Meshes and UV maps.. (a rant)

Okay, I'm sure this has been discussed to absolute exhaustion, but I just want to take a moment from aligning each individual stripe on a shirt pixel-by-pixel in a futile attempt to get my seams to match--just one moment to vent about the idiocy that are SL body meshes.

Who in their RIGHT MIND would UV map a torso and map the back side of it at a different scale than the front? WHY?

For that matter, who would place the mapping seam down the front of an arm, rather than the inside, like every other rational low poly mesh designer?

My three year old could map a mesh better than this, it just boggles the mind. Did LL find some random amateur to put this together?

Let me just put out this offer...

LL, I will remap your meshes, PRO BONO, to the point that they are just... flawless. They will be a BREEZE to texture. And I'll even throw in a revamp of the Butt-Injection-System-Skirt mesh. The fact that the skirt hasn't been taken care of... I mean, yes, in the grander scheme of things it's pretty irrelevant, but it's probably a 20 MINUTE JOB with even the most rudimentary meshing program. And it would make so many people so, so, so much happier.


Okay. *pants* I think I'm done.
*steps off box*

Thing is, such a change would mean all clothing and skins currently in SL would become useless at a stroke. Obviously this doesn't mean nothing should ever change (sometimes progress means uncomfortable adjustments, that's just reality,) but do bear in mind this small 3D model adjustment would set off a huge chain reaction of problems. The way things are has become very entrenched.

Also, I think you overestimate how much difference a better UV map would make. Yes, the seam placement on the arm is annoying, but painting textures for characters is never a breeze, it's always a tradeoff somewhere.

Feel the pain. Feel the pain!

If your software has mapping channels, use them. I'm constantly re-mapping to a different channel. If I ever get the yarbles for it, I might even re-map to something a bit more Poser friendly.

One thing I've been considering lately is doing my own rig. This would *so* help with projections.

Yes, I definitely understand the huge incongruity that would occur with even minor changes. However, as a designer myself, I would be willing to completely refabricate my entire collection of clothing and skins if it meant that the changes made would make it less excruciating for newer, less experienced artists to express their work--and to contribute to an overall better Second Life product. Yes, I know I don't speak for every creator out there, many of whom would have a much larger work load ahead of them than I would, but the argument that there is simply too much invested in how things are right now is a very narrow-minded, short term view.




I disagree. For an inexperienced texturer, yes it is always a struggle--but a great deal of that comes from a simple lack of familiarity with the way a texture lies against a particular mesh, or because the uv map in question is mathematically inconsistent. When a mesh is mapped properly, texturing is very much like sewing. If you measure the material properly, you forgo wasting endless amounts of time fiddling with it and making minor adjustments. Measure twice, cut once.

i'm trying to make a jacket right now that has a continuious pattern of braides that move from the front to the back and expand along the arms i'm pretty sure using sl's avatars its impossible i have been at it for days and nothing i do will make everything match up.

I came to the conclusion that what will have to do is make part of the jacket out of flexi prims or something (probably the collar and arms which does not thrill me to death) but its the only thing i can think of right now that will possible (not sure yet haven't tried) make it work

*wonders why he thinks up complicated things to make and just doesnt forget about patterns that span alond the chest and arms and back and arms his life would be so much easier*

*goes back to the non cooperative avatar for another 5 minutes in an effort to make it work*

*browses the forum again in an effort to avoid it as long as possible*

lol

Good luck with that one Sounds like a doozy.

The whole issue of baking with 3ds Max etc etc sounds fairly involved, but promising. All of my experience is with very low-poly meshing programs (such as Milkshape).. almost none at all with 3ds Max, so I think I have alot to read up on before I'm able to use that feature to my advantage.

Here, here...you go, girl!

Hehehe, good rant.

I'm not normally one to insult Lindens, but yeah, whoever mapped those UV's must have run out of crack that week or something. In other words, someone actually on crack would have done a better job.

In regard to the "remap = break content" concern, what we need are selectable mapping options for the avatar, and then there's no problem. When you make a garment, you simply press a button to select "default mapping" (the new, sensible one) or "legacy mapping" (the old, crappy one). That shouldn't be hard to implement at all.

