Macrocosm Draegonne

In your misplaced and mostly out of context berating you failed to imagine even a most basic use case? As one example, say an object with four materials, that are tilable textures, when this thing loads, you will download all four of those 1024 materials. Making an atlas out of those will not reduce the file size, but it will reduce the texture calls by three. Four textures become one, that is being more efficient with textures, BECAUSE YOU WILL DOWNLOAD ALL FOUR (or more) ANYWAY. That is a micro example, but larger builds and full SIM's could have a few specific sets which would be the sort used in most if not all areas of the SIM, or one for a certain area if that made more sense, its flexible, and only useful where it is, nowhere else. The really useful thing about an atlas (other than performance) is ability to tile the quadrants. Sure, when we get larger textures we can just add more uv/textures to less files, because there is more pixel space to work with, but the tilable quads wont work, we'd be stuck with a one direction of tilablity (in a combined atlas'ish layout), and only if the given uv goes all the way across.. The same can be said about BOM, baking all those skin related textures to the skin is great for performance, and there WILL be materials support, thats is not even in question. EDIT: Furthermore, to bake a grid of textures is hardly as difficult as all those avatar layers, surely it would be lighter work? Using the bake server would make it all more user friendly, but its not essential. Being able to use atlases and tile quadrants within a larger texture would be very useful though, far more exciting then having a bake interface, which is a moonshot in regards to atlases. lol

No dear, that would be your idea, one imagined so that you could argue against it because you're disappointed in the current beta-pre-release status of BOM. There are cases when a texture atlas is useful and does provide performance benefits, those are the interesting ones. Arguing against that seems silly, but you're not really arguing against that, you're going on about what its not good for, and I am not interested in the ways they would be less useful, or misused, I am interested in ways they can be of benefit. If they were totally useless they would never be used at all for anything. I dont expect BOM to be perfect day one, and I also do not expect they will ignore the big things it needs, they've addressed those needs enough to please me at least, I expect they will iterate and improve the system as they feel it out over time. There is no question at all it will have materials support, they've said that is a priority once the base is setup. I get the sense you're just going to hate it either way? Onion skins, or no onion skins? Let it suffice that people will choose with their dollars what they like best, I certainly wouldn't want a body/head that got rid of all the onion skin layers, but I would appreciate not having more than one additional, because the complexity will be lower, and texture calls too. Theres no reason to have all those images plastered all over an avater when its stuff that should be in the skin, baking all that to one is brilliant. The system clothes mean nothing to me, I tossed most of those out years ago, but I love OMEGA and customizing, I really dont want that to ever go away, only to improve and evolve. Maybe you dont think removing 1/3 or more of the onion skin layers is significant, but if you've got 100 avatars around you, those tri's can add up. and think of the textures (beards, brows, hairbase, tattoos, scars, freckles, makeup, etc) now reduced to 1 for the body, and 1 for the head. Plus their material layers too of course. How could anyone think thats a bad thing?? I can understand keeping LL feet to the fire on knocking the feature out of the park, but I dont see any reason to think of BOM in a negative light simply because its not perfected yet, how could it possibly be, its not even released yet lol.

Theres nothing negative in the translation of that tho, I am just chatting, I like memes, sorry not sorry. Translation "Good morning, cuties" lol I thought it was cute it said BOM, so sue me! And I do wish you a good morning, depending on your timezone? I love the peanut gallery here, the whole lot! Its precious. I never take negative comments personally, and I really do not intend to be hateful towards anyone, though I do get carried away rambling as im sure you've noticed by now. AND>>> unlike some apparently, I am very excited about BOM, I do not expect it to be perfect day one, but it is an excellent feature IMO.

OK lol, I was just teasing and checking, its hard to know when you're being sarcastic. So you know a texture atlas has its specific place and use, why then all the hysterics about things its not useful for? I am also not interested in those applications... I am interested in ways they can be used effectively. They are already in use in the current SL system at 1024, and of course there is no way to tile inner quadrants. BUT I am not speaking greek here, a texture atlas can do what it does great, and there are ways it can fit into SL to reduce texture calls, when it makes sense to do so, not in all the imaginative ways you can propose they not be used. ^^ And BOM is gonna rock your socks off! Get ready baby! ?

They do plan on adding material support including tinting, it was said many times in the BOM meetings I was at. I would never propose to remove all the onion layers, quite the opposite, but one is better than four+ Especially when most of those layers are pointless layers, AKA, only layers because there was no way to bake everything on the skin together. BOM should be a little bit of what everyone was looking for, i wont consider it fully released until it has material support though. It would need to bake a normal and spec too. Then the One or Two(though thats excessive) remaining onion layers can be used for those that want to wear some cool classic stuff they might have. But far more interestingly, those layers can be used to add effects and such on the skin for real time game animations and interactive. Stuff like the lalukta cotton candy, your face gets dirty and you can lick it off, both mesh chunks and a texture layer on the onion skin.

