing anything.  But, so far at least for wearable items I have seen nothing exciting enough to be worth using the mesh viewer..the viewer is beyond clunky and the mesh I have actually tried on has fit issues.

Link to comment
Share on other sites

I guess after 4 years here I've moved beyond caring much about wearable items.  Focusing too much on ones own avatar and appearance can be  pretty limiting.  There's greater world of SL then simply clothing. And  I will say that despite my not getting around to shopping much these days in SL, I HAVE seen some knock-out clothing and footwear items made in mesh.  More will come along as folks continue to come up the learning curve and add mesh items to their inventory.

Despite your perception that Mesh doesn't live up to expectations that it improves your wearable items, It might then be the case for you that the only advantage to mesh is the improvement of the general overall visual environment and the enjoyment of viewing all kinds of interesting complete avatars put together buy some incredibly talented people, and not improve the pixel paraphernalia you purchase to put on your avie.  Even with that limitation, it's pretty great. I'm using the beta viewer and I like it alot, there are a number of viewers that now view mesh so you have your pick.

Link to comment
Share on other sites


Kwakkelde Kwak wrote:

So in theory one avatar could be 80 x 255 x 131,070 + 131,070 faces =
2,673,959,070 faces

Haha, 2.7 billion faces!  Wouldn't that be something? 

That's roughly 33.4 million per linkset.  If we went one attachment at a time, I wonder how many we could attach before the viewer just up and decides, "You know what, I'm just not doing this anymore.  Say goodbye to me, your graphics driver, and whatever other GPU-dependant applications you may have running.  Good luck with the repairs.  Peace out." 

My money's on one attachment.  Anyone care to bet on two?

 

It actually could be even worse than those numbers, though.  Here are the maximum counts I get, for various primitive shapes, with 65,536 verts:

  • A capless cylinder with 255x256 divisions has 131,064 tris
  • A torus with 256x256 divisions yields 131,072 tris
  • On a sphere, it's slightly more complicated.  257 x 256 divisions yields 65,537 verts.  Delete one vert somewhere (but not from a pole), and we end up with 131,323 tris. 
  • On a cube, it's also somewhat complicated. 104 x 104 x 105 divisions 64,314 verts.  Subdivide a few hundred faces to bring the vertex count to 65,536, and we get 131,063 tris.

Of course, the worst example possible would be one in which every single grouping of three vertices is connected by a triangular face.  I don't have the patience to construct such a thing.  I did start by merging a few tetrahrdrons together, and of course, the face count very quickly exceeded the vertex count, by leaps and bounds.  But I gave up after just a minute or two, when the thing got complicated enough that I could no longer tell if there were any three vertices that didn't have a face in common.  In a world made only of three dimensions, it's something of a tall order to build such a model, and I don't have the mathematical knowledge to calculate what the total face count would be, without building it.

So, for now, I think we can call the the sphere the winner, and 131,323 tris, for 65,536 verts.  Multiply that by 80 x 255, and we get 2,678,989,200.  As a written number, it doesn't seem all that much different from the figure you'd come up with.  But in practicality, it's an increase of over five million faces.

Ain't inefficiency fun?

Link to comment
Share on other sites

Going up the learning curve is somewhat slow going for a lot of people whose only prior experience with 3d mesh was constructing sculpties. And in addition to to coming up to speed with one's 3d modeling programs of choice there's the added complexity of adapting to the somewhat changeable upload system this is both time consuming and tedious.  Ultimately I expect we'll see an explosion of mesh content in the upcoming months.

Link to comment
Share on other sites

@Chosen

I'm not quite sure about the 80 attachment points, or how different viewers handle this... the number is just a rough estimate and I'm sure I overlooked something.... I think you will not only fry your graphics card at a fraction of this rendermadness, you might take out a server or two in the process:)

Link to comment
Share on other sites


Kwakkelde Kwak wrote:

I'm not quite sure about the 80 attachment points

Understood.  No matter what the actual numbers are, it's an interesting mental exercise to figure out just how bad things could get, from any given set of limits. 

