Jump to content
Penny Patton

One-Texture Environment in Unreal / Modular Textures in Realtime 3D Art

Recommended Posts

I just wanted to share this interesting article I stumbled across where someone built an entire environment in Unreal with the goal of only using one single texture.

One-Texture Environment

And here's a thread on Polycount where the artist posted WIP shots over the course of creating the environment.

Granted, the artist here uses a lot of technical tricks that aren't available to SL users but it's still fascinating. And while this is on the extreme end it does highlight how much emphasis is placed on optimization for realtime 3D art outside of SL, something I wish we'd see more of in this community. While modular textures are common in game design, I can't say I've ever seen the particular trick used in content made for Second Life. If we ever want to see better performance in SL, and more reasonable hardware requirements for what we see on screen, it's something the SL community really needs to push towards.

The thread in General regarding the Crystal Cave skybox LL released for Premium accounts got me thinking again about how LL's in-house content should really do more to showcase good creation habits that users might learn from, or at least be able to see examples and the performance gains possible.

  • Thanks 2

Share this post


Link to post
Share on other sites
2 hours ago, Penny Patton said:

LL's in-house content should really do more to showcase good creation habits that users might learn from

Hahaha... Good Content needs Good Content Creators, and hiring those costs MONEY...

Hiring some freshly graduated inexperienced hack from a cgi course at the Buttburg College of Knitting, who's keen to work in the 'games' industry, and who'll accept 'apprenticeship' payscales, is much cheaper, and so... You get bad content...
 

Share this post


Link to post
Share on other sites
1 hour ago, Klytyna said:

Hahaha... Good Content needs Good Content Creators, and hiring those costs MONEY...

Well, there are good content creators and then there are good content creators. Most of the Moles are actually excellent content creators when they work with the materials they are familiar with. They are all excellent prim builders and very efficient sculpt builders and they have some really good texture makers and at least one brilliant scripter. But then there is this mesh thing...

It's not that they don't want to, nor is it that LL doesn't pay them to do a proper job, they are genuinely doing their best. I have one example to illustrate that - I suppose enough time has passed now it's safe to mention it without outing the poor builder.

I took part in a charity event last year. I don't usually do events but this time I was obliged to since they were decorarting half a sim with my plants and happy to since it was for the benefit of a cause I really, really support.

One of the main attractions at the event was a mansion made by a well known builder who also was - and as far as I know still is - a Mole. It was absolutely gorgeous - every "I <3 Goth as long as it's Pink" girl's dream. The problem? You guessed it. land impact at least twice what it should have been and the LoD was a total disaster. Possibly physics issues too - I never checked.

This was a house made by a Mole athough he made it under his regular SL account, not as a Mole. And it was so obvious it was a build he really cared for and had put all his heart and soul into. He failed not because he didn't care but because he didn't know how to do it.

Yes, I've been criticizing lots of unnamed (and hopefully unidentified) "mesh fakers" before and I do think some of them could do with a good spanking. This is different though, this is somebody who really is trying his best. What people like he need, is teaching. Tell them how to do it and they'll will. The problem is that he's supposed to be one of the teachers so we end up in a classic blind-leading-the-blind situation.

  • Like 2
  • Thanks 2

Share this post


Link to post
Share on other sites
12 hours ago, Klytyna said:
14 hours ago, Penny Patton said:

LL's in-house content should really do more to showcase good creation habits that users might learn from

Hahaha... Good Content needs Good Content Creators, and hiring those costs MONEY...

Hiring some freshly graduated inexperienced hack from a cgi course at the Buttburg College of Knitting, who's keen to work in the 'games' industry, and who'll accept 'apprenticeship' payscales, is much cheaper, and so... You get bad content...
 

Klytyna, Penny, Optimo, and ChinRey…

In order to reduce lag do you guys think it would have been better for LL to change the way mesh and sculpts are implemented at the core of it all rather than asking content creators to rein in their triangles?  I'm sure changing both ends would be great, but I don't see any information about changing the basic code. The end result is that we will end up having less value for our money if LI is calculated differently (unless they do allow more prims on the land too).

Share this post


Link to post
Share on other sites
1 hour ago, Luna Bliss said:

Klytyna, Penny, Optimo, and ChinRey…

In order to reduce lag do you guys think it would have been better for LL to change the way mesh and sculpts are implemented at the core of it all rather than asking content creators to rein in their triangles?  I'm sure changing both ends would be great, but I don't see any information about changing the basic code. The end result is that we will end up having less value for our money if LI is calculated differently (unless they do allow more prims on the land too).

