Well, the open beta is open for business, but you already knew that. So a few things will probably happen from here.
First up, the beta grid is going to be awash with the curious, who largely just want to see what mesh objects look like in-world. The poor old beta grid has already been straining a little at the seams.
That shouldn’t last too long, though. Interest in mesh objects will probably diminish pretty rapidly, except among the few hundred people who are actually intending to use it routinely.
Many of those will probably learn just enough to make simple pieces and components, which will doubtless take some of the rough edges off of a lot of prim-based content (so to speak). A few will go the whole hog, and get involved in more detail.
Building with prims is almost trivial by comparison to 3D mesh modelling, when you get down to it. Even so people who actually create more than the most simplistic prim-objects in Second Life are a relative minority. Scripters are rarer, and texture artists are rarer still. The least populous grouping are those who work with sculpties.
3D modelling is an even more rarefied skill. Some users who currently work with sculpties will probably make the jump to working with meshes, and some won’t. While pretty much anyone can learn to do it (in much the same way that most people can learn painting, welding, or electrical engineering) most people won’t learn to do it.
As with any skill, we assess the cost of acquiring it versus the perceived benefits. If the benefits don’t seem to noticeably outweigh the costs, we’re just not going to try. That’s how people work. The learning curve of 3D modelling is a steep one.
All of that said, it’s a feature that is actually fairly well in-tune with Second Life’s top content creators. While males generally develop 3D spatial reasoning that is useful for navigation, females tend to develop 3D spatial reasoning that is more useful for 3D modelling than their male peers. With Second Life’s top content creators trending primarily towards women, the learning curve for them is likely to be somewhat less steep than for the boys.
There are a couple of potential advantages though. Pulling meshes for prim-based objects out of the Second Life viewer is not egregiously difficult, which could give some prim-based creators a head-start in developing more finely detailed and clean composite meshes based on their existing work.
Even so, I’m not expecting prim-based products to disappear in a rush when the whole shooting match goes live. More, I’m expecting simple mesh components to become a part of prim-based content, alongside a smaller number of mesh-only products.
Yep. That’s a point I’ve been trying to stress to people overly worried that Second Life might radically change for one reason or another due to meshes. Any charm found in prim building isn’t going anywhere, because possibility != it’s going to happen. Most people probably aren’t going to bother exploring the full possibility meshes bring if it isn’t worth the effort.
On the 3D spatial reasoning bit, I thought men were ahead due to avoiding becoming t-rex food while women plain didn’t have any, with the whole driving thing. Good with colors for all the poisonous berry picking though, I think thats what you meant.
I think you’re underestimating the fact that meshes are, in many aspects, *easier* than sculpts. Sure, there are more options, so the sheer possibilities are daunting, but we also have many more tools, on very different levels, to work with.
I have a feeling that many who threw up their arms in disgust over sculpts and never bothered to learn the bizarre quirks and rules – including having to learn Blender or Maya (and shell out for the latter) – can now actually get in the game with things like, say, Sketchup.
It’s true that sculpts are very persnickety – and Sketchup may well help considerably. Can you imagine what a nightmare it would be if Linden Lab tried to implement an in-world mesh-editor?
Something I’d like to see is a modification of the LLSD export tool in Imprudence (and presumably other TPV) to have it export Collada files.
Like you say, building with prims is trivial by comparison to 3D mesh modelling, so being able to build a mesh in the viewer, export it to Collada & re-upload would be incredibly useful.
I do suspect that converting a linkset to a mesh would be easier than converting a linkset to sculpties
p.s. such an export tool would negate the need for an in-world mesh editor.
I’d be really sad if LL would put resources/work/time/money in an in-world editor. The few hundreds who will create regularly, will have nearly no impact on concurrency…even more, most of few create already off-line when making sculpties.
However, maybe such an editor comes later (but i doubt it at the moment). More likely that some TPDs build an in-world tool that allows editing of local meshes, and temporally hold the mesh in the users viewer…but somehow i doubt it, too
Just curious when mesh will go live…from what i’ve heard: ‘months’, notice the plural
I’m estimating by the end of December, but I’ve got no information to base that on other than a hunch.
