Jump to content

one prim mesh


KoUsagiHime
 Share

You are about to reply to a thread that has been inactive for 1096 days.

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

Recommended Posts

Do you mean, how does one make mesh objects with an LI of 1? The short answer is to make them less complex, make even less complex LoD models, and an even less complex physics model. The long answer involves thousands of web pages on ways of optimising mesh models for upload into SL.

Or do you mean that when you upload a mesh object it becomes a link set of more than one object? The only reason I know of for this happening (and I'm not an expert, so there may well be others) is that you've made the model with more than eight materials. Blender materials (I don't know what they might be called in other 3D modelling programs) correspond roughly to SL faces. SL objects can only have eight faces, numbered 0 to 7. So if you try to upload a mesh model with more than eight materials the uploader will break it up into smaller components each with eight or fewer SL faces. To avoid this you need to be efficient with your use of materials, using as few different textures as possible, for example, and making sure that all the Blender faces (as opposed to SL faces) using the same texture are combined in a single material.

  • Like 2
Link to comment
Share on other sites

7 hours ago, KoUsagiHime said:

i have seen several people make one prim meshes, how?

First, did you use the decimate modifier so that you model is less complex?  You might try searching on YouTube for some tutorials on this modifier.

Link to comment
Share on other sites

54 minutes ago, Dhyaanee said:

First, did you use the decimate modifier so that you model is less complex?  You might try searching on YouTube for some tutorials on this modifier.

Yes but it's not really that effective. The rule for Maya and Blender users who want to make mesh for Second Life is that you need to spend at least 90 percent of the time in edit mode, not object mode, because that's where you find most of the really useful tools.

Link to comment
Share on other sites

14 minutes ago, ChinRey said:

Yes but it's not really that effective. The rule for Maya and Blender users who want to make mesh for Second Life is that you need to spend at least 90 percent of the time in edit mode, not object mode, because that's where you find most of the really useful tools.

Since getting a more complex shape requires quite a bit of geometry (as far as I know), what would be the first step in Edit mode to reduce that geometry?

  • Like 1
Link to comment
Share on other sites

2 hours ago, Dhyaanee said:

Since getting a more complex shape requires quite a bit of geometry (as far as I know), what would be the first step in Edit mode to reduce that geometry?

It depends but a typical process can be:

  1. If the model is made from tris or from complex polys, convert to quads (Ctrl+T to triangulate and then Alt+J to ... ummm quadrify?) This makes everything so much easier.
  2. Eliminate any suprefluous edge loops/lines first since they can be hard to identify and select later in the process.
  3. Delete any hidden tris.
  4. Do a limited dissolve (Shortcut X) and set the Max angle as high as possible without distroting the model.
  5. Limited dissolve creates a lot of complex polys, so you have to convert to quads again afterwards.
  6. Repeat step 4 and 5 until the triangle count no longer drops or the model becomes distorted (when it does, remember to undo last round with Ctrl+Z).
  7. Check the model and see if there is anything else that can be safely removed. Look especially for "zero angle" vertices that can be removed without altering the model at all. For some reason limited dissolve isn't good at handling those.
  8. Repeat step 4 and 5 until the triangle count no longer drops or the model becomes distorted.

Do this for each LoD model, including the high one.

Here's one very useful trick I've never seen anybody mention: If you're unsure whether you've simplified too much, take before and after screenshots in object mode and compare them side by side. It's much easier to judge the differences that way than when you undo and redo to switch between the two in Blender.

Edit:

This is a 256x256 m surround mountain build so it can't be used in SL but it's what I happen to be working on right now and it should illustrate the power of the limited dissolve.

Before - 838 triangles:

bilde.png.c86385ed79d3d7bf69c9ab40a4cbf39d.png

Limited dissolve, 15 degrees Max angle - 777 triangles:

bilde.png.a5684a7d711c0ba5adee9316dd30ceb8.png

Not the biggest saving of course but this was a model I thought I had already reduced as much as possible (it started at 1922 tris) and it took less than five seconds.

Edited by ChinRey
  • Like 1
  • Thanks 2
Link to comment
Share on other sites

10 minutes ago, ChinRey said:
35 minutes ago, Dhyaanee said:

Since getting a more complex shape requires quite a bit of geometry (as far as I know), what would be the first step in Edit mode to reduce that geometry?

It depends but a typical process can be:

  1. If the model is made from tris or from complex polys, convert to quads (Ctrl+T to triangulate and then Alt+J to ... ummm quadrify?) This makes everything so much easier.
  2. Eliminate any suprefluous edge loops/lines first since they can be hard to identify and select later in the process.
  3. Delete any hidden tris.
  4. Do a limited dissolve (Shortcut X) and set the Max angle as high as possible without distroting the model.
  5. Limited dissolve creates a lot of complex polys, so you have to convert to quads again afterwards.
  6. Repeat step 4 and 5 until the triangle count no longer drops or the model becomes distorted (when it does, remember to undo last round with Ctrl+Z).
  7. Check the model and see if there is anything else that can be safely removed. Look especially for "zero angle" vertices that can be removed without altering the model at all. For some reason limited dissolve isn't good at handling those.