Traditional Saying: "If Wishes were dishes, we'd all eat like Kings"

Would it be nice if SL ripped out the kludge fixes of years gone by and implemented proper systems? Sure!

It would be great if SL supported mesh made with quads instead of just tris, like some 15-20 year old fossil application, but it's not going to happen, redesigning the core mechanics at this point wold break so much, and take forever, and cost almost as much as they wasted on Project Stupid.

It's like this threatened reworking of RCI scores for Fitted Mesh, because of a "exploited LoDs bug"...

The mesh haters are gloating because they hope it will stop us wearing all the mesh content of the last 4 years and drive us all back to system clothes and bodies and flexi prims, and their failed 10 year old stores will make money again.

The little league mesh makers who failed to compete because they didn't make anything anyone wanted to buy, are gloating, because they hope it will drive the big brands under, and "open the free market" for their badly made, unpopular crap.

Then there's the Bake-Fail-Propaganda-Cabal, who hope the threatened changes will force the big brands to go "No-Onions" so that Bake-Fail-on-Mesh will rule the Grid, and fossils with 10 year old failed system skin stores will live again!

...

And me? I don't hope... I fear... I fear that heavy handed "All Stick no damned Carrot" tactics will basically make 4 years wrth of popular content unusable, and that as a result, a significant majority of SL's active users will simply say "I ain't starting over from scratch, hell with it..." and vanish, EXACTLY the same way many people did when they were told that SL was history and that Project Stupid would replace everything, and we'd all have to start over.

Here's another Traditional saying...

"The Cure is worse than the Disease..."
 

  • Like 2

Share this post


Link to post
Share on other sites
8 minutes ago, Klytyna said:

Here's another Traditional saying...

"The Cure is worse than the Disease..."

lol too true...

and what's going to happen to all of us :(

I can't even remember all the changes that screwed things up for me, but I just found a workaround when I could. Sometimes you can't though.

I'd suck it up for progress, but will it be progress?

Share this post


Link to post
Share on other sites

I genuinely love feedback.  I have yet to see or have any one suggest to us how you would use your masterful expertise to do any of this differently.   As mentioned, please stop and think that we are operating within the confines of the same potentially flawed systems you all very much love to bash and point out.   On top of that you trash our content and content creators at a very deep, personal level, who are the very same people so passionately in love with the community and Second Life that they do this all for the love of it in many cases?   

I shouldn't have to justify our actions, but here is a great example I will share about what happens under the hood.   I tell my creators to build to the lowest common denominator and upload all LOD's on most things we do because the majority of Second Life users have what we refer to as 0 spec hardware.  Those are the pc's or laptops that have very weak integrated graphics that can't manage to render anything non-collapsed over 1.0 on a good day without suffering through 3 FPS and then normally would  only being able to see collapsed content, which is something in many cases we can not afford.  We will not have our builds non-view-able by the lowest denominator, they need to be able to see it.  Having something like Learning Island and it's builds for example in a collapsed form would be an absolute show stopper for the new user experience.  We must avoid the OLOD factor 4.0 epidemic that everyone so wrongly suggests as a setting because it is undeniably the wrong thing to do for so many reasons that I will not debate.

Since none of you have any insight in to how we do or why we choose to do things the way we do them, it would be great if you stop trashing the people who quite potentially are also your friends outside of being a Mole.   If you feel you have some meaningful way to show us how you would have precisely done anything, any better than we have, I'm all ears.   Four of them as so pointed out to be exact.   I invite you to come talk to us, and hopefully in a much better atmosphere than here.

  • Like 8
  • Thanks 4

Share this post


Link to post
Share on other sites
3 hours ago, Luna Bliss said:

In order to reduce lag do you guys think it would have been better for LL to change the way mesh and sculpts are implemented at the core of it all rather than asking content creators to rein in their triangles?

Well, sculpts should have either never been a thing, or carried a higher land cost than "1 prim" from the beginning. I'm sure a book could be written on how the pre-mesh tools could have been done better or improved over the years so I'll just focus on mesh, which is what most content is created with now. And with mesh I'd say Land Impact costs could be calculated better, and LL agrees. They're hashing that out right now with the goal that LI will more realistically reflect the rendering cost of content, but not discourage proper use of LOD like the current calculations do.