I'm usually so focused on finding the optimum way to do more with less on any given project, it's kind of fun to think momentarily about how to maxmize the concept of doing less with more.  I just hope no one's getting any ideas from this.  It's just an academic exercise.  Don't try this at home, folks. ;)

Link to comment
Share on other sites

Thought I'd wade in with some info on attachment points. Below is the most current info I've found to date :)

[ATTACHMENT POINTS] 30 attachment point on Avatar, 8 attachment points for HUDs.
[ATTACHMENTS] Maximum of 38 attachments can be worn, either dispersed among various attachment points or all on the one attachment point for Viewer2.4 and later. Older viewers are limited to 30 attachments.

Link to comment
Share on other sites


NaiadNightfire wrote:

Thought I'd wade in with some info on attachment points. Below is the most current info I've found to date
:)

[ATTACHMENT POINTS] 30 attachment point on Avatar, 8 attachment points for HUDs.

[ATTACHMENTS] Maximum of 38 attachments can be worn, either dispersed among various attachment points or all on the one attachment point for Viewer2.4 and later. Older viewers are limited to 30 attachments.

That's what I had thought, since that's the information in the knowledge base.  But from Kwak's comments, I figured the information was maybe outdated.  I suppose I should have tested.

After reading your post, I did test now, and I can confirm that the maximum number of attachments is indeed 38 (unless maybe third party viewers have a way of getting around this somehow).

So, if we want to redo those numbers now, I guess we end up with 38 x 255 x 131,323, which equals 1,272,51,870.  Aw, is that all?

 

By the way, if anyone wants to know a few more numbers.  A 1-meter sphere, with 131,323 faces yields the following:

  • The DAE file is 8.22 MB
  • The model takes 18 seconds to appear in the uploader window (and that's with a fast cable connection)
  • At default LOD's, and physics calculated from the lowest LOD, the weights are as follows.  Upload fee: L$293 Land Impact: 311.640, Dowload: 251.257, Display: 21199, Physics 311.640, Server 0.500.   Yes, 21199 display cost, for a one-meter object!
  • Display cost failed to calculate in the uploader.  The field wasn't even there. The uploader spit out an error about a missing block, which I'm assuming is the cause (effect?) of the field's absence.  To get the display cost, I had to upload the model, and rez it.
  • I tried to test with "same as above" for all LOD's, to fully maximize the costs, but I couldn't.  The uploader got all kinds of pissed off, and kept throwing up lots of really fun errors, including one about having too few vertices in the model, which I found particularly amusing.
  • On a private island, staring out at the water, with SLI enabled on my best desktop, I get a steady 60 FPS.  With this model in view, it drops to 55.  With the model selected, it plummets to 7.
  • When the sphere is stretched to 64 meters along one axis, the land impact skyrockets to over 4300.  I didn't dare stretch it along either of the other two axes, for fear it would cause other objects on the island to be returned.  At 64x64x64, this one object could well end up utilizing the entire 15,000 prim count of the region!

Anyone else think maybe LL aught to lower that vertex limit?

Link to comment
Share on other sites

I don't know what viewer you guys are using, but I am on "Second Life 3.2.1 (244227) Nov  1 2011 07:01:35 (Second Life Beta Viewer)" and I can definately attach well over 38 things, as I could with previous linden viewers, it's called "add" instead of "wear" and it allows 8 objects on one attachment point (not on all of them at the same time).

I must have counted all the links aswell I guess...duh... 38 it is, which brings up something new though. I have 40 attachment points, 32 on the avatar, including the avatar center (which is for avatar shapes?) and 8 on the HUD. A bit strange we can't use them all at the same time. ( btw this is good news, we can only have 1 billion faces to render instead of 2 billion)

Limiting the 64k? I'm not for that..I am however for a limit in prim equivalence or something. You can use 64k in the highest LOD and still get a reasonable object I think by using low numbers for the other slots, although I wouldn't know what kind of object it could be, maybe jewelry or something which won't show until you are very close.

Some limit would be nice I think, since this stuff is a griefers wet dream.

Link to comment
Share on other sites

I don't know what viewer you guys are using, but I am on "Second Life 3.2.1 (244227) Nov  1 2011 07:01:35 (Second Life Beta Viewer)" and I can definately attach well over 38 things, as I could with previous linden viewers, it's called "add" instead of "wear" and it allows 8 objects on one attachment point (not on all of them at the same time).