I love this idea. As a texture artist, I find the UV/seam issues extremely impeding.
Someone should open a proposition?

You're... you're a genius! This solution is beautiful in its simplicity.
Anyone knows whether LL has plans for a new mesh, by the way?

Clever solution! How do we get LL to read this?

If I recall correctly Phillip Linden actually described that situation of how they'd move to a new avatar mesh in either a town hall meeting or in an interview on www.secondcast.com about a year ago. He said there'd be a way to select which mesh you'd use just like Chosen explained it.

In the same discussion, Phil said that they are working on a new avatar setup, but he also stated that the avatar in SL is obviously really successful with how great people have made them look. So, the new avatar is a back burner thing at LL.

In their minds, it ain't broke. And some of you do look really good as it is! So, why fix it?

We deal with the tools that Phil and company give us because at the moment we have no choice. I don't expect that to last too much longer with the direction that open sourcing is going. I'm skeptical that LL will be the ones to move forward with this. I think it will probably be a 3rd party when they get the chance.

If we were all content with spitting berry juice and charcoal out on cave walls then this thread wouldn't exist. I think we're all in agreement that given the same number of polygons to work with, many people could improve the current avatar. I spend a great deal of my time dealing with the resolution and distortion inconsistencies spanning the (far too many) avatar seams. I could cut my work time down to a fraction of what it is now with better seams. We just need LL or someone else (when that becomes possible) to alter the platform a bit to give some of us a chance to work on this. Nothing needs to break as was mentioned above.

hehe, I share your pain. I think the avatar was mapped the way it was so that the template shapes would be easy for novices to recognize. If it were mapped for efficiency, relative polygon scale, and optimal use of UV space, it would be a lot harder for people with no previous 3d texturing experience to learn. Now that there are freeware av texture previewers that's really not as big an issue as it once was. If they ever take the redesign off the backburner again I do hope they'll ask for some feedback from all of us first. I'd hate for Avatar 2.0 to end up with only one arm or some other apoplexy inducing design decision.

Abu, would you please expand on this re-mapping to a different channel? Is this a technique for dealing with the front and back scale difference? I use Photoshop CS.

I'm hoping Chip Midnight comes out with the templates like the ones he currently has if the re-meshing thing goes through.

hehe, if they do redo the mapping I'll make updated templates for the new UVs, but there would probably be a few months of lag time before I find time to do it. We're just engaging in wishful thinking at the moment. I wouldn't get your hopes up too much. We've been begging for improvements to the avatar mapping and texture layering for a looooooong time now.



Unfortuantely it's not something you can do in Photoshop. What he's talking about is the ability of most 3d modeling programs to store various different UV mappings in different channels on the model, allowing textures to reference whichever set it's intended for. Say for example you were trying to make a sleave with a fairly complex pattern. You could wrap the texture around the arm cyllindrically using a different UV mapping, then bake the result back to SL's UV mapping. It's very handy for working on some of the trickier parts of the av. You'd need a program to supports UV editing, multiple UV's per object, and texture baking, such as 3ds Max, Maya, Lightwave, Cinema 4d, or similar.

I have access to 3ds Max, but haven't yet found the time to really break through the learning curve. My ability is quite rudimentary, and I'd need alot of hand-holding to make my way through the process, I think.


My opinion is nowhere near unbiased, as I am a texturer by trade--however, I believe that what has made SL stand apart from the crowd is the boundless opportunity to create anything you can imagine, as well as the novelty of a virtual user-created economy. The economy, of course, is dependent on the creation. SL makes the creation process much easier in alot of ways than other games, as it is the primary function of the game--however, the system is still quite flawed. Looking at it from a purely business standpoint (and it is a business, I have no problem with that), refining the creation process would be incredibly profitable to the virtual economy as well as the in-game environment. The end result would be a much more salable product.