Beyond that, textures are a bigger culprit when it comes "lag" than mesh complexity, for everything besides attachments at least. LI as it exists now isn't perfect, but does tend to keep triangles reigned in. And it does nothing to stop excessive use of textures. You can slap as many textures as you want on a mesh object and it won't affect it's LI cost at all. I've seen sims pushing 8-10 GIGABYTES of textures and and individual avatars are regularly pushing 200-500MB apiece. That causes a lot of performance issues.

Now, when it comes to avatars there's also no real restriction on object detail, and rigged objects are always shown at their highest detail level. So you see all of the really high-poly stuff being made as avatar attachments. Adding some sort of Land Impact like pool to avatar attachments could sort that out. Having textures affect LI could help bring down texture use overall. You don't need to hop and wish for content creators to reign in triangles or anything else out of the goodness of their hearts, you just need a system in place to discourage the worst habits.

And none of this would reduce value for anyone. We already gravitate towards low LI for the express purpose of getting more value. It would just encourage content creators to be more mindful of optimization, which means there would be more, better looking, content that won't kill framerates or create lag.

1 hour ago, Patch Linden said:

I genuinely love feedback.  I have yet to see or have any one suggest to us how you would use your masterful expertise to do any of this differently.   As mentioned, please stop and think that we are operating within the confines of the same potentially flawed systems you all very much love to bash and point out.   On top of that you trash our content and content creators at a very deep, personal level, who are the very same people so passionately in love with the community and Second Life that they do this all for the love of it in many cases?   

I agree that the personal comments against people aren't helpful. That's the curse of the internet, it's easy to forget that there are people on the other side of the screen, especially when we are frustrated. And there's a lot of frustration when it comes to SL.

I do feel communication helps, which is a point I was trying (and apparently failing) to make when starting this thread. Second Life us as much a content creation tool as it is anything else and I'm trying to say that content created by LL, either in-house or via the Moles, could be used to educate people about good building habits, avoiding bad habits, and generally showcasing what's possible in SL.

The simplest way to achieve that is by making good content and by making it modifiable so others can take it apart and see how it was made. With the inspection tools in Firestorm I can see that all of the furnishings for the new crystal cave skybox use a combined 3 textures, total. And since I'm assuming that they use spec and normal maps that means they use modular textures, 1 diffuse, one specular and one normal, and that's great...but no one can look at this furniture and see that, or how it was done, because it's no-mod. Now, the skybox itself uses a lot more but because it's no-mod I can't really know how those textures are being used.

And, again keeping in mind that SL is as much content creation software as it is anything else, I don't think it would be a bad idea to do a blog post about new releases that go into a bit of detail about the creation process. Explaining what is done and why. Maybe not with every new release but often enough to cover the basic aspects of content creation, answer community questions, and tackle some of the misinformation when it comes to content creation.

  • LL has to be aware that most of their userbase has no idea why the OLOD factor 4.0 epidemic is a problem.
  • LL has to know that a large portion of their userbase doesn't believe texture use can impact framerates at all.
  • LL has to have read some of the debates raging on this forum every day where people try and sort out what causes lag or kills framerates based on nothing more than their personal experiences and best guesses, often leaping to the wrong conclusions or wild speculation.
  • LL has to have seen that most of their userbase doesn't understand why scale is an issue for land use, value, performance, animations and visual consistency. And I'm not just talking about the average casual user who is afraid to resize things with the edit tools, these are problems within the content creation community itself.

None of this is a secret. It's right here on the forums, on every SL blog, I even see a lot of it coming up at Office Hour meetings. If there were an official creation blog where work by the Lindens, Moles, and even residents who understand these issues could be examined, explained, and content showcased I think it would not only help the tone of the discussion, but diminish a lot of the misinformation that is commonplace today.

  • Like 1
  • Thanks 2

Share this post


Link to post
Share on other sites

@Penny Patton Thank you - By and large I think you are on the same page with us regarding most of what you mention above.  We know and understand everything you have on your bullet points.  I read the forums and am by and large out and about in the community myself along with many other Lindens. 

The only thing amiss from all of this?   We can not use Firestorm.  I did read Beq's wonderful blog post today about LOD (someone pointed it out to me), and also how that tool set works.  As a (dormant) content creator myself, I am fascinated by it.    It seems we are missing this valuable tool out of our viewer and that's something I'm going to discuss internally coming out of this.