I must have counted all the links aswell I guess...duh... 38 it is, which brings up something new though. I have 40 attachment points, 32 on the avatar, including the avatar center (which is for avatar shapes?) and 8 on the HUD. A bit strange we can't use them all at the same time. ( btw this is good news, we can only have 1 billion faces to render instead of 2 billion)

Limiting the 64k? I'm not for that..I am however for a limit in prim equivalence or something. You can use 64k in the highest LOD and still get a reasonable object I think by using low numbers for the other slots, although I wouldn't know what kind of object it could be, maybe jewelry or something which won't show until you are very close.

Some limit would be nice I think, since this stuff is a griefers wet dream.

Link to comment
Share on other sites

"At 64x64x64, this one object could well end up utilizing the entire 15,000 prim count of the region!"

It shouldn't. The limiting physics weight can only stay the same or decrease with stretching, depending on physics shape type. So the increased weight must be the download weight. The radius of the version stretched on one axis is about 32. Stretched in all three dimensions, it would be sqrt(3) times that, which is about 55. But the download weight plateau is reached at radius = 43.44. That's only 1.36 times the present radius. Plugging the numbers* in the formula says the weight increase is about 1/0.59, so it should be 4300/0.59, which is just under 7300.

Go on .... don't you just have to try it now? :matte-motes-evil: :matte-motes-smile:

*assuming default 1/4 data size for medium LOD / high LOD

Link to comment
Share on other sites

"At 64x64x64, this one object could well end up utilizing the entire 15,000 prim count of the region!"

It shouldn't. The limiting physics weight can only stay the same or decrease with stretching, depending on physics shape type. So the increased weight must be the download weight. The radius of the version stretched on one axis is about 32. Stretched in all three dimensions, it would be sqrt(3) times that, which is about 55. But the download weight plateau is reached at radius = 43.44. That's only 1.36 times the present radius. Plugging the numbers* in the formula says the weight increase is about 1/0.59, so it should be 4300/0.59, which is just under 7300.

Go on .... don't you just have to try it now? :matte-motes-evil: :matte-motes-smile:

*assuming default 1/4 data size for medium LOD / high LOD

Link to comment
Share on other sites


Kwakkelde Kwak wrote:

 

Limiting the 64k? I'm not for that..I am however for a limit in prim equivalence or something. You can use 64k in the highest LOD and still get a reasonable object I think by using low numbers for the other slots, although I wouldn't know what kind of object it could be, maybe jewelry or something which won't show until you are very close.

 

I have to respectfully disagree with this line of thinking.  The only single object I could envision that would require so many verts would be something extremely large, like a sim-sized castle or something.  But needless to say, something like that would have to be built in pieces anyway, so there's still very little point in allowing so many verts on one object. 

As for smaller things, like jewelry and such, I'd have a hard time imagining ANY piece of jewelry, or other small item, that would benefit in any way from such tightly packed vertices.  Show me a piece of jewelry that has 64K vertices, and I'll show you a modeler who has no idea what the hell he's doing.

 

As things stand right now, the door is just way too wide open for inefficiency.  Just last night, someone who's fairly new to mesh modeling asked me to come look at his new build.  This person is intelligent and talented, so the build of course looked nice, and the land impact was quite low, even for large buildings.  But when I took a look at the wireframe, I got a rather striking reminder that no matter how much we TALK about efficiency, even the smartest people aren't necessarily going to get what we mean, without having been shown.

Here's what I saw.  The lowest-poly wall sections looked like the figure on the left in the image below, which means they had twice as many polygons as they needed.  Most walls looked something like the figure in the middle, utilizing at least three times as many polys as needed, and several had as much as 20 or 30 times more than that.  Not a single one looked like the figure on the right, which is how they all should have looked.

wallWireframes.jpg

The land impact of the whole building was only 14, which obviously was way less than it would have been had the building been made from prims -- my quick estimate, from the design, was it would have taken around 75 cubes to build the same structure from prims.  Needless to say, the creator was pretty happy with that 14.  But also needless to say, the display cost was many times higher than it should have been. 

In fairness, considering that a prim version of the same build would have weighed in at somewhere around 8100 polys (75 x 108 per cube), it's worth noting that the mesh version, even as inefficient as it was, did still come out ahead.  But obviously, simply being not as bad as one potential alternative isn't exactly the same as being good.  By my estimation, at least 90% of the mesh building's poly count (no exaggeration) should not have been there.

