Slow Loading Terrain Textures

Ask questions about creating worlds, using worlds, etc.
User avatar
Koshari Mahana
Posts: 358
Joined: Fri Feb 06, 2015 8:14 am
Has thanked: 279 times
Been thanked: 497 times
Contact:

Slow Loading Terrain Textures

Post by Koshari Mahana »

I have 64 regions and it takes so long for the terrain textures to load all the way up to the land in the north. The default is sand so that's what is seen until the grass loads (it loads the terrain textures south to north). On a single region or even up to 4x4s it took time but because they are smaller it didn't take as long to reach the northern land. Is there any possible way to speed up the terrain texture loading time or is it just the way it is?
These users thanked the author Koshari Mahana for the post (total 3):
Ilan TochnerSnoots DwagonChris Namaste
Image
VISIT FOUR WINDS ON THE MARKET
VISIT FOUR WINDS INWORLD: hop://grid.kitely.com:8002/Coopersville/909/907/21
User avatar
Ilan Tochner
Posts: 6503
Joined: Sun Dec 23, 2012 8:44 am
Has thanked: 4942 times
Been thanked: 4454 times
Contact:

Re: Slow Loading Terrain Textures

Post by Ilan Tochner »

Hi Koshari,

The viewer decides the order it asks for assets, including terrain textures. There isn't much we can do about it from our end. You'll note that the more assets that there are in your world and the farther away that they're located from your camera view point the longer they'll take to load.
These users thanked the author Ilan Tochner for the post (total 3):
Koshari MahanaSelby EvansChris Namaste
User avatar
Snoots Dwagon
Posts: 422
Joined: Mon Jul 30, 2018 9:45 pm
Has thanked: 442 times
Been thanked: 779 times

Re: Slow Loading Terrain Textures

Post by Snoots Dwagon »

I have found that going to make a cup of coffee mysteriously makes textures load faster. :mrgreen:
These users thanked the author Snoots Dwagon for the post (total 2):
Koshari MahanaShandon Loring
~~~~~~~
I'm a dwagon in real life too. (Ask my sister, who totally agrees.)

~~~~~~~
User avatar
Koshari Mahana
Posts: 358
Joined: Fri Feb 06, 2015 8:14 am
Has thanked: 279 times
Been thanked: 497 times
Contact:

Re: Slow Loading Terrain Textures

Post by Koshari Mahana »

Thanks!
Ilan Tochner wrote:
Tue Feb 16, 2021 4:27 am
Hi Koshari,

The viewer decides the order it asks for assets, including terrain textures. There isn't much we can do about it from our end. You'll note that the more assets that there are in your world and the farther away that they're located from your camera view point the longer they'll take to load.
Image
VISIT FOUR WINDS ON THE MARKET
VISIT FOUR WINDS INWORLD: hop://grid.kitely.com:8002/Coopersville/909/907/21
User avatar
Koshari Mahana
Posts: 358
Joined: Fri Feb 06, 2015 8:14 am
Has thanked: 279 times
Been thanked: 497 times
Contact:

Re: Slow Loading Terrain Textures

Post by Koshari Mahana »

:lol: :lol:
Snoots Dwagon wrote:
Tue Feb 16, 2021 4:35 am
I have found that going to make a cup of coffee mysteriously makes textures load faster. :mrgreen:
Image
VISIT FOUR WINDS ON THE MARKET
VISIT FOUR WINDS INWORLD: hop://grid.kitely.com:8002/Coopersville/909/907/21
User avatar
Koshari Mahana
Posts: 358
Joined: Fri Feb 06, 2015 8:14 am
Has thanked: 279 times
Been thanked: 497 times
Contact:

Re: Slow Loading Terrain Textures

Post by Koshari Mahana »

Oh, I should add... objects and object textures are loading so much faster than the terrain textures.
Image
VISIT FOUR WINDS ON THE MARKET
VISIT FOUR WINDS INWORLD: hop://grid.kitely.com:8002/Coopersville/909/907/21
User avatar
Snoots Dwagon
Posts: 422
Joined: Mon Jul 30, 2018 9:45 pm
Has thanked: 442 times
Been thanked: 779 times

Re: Slow Loading Terrain Textures

Post by Snoots Dwagon »

I am afraid as Ilan says this enters into the area of Firestorm... and the SL Viewer foundation upon which it was written.