I would like very much for us to establish some sort of creation blog, or maybe it's a specific forum to go over everything you mention and hopefully impact the tone of discussion in a positive and meaningful way.  I courage everyone on our team to get out and discuss everything we do, because we are open to present what information we do have.

  • Like 4
  • Thanks 1

Share this post


Link to post
Share on other sites
4 hours ago, Patch Linden said:

I genuinely love feedback.  I have yet to see or have any one suggest to us how you would use your masterful expertise to do any of this differently.

I can only speak for myself of course. (Note, some of the posts from the old forum may be a it confusing since the pictures are missing):

There's more - I still have 55 pages old post listings to go through but I think you get the idea.

 

23 minutes ago, Patch Linden said:

..

and hopefully impact the tone of discussion in a positive and meaningful way.

The tone of discussion used to be positive here too but there is a limit to how long you can bang your head against the wall before frustration takes over. Most of us reached that limit long ago and then the discussions turned sour.

Edited by ChinRey
Typos
  • Like 3

Share this post


Link to post
Share on other sites

@Patch Linden I hope those internal discussions go well. I know several of Firestorm's tool, plus more great resource management tools, have been submitted to LL over the years but I'm not sure the viewer devs always understand or appreciate content creation issues and their broad reaching impact on SL, that's always been a source of frustration for me. At the same time I also understand resources for SL development are limited right now so I'm never sure if they're giving low priority to certain features, or if there really isn't enough time and people to put towards them at the moment with all of the other projects underway. A lot of the recent comments from the LL over the past year or so have made me feel there's a positive change in attitudes regarding viewer development and I really hope they are open to input from you and your team. The art department should be vital in viewer development.

  • Like 2

Share this post


Link to post
Share on other sites
7 hours ago, Luna Bliss said:

Klytyna, Penny, Optimo, and ChinRey…

In order to reduce lag do you guys think it would have been better for LL to change the way mesh and sculpts are implemented at the core of it all rather than asking content creators to rein in their triangles?

Any improvement in software performance would be nice of course but no, it wouldn't help on its own and it might even make matters worse.

There is no known limit to how much resources you can waste with poorly optimized content. Improved software without improved content will probably only mean there's more to waste.

We can't do anything about the software anyway, that's all up to Linden Lab. What we can and should do, however, is to make the most out of what we have

  • Like 1
  • Thanks 1

Share this post


Link to post
Share on other sites

As far as texture encoding goes, LL had taken the route of embedding different channels in the same texture when they implemented materials.

It can't be stressed enough, the materials in SL aren't just a normal and specular map with a glossiness slider value to govern the specularity. It's not as efficient as UE4 multi-texture embedding and their shaders, but hey, that's UE4...

In SL we have:

  • Diffuse color texture can embed a Glow map in its alpha in order to render a specific area of a texture as glowing, keeping the remaining pixels looking normal. Basically, a per pixel glow instead of a face property. ALPHA DOESN'T MEAN ONLY TRANSPARENCY!
  • Normal maps' alpha channel embed a Glossiness map, which the build tool helps with by giving us a multiplier (needed to convert a linear greyscale into a SL readable format)
  • Specular maps' alpha embed an Environment map, which basically turns on/off and modulates the metallic shininess (again with a multiplier to help converting a linear greyscale into SL readable data).

The current use of materials i could see around IS a waste of resources already, because Glossiness and Environment maps are neglected by most, but they're still accounted for during rendering, even if those textures actually don't have any alpha channel they're being treated as if they had one! LL has tried it, but the problem sits in the lack of proper documentation that anyone, without a solid technical background, can understand and make use of.

However, LL is lacking on one, fundamental aspect: keep the development going! This is the point i have been advocating against the BakesOnMesh project: we've got a shader that gives us some material input slots, KEEP DEVELOPING THAT and come up with newer features to save on more resources on ALL items types that can use materials, instead of reviving old features that, at current state, address only one aspect of content creation, the mesh avatar bodies, in a crippling manner, since it doesn't address the newer texturing features and procrastinating their implementation to an indefinite time, if that will ever happen.

