PE and full-region mesh builds.

[Gaia Clary]

---

Medhue Simoni wrote:

What will bring more people are better full sim builds, more interactive environments, more fun stuff.  If you ask me, all the limits put on mesh is completely counter productive.

---

 Are you realy sure about that ?

 

• Has anybody ever tested, how a complete environment import would be charged ? 
• Has anybody ever even tried a full environment import ?
• And applied all possible optimization methods to the included meshes ?

  To see how far we could get ?

  e.g. split a huge tree into several (reusable..) parts to reduce render costs ?

  create linksets and set children's physics to none where  applicable... etc...

I have not done this. I even have no idea how to organize such an environment upload. Just thinking about recreating the different LOD's for the environment makes me anxious about the overall effort behind such a project. So maybe talking about full sim uploads is just not a practical method...

But if it where the way how we should do Sim development with meshes... then it would be interesting to see if we could get away with that approach...

 

[Marcthur Goosson]

And what about internet speed.

Doesn't that have a huge impact on what we are able to see and stream.

I can have the best graphic card but al the data have to reach my PC too.

Overhere with 1mb internet, things aren't that fast.

And 1 mb is no exception in the rural areas of this western country.

And yes all those fantastic PC games with super realistic graphics.

Don't they put some gig's first on your hard drive?

 

 

 

[Drongle McMahon]

"Is the avatar the way to success? I think not. IMHO..."

Interesting assertion. As a rather neglected avatar, I obviously share a greater concern for the wider environment, but if you consider the volume of commerce and creativity devoted to avatars and their adornment you might come to a different opinion. It clearly is very important to very many users, possibly the majority. I wonder what the turnover in attachments vs static objects is. Anyone know? Of course that may not be a good indication of what affects retention, but....

[Vivienne Schell]

Tapple, there is some kind of table on the "typical" fps of grahic cards in SL here:

http://wiki.secondlife.com/wiki/Typical_Frame_Rate_Performance_by_Graphics_Card/GPU

While i do not know what "typical" is. Going to a region crowded by 20+ fullly cladded avatars and full of particle wizardry and whatever else within deafult viewer range certainly is not "typical". But MAYBE the numbers are some kind of median performance. Given the fact that aynthing below appr. 20 fps is recognised as "stutter" by the eye, only about a good quarter of SL "typical" user GPU´s qualify for a somewhat enjoyable fps experience, at least in MOST occasions. I´d really like to see some more details there, like which GPU´s are preferred by they users (How many users actually get an average of 25 fps + and how many not). Nevertheless, this somehow limited table might serve as a pole-hint on the difference between Crysis2 users and SL users.

I am not sure how far bandwidth affects fps performance, but maybe someone knows?

[Vivienne Daguerre]

That was a helpful post and reference Vivienne.

[Vivienne Schell]

Vivienne, as content creators (mesh or not) we ALL are swimming in some kind of black laguna if it comes to design stuff for a certain target platform (SL typical GPU´s mainly). It would be helpful if LL would - at least - feed us with some more detailed info there. So that we know for what kind of existing user hardware we are designing our content and what exactly influences what - at least the typical fps performance - in which way. The more details, the better. So far we only can speculate on this decisive topic. Even given the strong suspicion that *most* commercial content creators only care about the linden dollar, looks and prim count - some kind of info on "recommended" specifications and target hardware in content creation would be more helpful to some of us than non transparent arithmetic formulas.

[Vivienne Daguerre]

I was not being sarcastic Vivienne. I truly found the link to the webpage showing SL performance on the top graphic cards to be interesting and helpful.

[Tapple Gao]

I have a 2002 laptop I frequently run SL on. If anybody ever wants my performance numbers on a certain region, just ask

[Tapple Gao]

I don't see fps as that big a deal as long as it's above 6 fps. that's what anime is rendered at, and it has quite a following

[Feynt Mistral]

What saddened me at first was that I had come back to the mesh discussion after several months of RL issues to hear the maximum size was constrained to 64m^3.  Ok, I can understand that, there aren't many builds which need more than that and those that do could be cut up (unhappily, as my current project requires a 170m tall prim, and if it were one piece the dimensions would be comparable).  Now what I'm reading is the tree I'm working on will be impossible due to the PE cost on something so large.  I fail to see how a non-physical (prim physics shape) model with a reasonable number of triangles (likely not more than 100k with foliage, and quite possibly much less) will impact someone more than the traditional hair, shoes, and jewelry we all are subjected to daily.  What does it matter whether 10k polys are compacted into a 1m^3 area or stretched out over 64m^3?  It's the same complexity to stream and render.