Texture rezzing priority is a complex concept at best. If I were to make a stab at it, I would try to follow a logical course:

1. I would first of all go back to the way SL originally did things... using a single texture to rez general shapes first. I would probably use a more modern bright cyan outline on medium blue forms. (Think TRON.) Everything everywhere rezzes fast and mean, using that one texture. That way people can at least see the shape of the world around them.

2. The Avatar is rezzed next... quickly and every time-- no failures (short of Internet collapse). Enough of this orange fog nonsense. When a user logs in the first thing they should see is their avatar rezzing quickly and accurately. That includes sculpties and meshes.

3. ON DEMAND textures would immediately take priority over all other texture rezzing, since those textures are having direct interactivity with the user at that time. Whether that is the user rezzing a box to ground and applying a texture to it, double-clicking a texture in inventory to see on his screen, or clicking an already-rezzed vendor to see the next product, that visitor needs to have the texture they want right now, preferably within 1-2 seconds.

4. Then rez the ground, since that's the foundation of the world and where we're going to be walking.

5. Then starting with the items closest to the avatar, begin rezzing out in a circle until reaching his maximum draw distance. If that draw distance is set too high to properly rez the world around him, automatically scale it back so that what he can see, is seen.

6. Calculate new textures to rez as the avatar moves, based on that same draw distance scale. World mechanics. Has to work, every time.

That these things haven't been implemented after all the years LL worked on the base code and all the following years the Firestorm team has worked on the code... appears in my book as MAJOR FAIL (sorry, no punches pulled in this one). I've been arguing this one direct to LL (and later, Firestorm) since 2005, pointing out they had a major texture bottleneck in their coding logic. Come to find out, they had several: 1) Reloading the same textures due to faulty cache access, 2) failure to prioritize texture loading, 3) failure to recognize on-demand textures as highest-priority, 4) failure to fully rez textures (resulting in "fuzzy" textures).

Texture rezzing is one of the most important user-perceivable functions of a grid. Poor texture loading has direct negative effect on user retention. If the textures don't rez, nothing works. But if properly prioritized (as we read above) and the system cache working properly, even if all textures haven't rezzed the textures that are of greatest interest to the user will rez, quickly and usefully.

If it takes a full minute or more for textures to rez clear across the region, so be it (if they even need to rez at that distance). The Viewer should be far more focused on the textures the user is trying to get to work right now: 1) avatar appearance 2) whatever object it is s/he's working with at the moment 3) The immediate world around the avatar (especially the ground and perceivable shapes). These are items of immediate user concern.

Failure to tackle this major issue with due diligence is something I see as a significant failure in Viewer team logic. Every graphics-intensive system on the planet knows that textures must work. If we would want do discuss the "Top Ten Things that Cause People to Leave Kitely".... textures not loading quickly enough would be-- from my experience-- in the top 5 items of the list... possibly the top 3. When the 2-D don't work, the 3-D don't work. Is it any wonder new users don't stay and invest their limited funds-- on a perceptibly faulty system?

I know folks have heard this before, possibly from me. I've been bringing this up since 2005. A well-known tech joined with me back in 2007 and we stood back to back on SL JIRA and fought will Lindens (and sycophants), proving with established data that the SL texture system was broken. But they outright ignored the data glaring right at them; in their minds we were newbs who didn't know anything (despite our decades of real life coding expertise).

It's likely I will continue to bring this up again until it gets through that be that some of the greatest problems with user retention is caused because the textures do not load correctly, logically, in the right priority... and a texture we just saw 5 seconds ago for some reason have to re-load from the asset server rather than our computer cache.

"Some people's systems can't handle it." We hear this regularly. I sympathize when I realize some users are still using systems their grampa used. I feel their pain... but if one is going to use a modern-day high-level virtual world, having a computer system, graphics card and internet feed to handle it is part of the criteria. If they can't update their equipment, maybe it's time to find a different, more affordable hobby (or alternately, turn their graphics settings down to the capabilities of their system). For the majority of us who have modern, 64-bit computer systems, good graphics cards and high-speed Internet... we need and expect the software foundation to warrant the equipment and grid-cost buy-in. Those who use World of Warcraft, Skyrim, Portal and other immersive environments and then come here and can't even get their avatar to rez... they're going to spend about thirty seconds looking around and then kiss us off and go find something more rewarding to do with their quite-capable computer systems.

That is a reality of computer life.

Please excuse me for speaking so boldly on this. It's a "Squeaky wheel" thing. Unless I and others speak on this and keep speaking and keep speaking... we can be guaranteed nothing will be done to get textures to load properly and in the proper priority.