  • Thanks 3

Share this post


Link to post
Share on other sites

Now that I've calmed down a little bit, it's time to reply to the whole of Patch' post, not just that one sentence that provoked me so much. I keep my first post though because I really mean what I said there and I think it's an important point: all the people who have contributed to this thread so far and several other active forumites have spent a lot of time and effort, trying to figure out the best ways to create efficient content for Second Life and freely share what they've learned with the SL community at large. If you've yet to see that, Patch, it's only because you haven't looked.

 

7 hours ago, Patch Linden said:

As mentioned, please stop and think that we are operating within the confines of the same potentially flawed systems you all very much love to bash and point out.

Yes, I at least understand that, and as I said in my reply to Luna, it is all about making the most of what we have.

 

7 hours ago, Patch Linden said:

On top of that you trash our content and content creators at a very deep, personal level, who are the very same people so passionately in love with the community and Second Life that they do this all for the love of it in many cases?

I can understand that but do you understand that the people who have been working hard to improve the content of Second Life for the love of this virtual world are hurt on a deep personal level by the way they've been treated too?

Also, there is a difference between voluntary amateur work and paid work. The moment money enters the picture, you have to deliver a certainl level of professionalism. Professionalism in your performance of course but also in the way you handle criticism - especially well founded and easily documentable ciriticism - no matter how harshly it is phrased and how unfair it may seem. I've certainly received my share of pepper both here and elsewhere but you know what: I appreciate it, it is very valuable for me. Oh yes it can be very painful, but in the end it is what helps me improve. It's a bit like going to the dentist, really: You don't want to, you really, really don't want to but it's so much better afterwards.

 

7 hours ago, Patch Linden said:

I tell my creators to build to the lowest common denominator and upload all LOD's on most things we do because the majority of Second Life users have what we refer to as 0 spec hardware.

Oh yes, keep doing that and tell everybody else too!

But do you believe optimizing for LoD factor 1 means high land impact?

There are numerous ways to reduce land impact and actual lag/load without sacrificing quality. Most of them should be listed in the posts I linked to above. I think I once collected them all in a single list too but I couldn't find that post in a hurry. I haven't posted all the practical details about them here of course, that would be way too much, but at least the basic principles behind each is explained. I think everybody who builds for SL ought to try their best to learn as much of it as possible. Anybody who builds for Linden Lab needs to be well and truly familiar with at least the most important ones. That's mandatory. It's not rocket science either, it's all about studying, learning, doing your homework.

I'm afraid that leads to some more criticism that may hurt somebody's personal feelings. I think it's neccessary though, it's well intended advice and I simply can't think of a gentler way to say it:

I haven't paid much attention to the premium gift thread, I just chimed in with one very obvious point. I hadn't even looked at the build until now. It's a delicious build, I really love it. The main srtucture is 161 LI though. I can tell you even after a quick glance that it could easily be done with a land impact below 100 and still be rock sol... - I mean crystal solid to somebody with their LoD factor set to 1.0. I ahve to study it closer to say more but I wouldn't actually be surprised if it was possible to get it all the way down to 40 with no noticeable quality reduction. I'm not talking about manipulating the numebrs here btw, I'm talking abut genunine reduction/redistribution of the load that would benefit everybody. That's what true content optimization is all about: not sacrificing LoD or other visual quality factors but rather making the build as easy to handle for the software as possible.

Edited by ChinRey
  • Thanks 3

Share this post


Link to post
Share on other sites

@patch linden

Those of us who are not so vociferous are looking forward as this work in progress we enjoy progresses and appreciate you are involved and your visibility mate. Bonus

Share this post


Link to post
Share on other sites
3 hours ago, Patch Linden said:

I did read Beq's wonderful blog post today about LOD (someone pointed it out to me), and also how that tool set works.  As a (dormant) content creator myself, I am fascinated by it.    It seems we are missing this valuable tool out of our viewer and that's something I'm going to discuss internally coming out of this.

(Empasis mine)

To me Patch, this speaks volumes. We have been actively using Beq's great tool to attack the bane of LOD abuse mesh for months and you only read about it today? That feature has been so widely talked about here, and it's caused me to go through and rebuild four regions in LOD2 friendly ways.

 

Edited by Callum Meriman
  • Like 1

Share this post


Link to post
Share on other sites
31 minutes ago, sirhc DeSantis said:

@patch linden

Those of us who are not so vociferous are looking forward as this work in progress we enjoy progresses and appreciate you are involved and your visibility mate. Bonus

I'm not vociferous, I'm quiet and shy. I even have it written in the profile of one of my alts so it must be true.