Once again, I'd like to reiterate that the creator of this particular build was no slouch.  Nonetheless, all signs were it simply had not occurred to him what "low poly modeling" really meant.  I suspect the same is probably true for nearly all SL builders who are now dabbling in mesh for the first time.  Most just haven't been programmed to think this way.

I'm not sure what the best remedy for that is going to be.  But I'm fairly certain that allowing people 65,536 vertices in a single object isn't likely to help the cause.

Link to comment
Share on other sites


Drongle McMahon wrote:

Plugging the numbers* in the formula says the weight increase is about 1/0.59, so it should be 4300/0.59, which is just under 7300.

 

Oh, OK, so it would take two of them and change to fill the sim, instead of just one.  Gee whiz, when you put it that way, Batman...

 


Drongle McMahon wrote:

Go on .... don't you just have to try it now? :matte-motes-evil: :matte-motes-smile:

I'd love to, if somebody's got an empty sim somewhere that would accomodate.  I tried going to a sandbox, and I couldn't even rez the one-meter version, haha.  "Unable to create object. The region is full."

Link to comment
Share on other sites

You are comparing apples and pears now, or the two of us are.

I was looking at it from a griefers point of view. I do not want to test too much of this or give peope any ideas, but if you have a mesh object and scale it up eventually the primlimit of the sim will be reached. Something will be returned... Make it bad enough and EVERYthing will be returned. Hopefully LL has programmed things in such a way you can't scale up when you hit the primlimit, or the object being returned is always the current object, but this is what I had in mind when posting.

I have made a celtic collar in the past, around 120 prims, almost all of them half torusses. (This was in my early days so yes I know it's renderhell, but it's still well worth it i think and the single prims are so small LOD makes sure it's not all that bad) to make a torus look any good it would have to have at least 24x8 sides, so 12x8 for half ones. this is 192 faces each.  In total this is almost 23k. It does look like quite a lot, but a slightly more complex celtic knot could easily double it, you are still a long way from 130k, I agree.. Anyway that was just an example, I'm sure there are other items that could use so many faces. I am building a house right now, most of it using mesh. I like building blocks, so it won't be one object, but it could have been 2 objects given the size. One doorway costs 800 faces at LOD high and to be honest it's quite simple, but it has an arch. There are around 15 doorways, I could use forty windows and I like nice roofs. Add some cornices and columns and I bet 130k is easily reached.

Whether you can use 64k vertices or 2k vertices in one object doesn't make a difference I think. Some people do know how to build efficiently, some don't. A limit like this isn't going to change that. (there are no surfaces like you drew in my builds btw, a wall is 2 faces, wall with a square door is 6, square window 8, and the wall section of my arched piece is 11 with some minor changes I could peel 100 from the 800, but they'd still be changes I don't want)

 

EDIT you posted while I was typing:) .....region full when trying to rezz? the whole thing was about worn objects, that should be possible right?

Link to comment
Share on other sites

Non-mesh viewer:

non-mesh.jpg

 

Mesh-compatible viewer (V3.2 Beta):

mesh 1.jpg

Pictures say far more than words ever can.

But what you never had you don't miss, and that's the truth of it. And I don't really mind that the dragon statue at Missing Beckett is still made up of torii after the strange "bug" affected LeTigre sims several weeks ago.  Non-mesh viewers will just see the world something akin to Picasso art, and isn't that just beautiful in it's own way?



Link to comment
Share on other sites


Chosen Few wrote:

[ Lengthy talks about the worst case object a person could create ]

 

The worst case objects you're coming up with are pretty scary, and I'm shure some one will try to build one. However things aren't quite as bad as they seem from the numbers, modern GPUs have a lot of tricks to cut down on unneeded processing. To start with during primitive assembly all vertices will be snapped to a sub pixel grid and degenerate triangles culled, then if enabled backface culling with remove all triangles with normals pointing away from the camera, then the GPU will do one or more hierarchal Z tests and cull occluded triangles, then after rasterization the GPU will do an early Z test and cull fragments (really quads) that are occlude. So in actuality only 1/3 to 2/5 of the triangles will be rasterized and of those only around 1/3 to 2/3 of the area they cover will have fragment shaders run on them (depending on how the object is made). Of course all this is assuming opaque faces, transparent faces will throw most of those optimizations out the window.