How seriously are people going to take a system where textures and avatars don't rez properly? Do we want this system to grow or remain stagnant because the rest of the computer world is passing us by? Do we make required changes and fixes... or continue to do things as they've been done for the last 20 years?

* We log in and our avatar remains a cloud of orange smoke for a lengthy period of time
* The avatar doesn't appear at all, remaining totally invisible, forcing the user to wear another av
* We use a vendor and it takes 20 to 30 seconds for the next product to rez clearly

New visitors coming in from other systems see this and they simply leave, having little tolerance for amateur-hour systems that have had the same bugs for as long as I've been on virtual worlds. Kitely has done more to change such things than just about any system I know, but there's only so much one grid can do without total cooperation of the viewer devs... a group that seems to consider Opensim a second-class system. ("For us to do that we'll have to get Linden Lab's backing" is getting to be a very tiresome response.)

I'm just sayin'. I'm a retired businessman with a 3+ decades habit of direct communications. The job either gets done, or it doesn't. Textures either load properly, or they don't. How many more years is it going to be before we start taking the Viewer situation seriously?. To be very honest, I find myself asking these days if I could be making better use of my time. If I'm asking that question, that gives us some indication of why retention industry wide is so low. If Second Life worked as it should, if Opensim worked well, we'd be growing-- significantly so (especially as these days of "Covid shut in" quickly approach a full year).

Want to improve retention? Improve the product. Improve texture rezzing performance. Give people a reason to stay.

And that's it for the night 'cos it's 2:46am... which may help explain the no-holds-barred podium-post. ;D
Last edited by Snoots Dwagon on Wed Feb 17, 2021 3:07 am, edited 3 times in total.
These users thanked the author Snoots Dwagon for the post (total 3):
Selby EvansAda RadiusKoshari Mahana
~~~~~~~
I'm a dwagon in real life too. (Ask my sister, who totally agrees.)

~~~~~~~
User avatar
Koshari Mahana
Posts: 358
Joined: Fri Feb 06, 2015 8:14 am
Has thanked: 279 times
Been thanked: 497 times
Contact:

Re: Slow Loading Terrain Textures

Post by Koshari Mahana »

Thank you so much Snoots for your thoughtful in-depth analysis of the situation. The problem is far deeper than I'd imagined. Is Firestorm code (this year) opensource or only the original SL viewer code from ages ago? I know absolutely nothing about code, I can make a website and that's all I can do. I wonder if there is someone out there that wants to create a newer, better viewer, more 2021 quality? I understand the devs at Firestorm are volunteers so it's somewhat understandable that they have different priorities than we do. I've tried other viewers and I don't like them. Even though Firestorm is imperfect it's better than all of the other options out there (for me personally). Is this a problem that can be solved or will these types of virtual worlds become obsolete? I hope not, I hope it can be fixed.
These users thanked the author Koshari Mahana for the post (total 2):
Snoots DwagonAda Radius
Image
VISIT FOUR WINDS ON THE MARKET
VISIT FOUR WINDS INWORLD: hop://grid.kitely.com:8002/Coopersville/909/907/21
User avatar
Ilan Tochner
Posts: 6503
Joined: Sun Dec 23, 2012 8:44 am
Has thanked: 4942 times
Been thanked: 4454 times
Contact:

Re: Slow Loading Terrain Textures

Post by Ilan Tochner »

Hi Koshari,

Firestorm is the most commonly-used third-party viewer for both Second Life and OpenSim and it uses the latest Second Life viewer code for both.

There have been multiple attempts over the years by various groups to create a new viewer codebase from scratch but all such attempts eventually encountered the reality of the amount of work that is required to do that. As a result, no one has created a viable viewer replacement for OpenSim that isn't based on the Second Life viewer code. The best people have done is creating proofs of concept or viewers with very limited functionality.
These users thanked the author Ilan Tochner for the post (total 3):
Snoots DwagonAda RadiusKoshari Mahana
User avatar
Snoots Dwagon
Posts: 422
Joined: Mon Jul 30, 2018 9:45 pm
Has thanked: 442 times
Been thanked: 779 times

Re: Slow Loading Terrain Textures

Post by Snoots Dwagon »

I fully agree with Ilan. My (somewhat lengthy) post above was merely to keep an idea alive: Opensim needs its own Viewer. Whether that idea ever sees the light of day... that's the ultimate question. (And this post is likely going to be equally long, because shut-in is driving me nutz.) :lol:

Ilan is 100% correct: it's been tried. And some of the viewers were pretty good (in fact, I'm rather sorry to see them no longer functional). Many died when SL instituted mesh; the viewers were prim-oriented and mesh was just too much of a change. Others died out because they weren't funded, didn't have a solid business base, were private attempts that were frankly doomed to eventual failure. So yes, as Ilan says, it's not that it hasn't been tried. It has... several times.