Assuming that there are a the same number of people building and using Mesh as there are using Sculpts today, what happens to lag?
http://www.flickr.com/photos/gracemcdunnough/4908837934/
@Grace: With any luck, things will get faster, since sculpts force a lot of suboptimal workarounds. The faces and UV mapping of meshes in particular can cut down *significantly* on texture memory, where sculpts need to uniformly wrap the whole object in one texture, at the highest detail/resolution needed at any one spot on the sculpt.
Side rant: Texture size is really the hidden killer in SL. Too many creators think bigger is better, since more details will be visible, and regularly texture with 1024×1024 pixels bitmaps. One thing is the rather insensitive belief that just *their* texture needs to be high detail no matter the performance cost; another is the somewhat counter-intuitive fact that too-large textures in fact looks noticeably worse! -SL’s on-the-fly resampling is fairly rough when it cuts down on size, so starting with a texture smoothly prepared for a size closer to the end result shown on screen gives better results. Try it; it’s easily visible to the naked eye.
@Tali I usually embrace luck with reckless abandon.
I would *guess* that most people who are capable of tossing down prims in SL have:
1) no idea what texture loading is or why they should care (and therefore don’t check),
2) don’t know how to (or can’t) scale down a texture if they do know,
3) don’t even think about the bigger picture because there is no motivation or trigger in place to help remind them.
And you lost me on the last sentence or I would try it!
Try making, say, a sign with some text, at 1024×1024.
Upload that, use it to texture a prim, and look at it from a distance where it’s drawn as, say, 200 pixels on your screen.
Then try resizing it in Photoshop or Gimp with a nice, smooth resizing algorithm to, say, 256×256 pixels.
Upload that, use it as texture, and look at it at the same distance.
Odds are good that the 1024 version comes out strangely jagged compared to the 256. (And the 256 is not losing any detail compared to 1024).
Both have too much information and must be scaled down on the fly to 200, so you don’t *gain* any detail by using the 1024, and Photoshop is better at collapsing *many* pixels into one than SL is.
So if your texture is roughly the size you need on screen (a little bigger, so you don’t have to *stretch* any pixels), SL can handle the rest well enough. But if it is a *lot* bigger, SL is not as clever as Photoshop when it comes to what to discard and how to average it, and may even just drop some rows and columns entirely without taking them into account at all for the final color.
Caveat: This is also subject to your graphics hardware and settings, so YMMV, but I have a feeling that SL does something crude to too-big-textures even before your graphics hardware sees it, likely to prevent abusing the valuable texture memory *too* badly.
Good thoughts in context about my thinking about
“How meshes will change the essence of second life?” beside technical things.
Well, we all await the final outcome of the mesh trials with bated breathe… altho the theory is that it will cut lag, the pudding has to be tasted before judgement is given.
re. textures, and slightly off topic, the size of the required texture is dependent on the size of the prim it is to be used on, and the number of colours (information) you need to use. One size does not fit all and there are many examples inworld where people have used a 512 or 256 where they should have used a 1024. The idea that this necessarily makes it slower to load is not completely correct as it depends on how many colours, and the detail of the texture as well as the size.
I have to agree that leaving it to the crude resizing by SL is a bad move… like you say Tali, get it right in PS/Gimp first.
I’ve worked with meshes outside SL, and made a few sculpts. The actual shape creation can be done with the same tools, but sculpts are a bit different, a bit limited in how you can work.
Texture layout is the huge jump.
The sculpt problem is that the UV-map is a fixed pattern. Regardless of the distance between vertices on the object, they’re a fixed grid on the UV-map. That means the texture has to be distorted to correct for the differences.
A Mesh can have an arbitrary UV-map. It can have gaps in the UV map. You can make parts at different scales, giving more texture pixels where you want detail. An example of what this might mean is a tailor’s cutting pattern, arranging the parts of a 3D article of clothing onto a 2D piece of cloth. And, unlike the tailor, you can use the same piece of cloth in multiple places.
The UV-maps we are likely familiar with in SL are the texture templates for the AV. The head and face are a relatively small skin area given a texturemap of their own. Both arms use the same texture.