[Medhue Simoni]

Well, I did not really means with our current PE, cause that is totally impossible. Tho you can import whole mesh scenes. So, like a whole mesh dining room set can be all imported as 1 scene, and each mesh might be small enough to not break the high PE barrier.

Even if I was meshing up a full sim, I'd probably not want to even render all that in 1 scene, in 3ds Max. If the PE levels lower, I will most likely be doing a full sim within the year tho. I need to buy another sim for our combat system, and It would really be kick butt to design all the structures, mountains, bridges, and whatever, all in 3ds Max.

I used to make MODs for Far Cry and liked to create my own unique combat areas for other online players to download and use, all on the original Cry-Engine. I never learned how to import models, but they had a great array of interactive items and structures to use. Would be cool to do the same thing but now make every item, and make them as interactive as I can. Sadly, LL doesn't want something like this,:smileysad:

[Medhue Simoni]

---

Drongle McMahon wrote:

"Is the avatar the way to success? I think not. IMHO..."

Interesting assertion. As a rather neglected avatar, I obviously share a greater concern for the wider environment, but if you consider the volume of commerce and creativity devoted to avatars and their adornment you might come to a different opinion. It clearly is very important to very many users, possibly the majority. I wonder what the turnover in attachments vs static objects is. Anyone know? Of course that may not be a good indication of what affects retention, but....

---

I guess, I should clarify. Yeah, I think every1 knows that avatar related stuff sells alot. The brilliance of SL, is fully customizable avatars, way beyond what any other game can do. Maybe not in total quality, which is limited by the tools, not the creators. I have praised this aspect to many potential SLers, and constantly amazed at how creative the creators are.

This comes back directly to my point. No matter how much better mesh attachments are, it is not really new or adding more to what the user can do or get. There are sculpty versions of most attachments already. To the end user, they are just getting better stuff, not really a much wider variety, if that is even possible in SL. When you think about it, even if the creators made every single possible thing you could attach, would that really change the appeal of SL? Full mesh avatars will be a step up, but this will not overtake all of SL.

Lets hit my point head on now. What SL needs, IMHO is more fun things to do, and a better way to rank and find those things. Mesh will bring the visual quality up, on avatars now, but LL hurts anything that uses a script. If mesh is so bad with a script in it, then how bad is a sculpt with a script in it?

Right now, I have an interactive system which is moving, rezzing, animating, and wearing sculpts like crazy. Waiting for a sculpt to load is ridiculous, but it all seems to function well. I can't really test this with mesh on the beta grid, cause I don't have enough prims, nor the room to do it. If mesh is going to create tons of lag, I will know it for sure with this system. But, I have to wait for release to actually test it.

Oh, and my original point also had something to do with how LL makes money on land, but only allows mesh for avatars. How strange is that? Mesh would probably make the biggest visual improvements on environments, or I think it would.

[Drongle McMahon]

"Lets hit my point head on now. What SL needs, IMHO is more fun things to do, and a better way to rank and find those things. Mesh will bring the visual quality up, on avatars now, but LL hurts anything that uses a script. If mesh is so bad with a script in it, then how bad is a sculpt with a script in it?"

I agree with everything you say here, especially the quoted bit. You must be close to the truth when you say just about every possible attachment is already available. A lot of these are incredibly inefficient, things like 12-sculpty shoes! Aarrgghhh!. Mesh certainly has a big potential to save client lag and the blob-wearing effects. It will be interesting to see whether that is sufficient to motivate replacements, and if so, how fast that will happen. Maybe a well-informed official publicity campaign would accelerate it.

Whoops. Wandering off my own topic. Sorry.

[Medhue Simoni]

You know, the sad part is, I look at my beautiful mountain, which has 3 levels, and a cave to walk around in. This I purchased from Insky on the marketplace. The work that it takes to make sure you can walk around on it, is crazy and adds dozens of prims to the build. This, as a mesh version, which would be a hundred times more efficient, is not possible because of the PE.

Kinda funny isn't it. I'm an animator, which you would think would be all about the avatar. Environments are organic tho, unless the structure is man made. Maybe the threshold for the PE jump could be push back a bit to make it so that objects smaller than 10x10x10 are few enough prims to be feasible. Currently, it seems that anything larger than 3x3x3 starts to get crazy.

[Drongle McMahon]

Make sure you try them out on the Mesh Experimental server*, which has the diameter-instead-of-radius bug fixed. The cost rise at half the rate with increasing size, although they eventually reach the same plateau. It won't help with very big things.

*Sandboxes 20,21,27,28.

[Haye Abbott]

Why not harm the existing platform? The existing platform is horrible, and some harm to it would do everyone good.