But what hasn't been tried (to my knowledge) is the entire Opensim community banding together and starting an actual not-for-profit, paid-employee company with an experienced, competent management team and a team of skilled coders that could write a viewer specifically for Opensim. They could start with the SL Viewer, sure. Why re-invent the wheel. But their goal would be to go in and "clean up the mess" and turn that viewer into what it should have been in the first place.

In place of individual devs deciding what's best for the community... the community would decide what's best for the community. That's where the Management team comes in: they take direct feedback from the community (votes if necessary) and decide what is to be implemented. Then the devs, being paid for their work, make those implementations regardless of personal whims... because they're being paid to do so (just like any other professional software development company).

Do we implement default to WASD movement (or at least put a prominent BUTTON on the screen to implement such)?

They fix the texture bottlenecks and prioritize texture loading as it should have been from the start.

They fix the years-old megabugs that are fully documented on several JIRAs... and they don't mess around doing so in their spare time. These coders will be full time employees.

Rather than trying to copy Linden Lab step-for-step and having OPENSIM decisions made by LINDEN LAB... Opensim separates itself entirely from Linden Lab, cutting the apron strings and not being afraid to go beyond and leave SL sitting in the dust (if at all possible). The goal: not to equal Second Life, but to do better.

Are these pipe dreams? Perhaps. Ilan is a realist and frankly, I am too. We both know how unlikely such a project is. However as someone pointed out... look what happened to Blender when they monetized the system. The software is now used to make movies... and it's free. How? Because both businesses and users like us saw the value in such a significant project, and were willing to finance it. Whether it's $1 or $5 a month donations from someone on a low budget... or a whopping grant from a business that really really wants a viable virtual world that runs like greased lightnging... there are ways to finance such projects. The trick is... it requires people who 1) Know what they're doing 2) Are well acquainted with Second Life, Opensim and the difference between the two 3) Have the brains and energy to spearhead such a project and 4) Check all egos and personal opinion at the door and have their minds open for new possibilities.

So that's the thing. Most users are aware that Opensim needs its own viewer. Many users will agree we need to break away from any control of Linden Lab... completely. Many users will agree that Opensim has potential far beyond the corporate greed of a shareholder, profit-driven company.

Please don't get me wrong. Second Life is amazing. The vision of that project was at the same time amazingly expansive, yet regrettably limited. The attitude however, has been totally profit-centric, even if it was to the detriment of their customers.

Opensim however, is not profit or greed driven. It was designed from the beginning to break away from such concepts and be heavily user-oriented. That's the prime difference. For a long time Opensim was nearly unusuable due to lag, bugs, incomplete functions etc... but they kept plugging on. In October 2018 (I remember it clearly) something happened. They released a long-awaited update and Opensim finally triumphed. We finally had a system that was technically equal to Second Life. It was like flipping a switch from "Okay, it's kinda working" to "Wow... what happened?"

I had just joined Opensim seriously a couple of months before that. I watched this happen with my own eyes. It was amazing.

The thing is... Opensim was not Linden Lab code. Many people would have considered the project impossible and doomed to failure... but look at it now. So that brings one question, and is why I left the above post:

If such could be accomplished with Opensim... why not the Viewer? In fact we have a head start on the Viewer because we're not starting from scratch. We already have a viable viewer; it just needs fixed.

So what we really need is a team of people who are willing take Linden Lab and Second Life out of the equation... and make Opensim its own super-dynamic system... equal to any immersive environment on the market. I'm not the person to do it. But there are those out there who have the ability... if they have the will, the drive, and the belief that this needs to be done. Set up a professional not-for-profit company. Arrange monetization. Hire paid programmers. Assign projects to a full-time staff. And see if we can accomplish in the Viewer and Opensim what has been accomplished in Blender and other projects that chose this direction.

So yes, I agree with Ilan. It's been tried and failed. But I sure would like to see someone surprise us by founding a Viewer "NASA"... intent on landing on the Moon. : )
Last edited by Snoots Dwagon on Wed Feb 17, 2021 3:53 am, edited 1 time in total.
These users thanked the author Snoots Dwagon for the post (total 2):
Ada RadiusKoshari Mahana
~~~~~~~
I'm a dwagon in real life too. (Ask my sister, who totally agrees.)

~~~~~~~
Post Reply