But seriously, us vociferous people are also looking forward to this. We've been looking forward and fighting for it for years and it is good to see that Linden Lab finally seems to start catching up. But they have to face some serious criticism because they are darned late to the party and they should have been the ones who organized it in the first place.

  • Thanks 2

Share this post


Link to post
Share on other sites
1 hour ago, Callum Meriman said:

Those numbers are crystal clear, and a huge part of why I am confused by the lab and your statement that you request LOD1 capable mesh. These figures don't back it up.

No, those figures do actually back it up. I explained it in the discussiion about that particular build but I can repeat it here.

They made the build LoD resistant by making the whole complex structure out of a single lump of mesh, big enough the LoD models will never actually be used at all (unless somebody has turned their LoD factor all the way down to 0). That is a very crude way of doing it though. It is the reason why the land impact is so high and since it essentially disables the entire LoD system, it also causes serious lag issues - or as serious as a mesh with less than 10 ktris can cause anyway.

Which brings me to...

No, I suppose there's no point in rubbing it in further.

I've worked most of my life as a musician and music teacher and some of the worst jobs we get are when we are given a student who has learned a few simple tunes and think they are seasoned pros. It breaks my heart every time it happens and it's much worse for the student than for the teacher. But they have to be taken down and forced to face the facts. It's the only way because you can't go forward unless you have a reasonble idea where you stand.

However - it's that dental phobia metaphor again - it's bad and it gets even worse but if you can make it through, it will end up far better than it ever was before.

Edited by ChinRey

Share this post


Link to post
Share on other sites
31 minutes ago, ChinRey said:

They made the build LoD resistsant by making the whole complex structure out of a single lump of mesh

I edited it out 30 minutes before your post, because I thought there was no need to bash a Linden more then he has been.

Share this post


Link to post
Share on other sites
17 minutes ago, Callum Meriman said:

I edited it out 30 minutes before your post, because I thought there was no need to bash a Linden more then he has been.

Ok, good point. I was actually adding a new paragraph to my post but it's even more suitable as a reply to yours. There is one thing we all have to remember through all this:

When a new tool is introduced to a work place, the employees are taught and trained how to use it. The Moles never got that training. I know this not only because it's so obvious but also because Michael once told me how they had slowly tried to unravel the Mysteries of Mesh. He was quite proud of it and for good reasons. All things considered they made remarkably good progress. If we compare the first Mole meshes to what they produced only a few years later, it's like night and early morning.

The problem is that they never got the help they were entitled to and that's not their fault. The responsibility lies higher up in the LL hierarchy, probably much higher up.

Edited by ChinRey

Share this post


Link to post
Share on other sites
1 hour ago, Callum Meriman said:

(Empasis mine)

To me Patch, this speaks volumes. We have been actively using Beq's great tool to attack the bane of LOD abuse mesh for months and you only read about it today? That feature has been so widely talked about here, and it's caused me to go through and rebuild four regions in LOD2 friendly ways.

 

Sorry, I can't be every where at once?   There are volumes of information out there that no one person can keep up with.  As I said, we can not use Firestorm, so I'm not sure that it's fair I should have specific interest prior to now without it being called out at me.  I can only grow and gain knowledge when someone calls it out to me or I stumble upon it myself.

As a former content creator myself, I get it.  All of it. 

Share this post


Link to post
Share on other sites

Hey! I´m Ancient Mole, the creator of the Crystal Fortress.

 

I´ve been reading along and would like to give you guys some insight into my work and thought process while creating it.

 

The whole Crystal Fortress Premium gift is made out of these 13 crystal clusters.

 

8779048519b79146619185e08acc8099.png

 

Each of those has exactly 108 Polygons and 3 Land Impact with the following LOD steps seen here:

 

0d5547158b02b29f10b150376339e26a.png

 

I have used those to build the first version of the Crystal Fortress.

I didn't save that version, so here is a simplified recreation of it using one of the clusters.

The Crystal Fortress has 9984 Polygons all together, so around 92 of those crystal clusters.

 

2ec2a2ad52b411c7d2e083bdd71322c5.png

 

As you see I ended up with 394 Land Impact which is way too much.

 

Just for a test I tried some lower LOD numbers seen here:

 

0e3648402ca987f635d5bf1bc608bf77.png

 

Each piece now now 2LI which results in 249LI and is still too much.