Do this for each LoD model, including the high one.

Here's one very useful trick I've never seen anybody mention: If you're unsure whether you've simplified too much, take before and after screenshots in object mode and compare them side by side. It's much easier to judge the differences that way than when you undo and redo to switch between the two in Blender.

Thanks.

Somebody should sticky this!

Link to comment
Share on other sites

9 hours ago, KoUsagiHime said:

i have seen several people make one prim meshes, how?

I suppose you've figured out by now that this can be a rather complex question to answer. ;)

But the first thing you need to do is learn how land impact is calculated.

Log on to SL, right click on an object in the scene around you (any object will do for now) and select Edit. In the Edit window, click on the "More info" link. That'll bring up a list with some data and there are three items on that list that matter here: Download weight, Physics weight and Server weight.

bilde.png.55df160e05afdfe9f3e246f38ad2d17c.png

These are estimates how much load the object puts on various parts of LL's network and it's the highest of these three weights that counts as land impact!

In my illustration, it's the physics weight that is the highest so that's the one that needs to be reduced to lower the LI. Reducing the server or download weights won't help in this case. That's actually quite unusual but in any case: work on the highest of the three weights if you want to reduce LI.

Server weight

That's the easiest one to understand. The server weight is simply the number of parts in the linkset divided by two. (Plus sometimes a bit extra if the object is scripted - it depends on what kind of scripts it runs.)

As KT Kingsley mentioned, if your mesh uploads as a linkset, this may be the one you need to handle. The solution is to merge objects before you export from whatever mesh editor you use.

Download weight

This is usually the big one and it depends mainly on the number of tris and vertices of each LoD model and the size of the object.

You may know that a mesh in Second Life isn't actually a single mesh but up to four (no five but we'll deal with the fifth later) different ones. Those are the LoD models, different versions of the mesh with different amounts of details intended for different view distances. You don't need all the finer details on an items seen from afar of course and there's a lot to be saved by simplifying those. I'm not sure if SL would even be possible without them.

The trick is of course to figure out how much detail and which details you need at the various view distances. Remove too much and you get those horrible collapsed meshes you see all over SL. Don't remove enough and you get "Mole mesh" - looks ok but takes up far more computing power and bandwidth than it should. Remove the wrong details and you get both problems.

The uploader is kind enough to offer to generate those LoD models for you but unfortuantely it's horrendously bad at doing the job and it always removes the wrong details. So if you really want to make good mesh, you need to ake them all yourself.

This is one of those "a minute to learn, a lifetime to master" skills. Anybody - I mean literally anybody - who knows how to make mesh at all can do better than the uploader's autogenerated models. All you need to do is engage an eye or two and a few brain cells. Do it that way and you're making mesh you can be proud of. But there's always room for improvements. I've been doing it alsot every day for more than six years now and I'm still learning.

Physics weight

I said there were five models. The fifth one is the physics model and it's the one your avatar crashes into and sometimes walk on. As you've probably guessed, that's the one that determines the physics weight.

Second Life uses the HAVOK physics engine and it has a ton of strangely counter-intuitive quirks and if that's where your problem is, you better post some pics and such so we can look at it. The basic principle is the same as for the LoD models though: Keep it as simple as possible.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

2 hours ago, ChinRey said:

Here's one very useful trick I've never seen anybody mention: If you're unsure whether you've simplified too much, take before and after screenshots in object mode and compare them side by side. It's much easier to judge the differences that way than when you undo and redo to switch between the two in Blender.

 This one made me chuckle. 🤭

I usually end up making a fifth LOD model. So I have them side by side in the viewport. Because I always have a hard time taking away all the hard work I have put into the model in the first place. 🤪

Land impact will force me to use the super low version anyway, but I can't help it, my initial lowest LOD is almost never simplified too much. 😁

  • Like 2
Link to comment
Share on other sites

26 minutes ago, arton Rotaru said:

Because I always have a hard time taking away all the hard work I have put into the model in the first place. 🤪

 

Oh yes it hurts and maybe that's an important reason why we so so much poor LoD in SL. It's not that people don't know how to make good LoD models or that they don't understand how important it is. It's jsut that they can't bring themselves to do it. Much less painfull to either overload the models with details or look away while the uploader does its butchery.

 

30 minutes ago, arton Rotaru said:

I usually end up making a fifth LOD model.

You mean you only make one extra LoD model??? 😮

 

Link to comment
Share on other sites

1 minute ago, ChinRey said:

You mean you only make one extra LoD model??? 😮

I make the High, the Medium, the Low, and the Lowest. Plus an ultra Lowest. Indeed, I first load the model with the regular Lowest in the importer. Land Impact will make me cry, though. So I load the ultra Lowest, and see what difference it makes regarding land impact.

The good thing is, once in-world, there is hardly any noticeable visual difference between the 2 versions, when zooming out.