Link to comment
Share on other sites


leliel Mirihi wrote:

 

The worst case objects you're coming up with are pretty scary, and I'm shure some one will try to build one. However things aren't quite as bad as they seem from the numbers, modern GPUs have a lot of tricks to cut down on unneeded processing. To start with during primitive assembly all vertices will be snapped to a sub pixel grid and degenerate triangles culled, then if enabled backface culling with remove all triangles with normals pointing away from the camera, then the GPU will do one or more hierarchal Z tests and cull occluded triangles, then after rasterization the GPU will do an early Z test and cull fragments (really quads) that are occlude. So in actuality only 1/3 to 2/5 of the triangles will be rasterized and of those only around 1/3 to 2/3 of the area they cover will have fragment shaders run on them (depending on how the object is made). Of course all this is assuming opaque faces, transparent faces will throw most of those optimizations out the window.

All true, but I think it important we be careful in how we present such information, lest it become an excuse in the minds of some not to bother optimizing.  The same exact processes will also apply to light-weight objects, to much better effect.

Plus, there's no getting around the higher download costs and memory requirements for the more poly-heavy objects.

Link to comment
Share on other sites


Kwakkelde Kwak wrote:

You are comparing apples and pears now, or the two of us are.


 

Please explain.  I thought we were pretty much on the same page.  Even though we disagree about the limit, my impression had been that we were both talking about the same subject, and we each understood what the other meant.  If that's not the case, let's figure it out. :)

 


Kwakkelde Kwak wrote:

I was looking at it from a griefers point of view. I do not want to test too much of this or give peope any ideas, but if you have a mesh object and scale it up eventually the primlimit of the sim will be reached. Something will be returned... Make it bad enough and EVERYthing will be returned. Hopefully LL has programmed things in such a way you can't scale up when you hit the primlimit, or the object being returned is always the current object, but this is what I had in mind when posting.

It looks like some protection is in place.  When I tried to set that sphere to 64x64x64 by typing the numbers in the editor, it did refuse to take any further input after the first entry went in.  After setting X to 64, when I tried to also put 64 for Y or Z, it would just revert back to 1 as soon as I hit the Enter key.  I'm hoping that's a deliberate feature, to protect the other content on the sim.

However, when I tried to scale the object visually, I was able to stretch it uniformally, all the way up.  I did not let go of the mouse, though, because I didn't know if it would just snap back, or if it would nuke the existing build.  I didn't want to take the risk.

 


Kwakkelde Kwak wrote:

I have made a celtic collar in the past, around 120 prims, almost all of them half torusses. (This was in my early days so yes I know it's renderhell, but it's still well worth it i think and the single prims are so small LOD makes sure it's not all that bad) to make a torus look any good it would have to have at least 24x8 sides, so 12x8 for half ones. this is 192 faces each.  In total this is almost 23k. It does look like quite a lot, but a slightly more complex celtic knot could easily double it, you are still a long way from 130k, I agree.. Anyway that was just an example, I'm sure there are other items that could use so many faces.


If you want to look at it that way, we could break down the math pretty easily, with just toruses.  One SL torus has 576 vertices in it.  So, 65,536 verts is roughly equivalent to 113 toruses.  We all know there are avatars walking around the grid with at least that many attached.

But is that a good thing?  I really don't think so.

To stick with your example, a Celtic knot, as I think you know, can be a very good case in point of how not to model for realtime.  No game artist in the world would build that kind of item the way you did it back then as an SL newbie. 

If someone's goal is to actually construct all those details, rather than just to suggest them with texturing, that's fine, but they should do it for a CGI film, or for a still render.  Realtime 3D isn't about making everything be as mutli-dimensional as possible.  In game art, a good artist is one who strikes the optimum balance between surface detail and performance considerations.  In the majority of cases, that's likely to mean slapping a picture that Celtic knot onto a flat surface. 

In game engines more sophisticated rendering options than SL has, we'd most likely use a normal map or a bump map to make the features of the knot appear to have depth. But ultimately, the real geometry would still be just a flat surface, and the detailing would still be just a texture.

