Using plural materals increases PE

[TANAKAAKIO Inshan]

Hello everyone.

I usually paint meshes with plural materials so that bake textures in Blender.
During working hard to model and bake texutures, a number of materials is over the limit 8 before I knows, after especially joining objects.
As you know, materials in blender become faces on prims in world.
Over limit of materials make prim's surface funny.
Furthermore using plural materals increases PE.
And affects physics weight also somehow.

If not a specific case, one material(face) is sufficient to paste a surface texture.
I think specific case are
  to use different texture settings(trans, glow, fullbright, shininess bumpness and so on) on each face.
  to paste different textures on each face.
Accordingly, I copy an object then reduce matarials from cloned copy as far as possible to upload meshes driving down PE　in Blender.

 

Uploaded by defalut  separately.



[Drongle McMahon]

A nice illustration of this effect.

Each material has a separate sub-mesh in the internal data format. The sub-mesh includes both triangle and vertex lists. This means that the vertices where two materials meet have to be replicated in the separate lists. The more fragmented the material patches are, the more replication. This increases the size of the download data from which the streaming weight is calculated. Fragmented UV maps and sharp edges have a similar effect, although here the replication is within a submesh's vertex list.

The case of the ninth texture should really be caught as an error and prevent the expensive upload! I seem to remember a jira for this, although it might be worth a new one. People will get angry if they find they are wasting upload fees because the error is not caught.

[WADE1 Jya]

Good catch really TANAKAAKIO!  That might not be an error.... it could be an actual intended limit but, if so, it should definitely be clarified as such!

They might not want people using multiple materials (except where design makes it unavoidable) & penalize that just so materials are always used as sparingly as possible. A limit might be fine to have really.

In many of these types of cases, further optimization is possible, and this should be encouraged - often you can do similar effects all with just one texture in the end - standard practice in the video game industry.

Textures are more expensive than the vertices they are stretched around.

[Drongle McMahon]

The 8 material limit has been there since the beginning. However, when I looked in the wiki to find a link, I could not find a mention of it, not even in the section on uploading multi-faced meshes or in the internal format definition. That certainly is an omission. I guess a mesh limits page might be useful there to include this and other relevant information?

[Gaia Clary]

I would expect that the importer did tell something like: 

"Too many texture faces: 9 used, 8 allowed"
"Too many physics vertices: 450 used, 250 allowed"
"Too few texture faces in medium LOD:  5 used, but 7 needed (see highest LOD)"
"Unweighted vertices: 7 in mHipRight"
"Bones missing: mHipLeft, mAnkleRight"
"Bones unknown: m_hip_left, m_ankle_right"

And so on...

That would be the simplest and most straight forward "documentation". And IMHO even the most efficient way to document. From my experiences with Primstar i have the impression that most users seem to either not find the docs (my fault :matte-motes-agape: ) or don't read them (their fault :matte-motes-grin: )

[Failed Inventor]

I thought it would be terrific using multiple materials, like prim faces, but find that you can't really use them like sl prim faces if the ojbct is complex, so really there is no plus only penalty to multiple materials. Unless you need the multiple material for scripting purposes, say targeted effects etc, shine, glow where others would exist.

But if no scripts or no special LL effects then you would want to just make an extremely well made template for your object with psd layers with a single material, cause any attempt to use the multi material feature will result in additional penalties.

[Drongle McMahon]

I don't think I can agree with you there. I find the extra versatlity of materials to be one of the most valuable aspects of mesh, for many reasons. Perhaps the best is the versatlity of co-ordinated retexturing with discontinuous material patches. With appropriate UV mapping, this allows the end user much scope for simply badapting a mesh using shared tileable textures to decrease texture lag. For example, this might be making all the windowframes in a house match, or recolouring all the candles on a mesh christmas tree. On the other hand, the materials allow eight times the baked texture detail to be applied to large meshes (perhaps not to be recommended as this will increase texture lag). The PE burden should be little more than that resulting from UV seams or sharp edges, especially if UV seams ams and material edges coincide, as is often the case.

 

[Da5id Weatherwax]

I have been experientiing with the same thing and I find the only place to use multi materials is for face-specific features like glow or shiny. for the rest I take my high-LOD model, subdivide it like crazy to create details (usually labelling this one "insane LOD" and split it into as many materials as I want to bake the texture.. then go back to the models I intend to upload, set all faces smooth and upload as a single submesh.Get the lighting right in blender before baking and the hard edges show up just fine even though they are a smoothly rendered surface inworld.

[Failed Inventor]

Seems a well made uv unwrap with psd layers could achieve the same. I am still not seeing any upsides other than application of shine,glow, transparency to specific target areas.