  • Like 2
Link to comment
Share on other sites

1 hour ago, arton Rotaru said:

Land Impact will make me cry, though.

The trick to avoid that is to have realistic expectations of what is possible. I always have but unfortunately the universe doesn't always agree with me what is realistic. Right now for example. I'm trying to make a group of 50 lovely smooth round rocks from 200 tris but for some weird reason the universe won't let me. Is that fair???

 

2 hours ago, arton Rotaru said:

The good thing is, once in-world, there is hardly any noticeable visual difference between the 2 versions, when zooming out.

Another important rule to remember:

Kill your darlings!

All content creators suffer from Builder's Blindness. That's when we get so focused and attached to every little detail of what we're making we miss the whole picture. This is not actually a bad thing because we want and need it during most of the process. But in the final stage we need to shift focus and look at the big picture and once we do that, suddenly a lot of thsoe finer details become superfluous.

  • Like 1
Link to comment
Share on other sites

18 hours ago, ChinRey said:

The trick to avoid that is to have realistic expectations of what is possible.

I was dramatizing a little bit. I don't actually cry. 😄 It's more in the context of simplifying too much. Before I simplify too much, I shift drag a copy and continue reducing the geometry on that. I'm also not obsessed with 1 LI objects. It's only when it is at the edge of 1.5 - 1.6. I try to push those below 1.5.

Link to comment
Share on other sites

On 4/27/2021 at 8:16 PM, arton Rotaru said:

Land impact will force me to use the super low version anyway, but I can't help it, my initial lowest LOD is almost never simplified too much. 😁

If you already know your LOD swapping distances, it is very important that when you look at your lods, you look at them at the camera distance they are intented for, and that you do so using the LL viewer default settings, since that is the "default" experience.

Most low lods tend to look horrible from up close, but they are NOT ment to be seen up close, they are ment to be "there" and to avoid drawing too much attention when the higher lods pop into view.

The two high lods tend to be the more difficult ones to get right because they are usually close enough to the camera that a sudden loss in detail will be much more noticeable, and that's the ones where you have to be clever if you want to hit that 50% reduction from highest to high.

Link to comment
Share on other sites

2 hours ago, Kyrah Abattoir said:

Oh my bad,  I'll remember to avoid being helpful in the future ^_^.

I just couldn't resist to reply in the same tone as I read pretty much in any of your being helpful replies. Total lack of humor. Which your conjectural inability of reading the smileys in my posts demonstrate.

Edited by arton Rotaru
Link to comment
Share on other sites

6 hours ago, arton Rotaru said:

I just couldn't resist to reply in the same tone as I read pretty much in any of your being helpful replies. Total lack of humor. Which your conjectural inability of reading the smileys in my posts demonstrate.

Well, I'm sorry. I tend to be pretty bitter even when trying to help because of this feeling that nothing I do or say will sway anyone in the right direction because other than increasing their profit margin or getting done faster, people don't appear interested.

I tend to take this way too seriously, but I've lived through SL for so long it is just hard to let go. :/

  • Like 1
Link to comment
Share on other sites

On 4/28/2021 at 12:26 AM, ChinRey said:

Kill your darlings!

Ah well, i wish more than 1% would at least try to do so. For clothings nearly nobody do any optimizations and as long as they don't have to care about land impact, they think it's not a problem, because the uninformed and/or ignorant customer will buy it anyways, no matter if it's game optimized or not. Best example: Thats an attachment i found on one of the big events, the lower textbox is the answer afteri told the creator that this sh*it is the reason why sl is a lag-fest. I removed names, so i hope thats okay with forums rules.

sl.jpg

  • Sad 2
Link to comment
Share on other sites

"Slightly more robust graphic card" is such a bull***** take. Unless you are on a really underpowered laptop, Your GPU is basically asleep when running SL.

You can literally mine ethereum in the background and see no change in FPS on a gtx1070.

@Tonk Tomcat1.3M triangles and 100Mb of textures too. Why compromise when you can have both! Please tell me you left the review it deserved.

Link to comment
Share on other sites

On 4/27/2021 at 1:24 AM, KoUsagiHime said:

i have seen several people make one prim meshes, how? 

can someone guide me in the right direction or show me ;.; i'm about ready to lose my mind.

thanks :)

You might try the classes @ Blender Benders, as I remember they dealt with reducing polycount -- mainly using limited dissolve and decimate. I'm not sure they got as detailed as ChinRey has, but you could ask in class. A new class started a couple months ago but I imagine you could enter now if you have a bit of experience with Blender.

* The nice thing about this class is that they cater it to what works best for SL.

Edited by Luna Bliss
Link to comment
Share on other sites

5 hours ago, Tonk Tomcat said:

Of course, i IMed the creator and told him, but yeah, the answer is what you see in the textbox and thats literally all he said. I guess all his new items will be pretty much the same.

Is it even correct to call such people "creator"? More like destroyer

  • Haha 1
Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 1096 days.

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

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...