In SL, we don't have those kinds of options (yet), so I do understand why people are tempted to make such things from actual geometry.  But in most cases, a well made texture will still do the job more than adequately.   When you can eliminate 99% of the poly count, while still having the item look 90% as good, there's just no way to justify all that geometry.  (Plus, in a great many cases, the well made texture will look BETTER, not worse, than the collection of prims, so it becomes a double win.)

That said, if you really did want to construct the knot as a mesh model, you could do it for a hell of a lot less than 23,000 polys.  Unlike with prims, you can control where your hard and soft normals are, which means even without such luxuries as normal maps at your disposal, you can still model with great efficiency, relatively speaking.

I just made the two knots pictured below.  They're each just 893 polygons, and 669 vertices.  That's less than the poly count of a single torus, and just a hair over the vertex count.  Clearly, it doesn't have to take a ton of polygons to make even a complicated shape.

shieldKnots2.jpg

The reason I made two was to illustrate an important point.  You indicated you beleived a torus needed to have at least 8 sections around its minor circumference in order to look good.  I assume by "good" you mean sufficiently round. from looking at these two knots, I think you'd agree the one on the right appears to be made from round cord, while the one on the left appears to be made from angular cord, right? 

Well, appearances can be deceiving, which is a principle game artists realy on all day long.  Yes, one appears round, and one appears sharp, but but in actuality, the two models are exactly the same.  The only difference is the one on the right has soft normals on the central edge loop, while the one on the left has hard normals on that same loop. 

Here are the same two models, with their wireframes showing:

shieldKnots_wireframeOnShaded.jpg

As you can see, in both cases, the "cord" is just two-sided, not eight-sided.  That's why the poly count is so low.  Without the wireframe showing, I would challenge anyone to be able to figure out how many sides are actually there.  Normals are powerful things.

 


Kwakkelde Kwak wrote:

I am building a house right now, most of it using mesh. I like building blocks, so it won't be one object, but it could have been 2 objects given the size. One doorway costs 800 faces at LOD high and to be honest it's quite simple, but it has an arch. There are around 15 doorways, I could use forty windows and I like nice roofs. Add some cornices and columns and I bet 130k is easily reached.

130K for a large object like a house might be perfectly reasonable.  It also might not.  It depends on the house, of course.

From your description, I do have to say I have hard time imagining why an arched doorway has to be 800 polys.  To really require that many, it would need to be one heck of a decorative arch, but you say it's simple, so I'm confused.  If you'd care to post a screenshot, I'd love to see it. :)

 


Kwakkelde Kwak wrote:

 

Whether you can use 64k vertices or 2k vertices in one object doesn't make a difference I think. Some people do know how to build efficiently, some don't. A limit like this isn't going to change that.

Yes, some people know how to model efficiently, and some don't.  But that doesn't mean the hard limit doesn't make a difference.  To go with your numbers (which I do realize you just made up for emphasis, so don't worry), if I can't use more than 2K per object, I'm probalby going to be very careful about how I utilize each of those 2000 polygons, especially given that I can only have 255 objects in any single linkset.  At 64K, if I don't know a thing about rendering concerns, then what do I care how many polygons I waste?  The sky's the limit.  I've got no enforced reason to care.

Of course, as I said in another thread on the topic a few weeks ago, I'd like to think that, if nothing else, at least market forces will eventually prevail, and those making lagtastic stuff will either learn to do better or quit.  But that may just be wishful thinking.

 


Kwakkelde Kwak wrote:

(there are no surfaces like you drew in my builds btw, a wall is 2 faces, wall with a square door is 6, square window 8, and the wall section of my arched piece is 11 with some minor changes I could peel 100 from the 800, but they'd still be changes I don't want)


I didn't mean to imply I was talking about your builds in particular.  I was speaking generally, and comiseratively (if that's a word), as in "Here's something we can both agree we think needs improvement."

 

 

 


Kwakkelde Kwak wrote:

EDIT you posted while I was typing:) .....region full when trying to rezz? the whole thing was about worn objects, that should be possible right?

Yeah, definitely possible to attach the object, no matter what the state of the sim.  But I wanted to see what the land impact would be at full size, and attaching it won't tell me that.