 

ec63f3e9dc31523da85e17c80309bd0a.jpg

 

I didn't take any pictures for this one but I also tried the following LOD steps: 108,28,9 and 2 with the LI per cluster being 1 now.

Eureka! We got the Crystal Fortress down to 118 LI! Lets look at it with the LOD Factor of 1 :

 

Little video of the results

 

This is not what I was looking for I fear and that would not pass QA.

 

So this is what I ended up doing to have the Crystal Fortress have 163 Land Impact while not collapsing at even LOD Factor 1.

 

I turned all of the Crystal Fortress into a single object which I then uploaded without any extra custom LOD´s.

 

This might be frowned upon in some cases but in this one, it allows us to have a decent Land Impact as well as it being resistant to collapsing.

 

Since the Avatar is basically located inside of the object it stays in shape even on the lowest settings.

 

That of course is only ok because the Crystal Fortress has only 9984 Polygons which are pretty easy to handle for your system.

 

Your basic avatar is already 7200 Polygons and that is being naked :D

 

I´d love to give you guys some more information if you got questions about how I or my Mole family do things!

 

Byesers for now :)

Edited by Ancient Mole
  • Thanks 3

Share this post


Link to post
Share on other sites
3 hours ago, Ancient Mole said:

As you see I ended up with 394 Land Impact which is way too much.

Yes, you have to be a bit careful how you split a mesh, there are certain "sweet spots". The basic ones are easy enough to find once you know what to look for and understand how LoD and land impact works. The more advanced ones require a bit of out-of-the-box thinking but they can be interesting challenges and big LI savers.

The most basic solution would have been:

  • Outer walls one single separate mesh. Use the same model at all LoD models.
  • Inner walls in segments aproximately 16x8 m. All LoD models zeroed out. That would still have kept the walls at full LoD everywhere inside the box but eliminated them as early as possible outside the box.
  • Floor and roof ideally 16x8 m sections too and with all LoD models at zero. You'd probably have to compromice a bit there though for practical reasons.

That would have been a fairly easy way to achieve a moderate LI reduction, probably down to ... say 120-130 or thereabouts.

There are more advanced solutions taking advantage of LoD models of course. I really can't go into details, at least not without having the dae file and spending a bit of time but here's one example:

3 hours ago, Ancient Mole said:

0e3648402ca987f635d5bf1bc608bf77.png

That particular section has LoD swap distances around 16, 66 and 132 m. That means you could safely have zeroed out the low and lowest models since they would never have been visible inside the box anyway. The lowest model doesn't add to the land impact either but the low one sure does.

As for the mid model, it would be seen halfway across the skybox and it is an awkward shape, so it would have taken a bit of work to find a good LoD model for it. The first thing to do would be to remove the faces completely covered by a neighbor crystal (actually you should have done that on the high model even). Next step would be to remove the faces from the back of the crystals - the ones facing the outer walls. Those faces will never ever be seen in anything but high LoD (and hardly ever then) anyway, so they are superfluous in the mid one. That may be all you could do without any visible distortions at all although you might have played a bit around with the smaller triangles to see what happened. But even so, it would be 84 tris for the high model, still about 40 for the mid (but with no visible distortion) and 1 or 2 for low and lowest.

As I said, I can't possibly give a full analyzis here but those two examples should give you some ideas what is possible with mesh optimization. So, study, experiment, learn and make sure every new build you make is better than the previous one! And since you're Ancient Mole, that means doing better than the Horizons Tree 1, the most technically sound Mole mesh I've yet to see.

Edit: On closer look, the section in your picture may not have been the one I thought it was, so the polycounts for the high and mid models may be different from what I thought. I'm not sure but in any case, the principle is the same.

And since I edit the post anyway: Even if you zero out models, don't leave it to the uploader. It's fairly easy to make a more efficient one manually. It's not that much difference but in a complex build it can actually add up to quite a bit of download weight. And looking closer at the pictures, I'm beginning to suspect you used uploader generated LoD models even for those meant to be seen. I won't ask and you don't have to tell but I think we both know that is a definite no-no for something like this. :P

Edited by ChinRey
  • Thanks 1

Share this post


Link to post
Share on other sites
15 hours ago, Patch Linden said:

I have yet to see or have any one suggest to us how you would use your masterful expertise to do any of this differently.