As you even said the one viable use it does have is not recommended as it will cause increase texture lag. 

I generally agree with you on most things, but I think people need to update from 2003 a little as optimized texturing and light baking is best done in the 2D and 3D exterior programs of your choice. While many may not like this fact, Second Life simply doesn't have the capabilities of professional editing platforms.

Rather than trying to stick to old habits and limitations people would greatly benefit from 1. Learning a real life skill sets, by working in the same or similiar workflows as professional artist.

2. Ability to multi market to variety of platforms, as Second Life won't always be the top nacho, and when that day comes you want to be able to use your skill set to easily adapt to the new platform.

But for some I understand this is just a hobby, but, if you are going put less effort into your profession or hobby, expect less results and ability. 

 

If anyone believes that someone with less effort should have same capacity and results as a person who invest their entire days, weeks, and lives to their mastery then what would be the point of skills, knowledge and experience anyways.

[Drongle McMahon]

I don't think this has anything to do with effort or profesionalism, just simple facts. If you want to use baked textures, as you seem to be recommending, then you can have eight times the detail resolution by using multiple faces. This improves the quality of large objects, whatever other techniques you might apply at the same time. But baked textures are bad for lag and worse for adaptability. If you decide not to use baked textures (hard to resist though, for the reasons you give), to save bandwidth and/or to allow re-texturing with simple reusable textures, that can be done either with multiple materials, or by breaking an object into multiple tiny objects. The latter is essentially ruled out because of the minimum prim equivalence and the the minimum upload fee. 

Thus from my point of view, multiple materials provide me with a useful feature whichever texture strategy I choose. I have also used them to overcome the absence of repeats on animated textures (also doable with UV stacking, at the same PE cost), for providing super low-LOD alpha representations without requiring the high LOD texture to be alpha, and for making invisible flaps to shift the object rotation axis, again without needing to use alpha for the main texture.

If your purpose in using the SL is the development of transferable and marketable skills, then of course you must use it in the way you see most appropriate to that end, and you are welcome to do so. That might quite rightly override other factors. Despite being seriously ignorant of such matters, I can't help suspecting that SL, with all its idiosyncrasies and limitations, may not be the ideal platform for that endeavour. The adaptability consideration, for example, is possibly unique to SL, but important the context of enabling user creativity.

[Vivienne Daguerre]

One must remember that there are different uses for 3D skills. You may be making illustrations or static 2D art. You could be making characters and backgrounds for animation for movies or machinima. In these two cases, polygon count does not matter as much but they do increase render time. You may be making assets for gaming, and in that case polygon count matters a lot. Good texturing skills matter for all three of these uses.

Making things for Second Life is like making assets for gaming. I have seen all kinds of things from WOW on the beta grid, and they are very low poly with highly detailed textures on them.

The skills you use for Second Life are transferrable to other areas in my opinion. I have considered putting some of my mesh things up for sale on Turbosquid or Renderosity. I can provide them in .obj and .dae format, but not the formats used by Maya or Automax, but they can import the .obj and the .dae. I plan to look into that.

[Zalandria Zaius]

I think the biggest problem you'll run into with baked textures is if you sell items. Inevitably someone is gonna retexture it not realizing it's one big texture. It's also easier to customize if you use the materials to make the faces for something someone might just wanna change a color of here or there but leave the texture alone.

They all have their uses and working together can be great. You can now set 1 material for all of your windows and easily change them at once without having ro run around clicking each one.

This will be great for clothing. You'll be able to set seperate materials for well the different materials so color changing will be done by faces but the texture can be baked greyscale.

The possiblities are endless but there is that 8 material limit that they don't seem to be catching and erroring on.

I did notice the vertices duplicating wherever materials touch to, and all the LODs have to have the same number of materials, so make sure you have a clean, set smooth, mesh for your physics and don't just use the lower LOD if you are using materials.

[Failed Inventor]

I understand and I think you have some very good proints drongle, thank you for discussing it further.

[DanielRavenNest Noe]

---

Zalandria Zaius wrote:

The possiblities are endless but there is that 8 material limit that they don't seem to be catching and erroring on.

---

Just a note for anyone testing items on Aditi - only 6 of 8 materials are functioning at the moment, the other two remain grey.  That was reported as an issue, and Kelly Linden noted that it was fixed in the JIRA, but the fix has not made it to the servers / viewer code that I know of yet.  I have some 8 material objects I check with periodically.

EDIT: the problem seems to have been fixed now, visiting Mesh Sandbox 28 and going back to Sandbox 30 may have done something, I had usually worked in Sandbox 30.