It’s a tradeoff. The standard AV template does have weaknesses. But a well-designed UV-map makes it much easier to have a texture that looks right. Emulate the tailor, and different fabrics are easy to create. Baked shadows are done as layers.
I’ve a feeling that I might do best to switch to Blender, but Meshes are something I’ve worked with before.
@Grace
Even if a lot of basic meshes like a four vertex, two triangle, one-sided plane was passed around and used for walls and floors instead of the basic prim cube, rendering costs would be orders of magnitude less expensive.
This should definitely show in hair, where a thousand or more triangles from a prim is making up one tress when it could really just be four or so.
Another thing to consider with sculptmaps going to way side is their the fact they required a texture of their own. While I’m sure the majority of sculpt users kept their map sizes to 64×64, I wouldn’t doubt if walking around us at times were 512×512 pony tail sculpts with a 1024×1024 texture map.
And on texture maps, we control UVs now. For something as simple as a couch it won’t require a texture map for each of the cushions, another for the back rest, one for the rest or however it’d have been split up. We can thoughtfully layout UVs and work with just 1 map. Probably will be a lot more 1024×1024 maps roaming about consequently, but it seems like there’ll be far fewer HTTP requests for textures, far fewer cached, and hopefully far less waste.
But..these are the kinds of things that aren’t going to happen over night. I’d be surprised if by next years Hair Fair, the majority of stores are using mesh for example. I don’t think that’ll be the case, even if working with meshes are far easier and far less restricting than sculpties.
Don’t all those low-poly objects look a mess in SL’s lighting system though? Or am I wrong?
Tateru, you’re right. A mesh that is too low poly may not look too good in SL’s default lighting mode, a mode known as vertex lighting. This type of lighting needs vertices present in order to light them up. This is why SL’s primitives are made from more polygons that one may expect.
SL has another lighting mode that uses GPU shaders, this mode doesn’t rely on vertices but requires much more GPU power to render. This lighting mode is the one that is activated when using shadows.
In future when everybody is using the more advanced lighting mode then low poly objects will look fine.
My understanding was that the Lab had scrubbed that development project. Again, I could be wrong. I’ve not been able to get the status on it in some time. Last I heard it was indefinitely suspended.
Pushing the graphics detail up in the mesh “Project Viewer” shows shadows. No debug setting or hack to enable it; it’s part of the high setting.
(The “advanced” checkbox is “Lighting and Shadows”. There is also an “Ambient Occlusion”, which to my knowledge is all new).
I think you’re substantially right in your assessment of the effects of mesh on in-world designers.
At first hand I linked together a mesh, a prim and a flexi-prim into a linkset just for the lolz. It then functioned more or less exactly as any other linkset would work.
From that perspective it’s likely that the “impact” on in-world creators is likely to be negligible and all that will happen is that the majority who use pre-made sculpties and link them together with flexi-prims (in the case of hair) will likely instead use pre-made meshes in place of sculpties but still use flexi prims for the swirly parts of the hair. (Mesh can bend easily but can’t swirl easily)
On the other hand, boot and shoe makers will benefit greatly from the ability of mesh to bend instead of the ugly way your boots twist off your feet during an animation.
On the positive side for SL consumers, lag should drop significantly as e.g. a flexi-prim + sculpty hair of over 150 prims goes down to a mesh plus 30 flexi prims, likewise for sculptie pants etc which bend rather than swish.
Also: when finally we start to see alternate avatars start to trickle in once the workflow for rigging non-standard meshes to the default SL-rig is standardized it can only get better.
“Can you imagine what a nightmare it would be if Linden Lab tried to implement an in-world mesh-editor?”
Actually thinking out of the box all we need for a very excellent *non*-nightmare mesh-editor-surrogate is the following:
Said in-world-creator builds sculpty linkset necklace in-world. Then runs *script* or *tool* to convert the linkset to *mesh*.
Mathematically that’s trivial compared to the position we were in previously of having to sub-divide meshes into prims.
I think if someone actually goes ahead and writes that script or tool then second life could flourish as it never has before.
Or to put it another way: Imagine how *amazing* it would be to be able to build complex meshes in world using nothing more than the existing tools plus a “click to convert to mesh” script!