I did (and still do) alot of work for The Sims, starting back before The Sims 2 was released. The release of TS2 and its dramatically improved graphics and creation resources forced all of the custom content creators to improve our skills and really up the ante. Content creation in many ways stopped being the sole realm of the very graphically and computer savvy, and the community received a flood of new creators who had a great deal of vision and artistic skill, but weren't quite able to get through the lengthy learning process required by the original Sims game.

While this may not seem ideal to alot of creators who have the game in the bag right now, again there is the matter of taking the long-term view. Choices should be made for their benefit to SL in five years rather than simply their immediate success. Countless businesses have failed as a result of taking the buck now over the hundred later. And as I'm rather fond of my SL, I'd really like to see it thrive, and see resources invested in its future.

Yes, I'm a total windbag. So sue me

I'm pretty new here, but I've already lost count of how many times I've wished I could just get the av mesh into milkshape3d and redo the uv mapping - even I could improve what we have now. And I was thinking that once we get alternate meshes we could also have ones for clothes that don't look like they're just painted onto a naked body...

I've got some tutorials about using mapping channels in 3DS Max. Couple other tutorials tossed in there.

Max Slop

Geared for beginners.

Unfortunately, I kind of dropped it during a round of editing. So, there may be a few things in there that don't quite make sense.

Mapping channels can be hard to get the hang of. But once you have that epiphany or Ah-Ha! moment, you should be good to go. It gets really fun when you start mixing in other kind of channels and material IDs.

You're preaching to the chior, Arielle I would love to see the SL avatar get remapped, even if it's not a huge change, just fixing some of the really crappy areas (shoulders, groin, the top of the head and adams apple seams being within bleed distance of each other, only one arm), I'd be a very happy camper - as long as they build in backward compatability.

Ideally they should do two new UV maps - one for beginners that's created similar to the existing templates so that the shapes are easily recognizable, and one for experts where it's mapped for efficiency and correct polygon scale. Even if you go through the trouble of remapping the av model yourself, as soon as you bake back out to the SL mapping you're right back to all its limitations.

I DO think that someone should open a proposition on this, despite the possibility of 3rd-parties being introduced into the mix. Frankly, the reason these things stay on "back burner" is because we put up with it. Software development and bug-fixing is a classic case of "the squeaky wheel gets the grease."

I am not a Linden basher in general, but I will say that I think it's ridiculous that they seem to be counting on third-parties to fix their bugs and design flaws. As the creators of the software I think the Lindens have a certain responsibility to properly maintain their architecture and code base, and frankly, there is no guarantee that a third party won't just make things worse.

So...will someone with the knowledge to do so please consider creating this as a feature request/bug request in the public forum for voting? Because I for one am willing to put most of my votes into it.

Any reason you don't want to be that "someone"?

Since I obviously have no problem shooting off my mouth, I'd be happy to be the person to open this proposition. However, there have been numerous cases where a single individual has presented the problem and been entirely ignored. A good deal of that is probably result of the individuals concerned only complaining about the problem without offering a viable solution, and because they went in without backup.

While I have no doubt that this is an old issue, whining isn't enough of an inconvenience for me to be deterred by low chances of success

Just to be clear on exactly what we'd like to see:

Two new UV map options, one simple and easily navigable for beginners, and one more advanced with a correct polygon scale and mapping for each body part (both arms mapped seperately). In order to incorporate backwards compatability, clothing creators should be able to choose from either of the two new maps or the old one when creating a new article of clothing.

Any other thoughts?

I definitely agree with the squeaky wheel bit. There are enough clothing texturers in SL that if we all squeak in unison, the effect is going to be more than a little bit annoying.

I dunno, I used to make textures for Irrational's Freedom Force game, including three base skins (one make, two female) and I haven't had that much problem with the SL models. The only issue I have is that the UVMap is for the female model, and takes a bit of imagination to get outfits to work for both sexes. Other than that, my texturing transition has been, pardon the pun, seamless. I still have yet to finish my base skins for SL, but I already have a clothing base. It just takes a bit of patience. Though I would probably use other models two. Personally my mapping issues are less with seams and more with the mirrored arms and feet, which annoy me to no end. Still I would rather have my demibuttoxed UVMapped avatar in a working SL than a pretty model which won't rez so I can see it.