 

 

Link to comment
Share on other sites


Chosen Few wrote:

 

Please explain.

 

 

Apples and pears...well You are looking at it from a builders point of view, where you obviously are after as little data as possible, maximum looks with as little faces as possible. This can maybe..maybe be stimulated by lowering the number of available faces per object. I appreciate your faith on this matter, although I don't really share it as I said before.

A griefer on the other hand is after as much data as possible with the least amount of efford.

An often used griefers tool is the self-replicator or how do you want to call it, a real pain when you don't have estate rights. So this is what I had in mind when I made the comment. Come to think of it, maybe as little data as possible per object, but as many objects as possible is more rewarding for a griefer, for visual purposes. In that case my line of thinking wasn't right.

but...

By lowering the maximum number of prims in a mesh object (or prim equivalence really) to let's say 255, you will make sure nobody makes a rediculous object as we both suggested, the 130k ones. I have no idea what kind of object would be 255 and how much detail it would have, but after playing around with numerous kinds of meshes in SL, I'd say 255 allows a fairly nice build. At not even that big a size you said the 130k faced object exceeded the maximum supported prims on a full region. with 255 as a max you can rezz 59 objects. So making the restriction the same by lowering allowed faces per object, that would result in 131000k / 59 = 2200 faces. Now that sounds really really low to me.(I have the nagging feeling I am completely miscalculating this, but my point is a couple of lines down the post..)

The thing that really worries me, isn't land impact though or griefing. The real renderhell always was and always will be in attachments. So with a 255 PE as a limit for a linkset, you won't come close to the horrific 255 x 130k = 33 million faces you can have now for every attachment point. I think it all comes down to that. No idea what is reasonable for one attachment, but let's say 10k faces. That's 33k times lower than what's possible now. So 130k faces / 33k doesn't leave a lot of faces per object to say the least....:) Now I am curious how many faces a 255 PE mesh can hold at the small size of an attachment. I'm sure it's a LOT higher than 10k, but closer to that than to 33 million.

Anyway, I think we're on the same page, just a different paragraph.

 

For the celtic collar, well I think what I have built back in the day can easily be described as overkill, I wouldn't have brought it up if I thought otherwise. But this is SL, not a game. You can't compare the two one on one I think. There are a lot of similarities, they use the same kind of software and push the same kind of hardware and building for either one means balancing looks and resources. But as I understand people want 80 fps for a game, where I think the 20-30 fps I normally get in SL are very acceptable for the purpose of SL, which is a social platform rather than a high performing game. 10-15 fps occasionally is no problem for me, I get annoyed when it drops lower. So following that logic I think you are "allowed" to push the sytem a bit more if that's what it takes to obtain a certain look. Alpha layers can't come close to actual prims with metal shine. I do agree that when you can build an object at the fraction of the weight (and this is ofcourse measurable) and at 90% of the looks, you should go for that option. But the 90% is very subjective in two ways. The number itself is subjective, some might say 40% of the rendering cost and 60% of the looks should be the turning point, others 5% and 95%. Next to that it's very subjective what "90% of the looks" means. As for your example, first of all, looks good and sounds very reasonable to render. But it's nothing like the tubular celtic thing I have made. Mine is a real mesh (yes let's confuse everyone by using this term..I mean in a chickenwire way of mesh, not a 3D program mesh), so two polys per section won't cut it, maybe 6 would be enough rather than eight , but two certainly won't.

I'd post a pic if I knew how..something is wrong...grrr ....seems to work now...

screenshot - collar.JPG

I'd post a pic of the doorway aswell....grrr..

screenshot - arched doorway.JPG

The doorway (5.0x0.9x7.0m which is rather large for a RL door, don't blame me, blame the person who wants it:))

The arch has 16 sections I think 24 sections, which look like the above in the picture below...sketch section.JPG

 I could make the section slightly different, using 20 instead of 24 faces...or use a box with textures, which would still be 8, again this is very subjective.

 

All I ment when I said the doorway doesn't have sections like you descibed, was to show out of the 800 faces, none are useless. I never thought you ment me, afterall you've never seen the thing:)

 

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 3605 days.

Please take a moment to consider if this thread is worth bumping.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...