SL's LOD display side, the part that has to work fast and every user uses constantly, is mostly OK. Major problems mostly come from the mesh uploader. The uploader generates low-LOD models using an extremely dumb algorithm, throwing away arbitrary triangles. That's where those low-LOD models with holes in them come from. Those are not the fault of creators; they're the result of an obsolete algorithm in the uploader.

Mesh reduction is a hard problem. but there are solutions. Sansar bought a commercial solution. Microsoft is now selling mesh reduction as a service. (It's free if you let them keep a copy of your mesh). There are lots of algorithms in the literature and much open source code. I've played with some of it. (The academic code tends to be brittle - any flaw in the mesh, like a hole or a degenerate triangle, and it blows up. Work would be needed to use that.) There's lots of stuff you can do, like convert fine detail into bump maps to get rid of small-scale sharp angles before doing the mesh reduction. You can recompute the topology and generate new textures. You only do this once, so it's far more efficient than anything done at display time.

A decade ago, people mostly did LOD content generation by hand, but there's been progress since then. Most game assets are now ground through an automated pipeline to generate the various LOD models. In the game world, this is no longer the job of asset creators, except for important hero models.

SL really, really needs an impostor system. Something that generates a low-LOD model by taking pictures of the high-res model and pasting them onto very simple geometry, like planes and cubes. I took a look at the code in Firestorm. The viewer's uploader almost has enough machinery inside to do this, because it can render a model being uploaded in a little window for preview purposes. OpenGL can take a picture of it from any scale and angle, and that can be used to create a texture for the impostor. But this will look good only for objects that are flat or rectangular. SL doesn't support billboard impostor display, where you have eight planes around a vertical axis, each with a flat picture of the object, and only the one facing the viewpoint is drawn. That's a basic LOD feature missing from SL.

(An amusing impostor plan - instead of showing water for sims not being rendered, show the world map.)

Share this post


Link to post
Share on other sites
1 hour ago, animats said:

(An amusing impostor plan - instead of showing water for sims not being rendered, show the world map.)

The rental companies with sky sign ads would love that. :P

A better LoD model generator for the uploader, better handling of bump maps/normal maps, impostors, those would all be great to have of course but as Patch said, we all have to work within the limits of what we have.

SL's LoD handling system was seriously compromised by the haphazard way sculpts were implemented, further messed up with the introduction of mesh and then it went down the drain with fitted mesh. Fixing it all to implement all those features might well require a complete rewrite of all the visual assets handling code both server and client side. It might also break a lot of existing content or at least make it so redundnant that nobody would touch it. Ideally this should have been done in 2008 and it definitely should have been done in 2011.

Incidentally, that ties up with something I wrote in a previous post:

8 hours ago, ChinRey said:

...

The Moles never got that training.

...

The responsibility lies higher up in the LL hierarchy, probably much higher up.

What I think happened (Patch may or may not confirm or deny this) was that the developers and LLDPW didn't communicate. The developers were unable to take advantage of the unbiased end-user perspective active, professional content creators employed by LL could have given them and the Moles were not informed about the progress of the development.

The result was that the developers misunderstood quite a few things and they also failed to provide relevant documentation, partly because they didn't really know how to use this tool for practical building, partly because they didn't really understand what was relevant and what wasn't.

When mesh was launched, the Moles - like all other builders - were thrown in at the deep end, being asked to handle a brand new but flawed tool with no manual.

Fortunately, some forumites did have the time and skills needed to unravel it all and pioneers like Drongle soon turned this forum and its Mesh specific sister into the ultimate SL mesh resource. There really is no other place like this. This is where all the info about SL mesh relevant to a builder can be found. If you can spend a few ages shifting through all the posts that is - or if you are actively following the forum so you can pick up the tidbits of info as and when they turn up.

The Moles are not forumites - except Rolig of course but she's mainly a scripter, not a builder (although I have to say I've seen some really clever prim builds by her Mole alt). They didn't have access to all this info. Nor did they have time to do the research themselves. They had deadlines to keep and schedules to follow. So they were lagging behind. And even worse: they didn't realize it. Until now that is - or until very soon. Let's see what tomorrow brings.

As for who's responsibility it is, it's the CEO's job to ensure the different departments communicate with each other. He doesn't work here anymore so it doesn't really matter.

Edited by ChinRey

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...