A texture atlas is generally good for performance on many fronts. That will not fix bad textures, you're throwing a red herring into the convo. ***** bad textures, they will be bad no matter what. What I am speaking about is the correct use of texture atlases, which if you've been paying much attention you'd see are all over the place already, albeit minimally, many creators combine various parts into one texture, and some are quite efficient with using available texture space too.

Im talking about texture atlases in case you're lost? Do try and keep up dear. https://en.wikipedia.org/wiki/Texture_atlas

There will always need to be at least one, its too useful not to have at least one onion skin. but a lot of the others can go, if its things that are supposed to be in the skin anyway why waste an onion skin and all those texture calls anyway? Scars, Tattoos, freckles, makeup, or any other skin stuff seems a perfect fit for BOM Avatar Bakes. The system clothes are not really relevant no matter what way ya shake it IMO except maybe the few types of things that look good skin tight. So, no I dont think Omega is going away, but it will have to adjust to the change. Nor do I think all the onion skin layers are going away. They're a make-due until we get more shaders, they're not only used/useful on avatars either.

Seems to me BOM will reduce some onion skin layers, which is great for the avatar skin use case. I suppose my post was slightly off topic perhaps, but it is related to the bake servers, which are used for several things already. Atlases will be the same textures, so its not any additional images, and they could actually save texture space wastage quite a lot, they can even do that automatically if programed to do so. Atlases are not some cure all, or needed for every thing, but they're excellent for certain things, theres no good reason not to use them. The larger texture size will happen, theres no reason OZ would say such a thing if it were not in the realm of possibility, he did not say when though, of course.

Yes, I realize the current use of the bake systemfor BOM is for Avatars. Im sure whatever oddities there are will be worked out to make the best of it for that use case. I am not however speaking to that use case, I am speaking about the bake system though, and its ability to composite images as it could relate to making atlases. Also, FYI we will have larger than 1024 images soon enough, I saw Oz say it himself. Its only a matter of time, who knows how long, but thats not the point.

Is it not better performance to load one 2k texture rather than four separate 1k textures? The file size would be the same basically, but it would be three less calls for a texture. Yes I realize the current implementation is geared towards avatars, but that does not always have to be the only use case. The terrain is another texture bake the system does. Combining several textures to make an atlas does have some benefits, although its only useful in specific cases, and would need to be well structured to prevent abuse. However, its a standard performance enhancing technique thats easy to do, even if we just upload our own self created texture atlases that would be a huge blessing. That said, using the bake system more enhanced performance minded (and automated) techniques can be implemented to ease the viewer, network, and data drives too. All with the same textures already in use today, only combining them when it makes sense to do so. EDIT: My point is really about the bakes way of things though, since texture atlases are nearly a moot point, we will soon'ish have larger than 1024 textures allowed, so we can make atlases on our own, and would then only need a shader bit to be able to tile the quadrants that are referenced in the atlas. With bakes though, there may be some cool cache automated atlas thing that could run on a schedule, choosing the best combos for an area and building the cache with it, updating as needed, and dumping outdated atlases, all automatically. Along with that there could be an atlas building UI, perhaps in the mesh/texure uplaoder, maybe the SIM management level, or who knows where else. The power of Atlases. A great way to see the value of atlases is when heavily modding skyrim, there is a 3rd party app called DynDoLOD you simply must use when you put many hundreds of disparate mods into the game (like SL) You end up with weird LOD that performs bad, and looks bad. Run DynDoLOD and it automagickally builds a perfect array of texture atlases (among other things) in logical sets, this considerably improves game performance, and massively improves the visual detail, all while making the LOD amazing too.

Seems well enough to set the slider at something reasonable, a few hundred thousand, if someone is too much I can just derender them, I do see jellydolls sometimes, but I think many people have learned about over complexity. Even the creators, bodies seem to be optimizing a lot, and BOM will remove at least some of the onion layers too, so things will get lighter, just as LL is improving the infrastructure , perhaps this time next year things will be quite different? I dont really lag from avatars, although im not in busy places often.

The stars seem to be just a solid image, that gets dispersed, white so it can be tinted. I would not doubt for one second they can tweak the stars too! I could have sworn they did already for EEP, although ive not tested this feature viewer yet. Allowing us to further customize stars, and to make them twinkle too would be really sweet!

Oh dear, I have some that are 100+m variations I cant recall if they're very much higher than that except maybe a giant sphere or two that are insane big lol, id have to dust off some olde inventory to see, I hope to use some of them for giant backdrop things eventually, off in the distance stuff.

They're so cool, I think I deleted all of the ones that were known to be a bit buggy, but I still have a bunch, nothing else can compare size wise, you'd have to upload tons of mesh to equal one of the big ones. The can be great for filler and buildings.

On what other things besides avatars can we bake textures? Could BOM potentially be utilized to create Texture Atlases that are still tileable textures? I know these would only be useful in some situations, but they could add better performance to a build if used correctly. We would need a new shader for that to work I would guess, in order to still have tileability on the individual quadrants. Perhaps the same Viewer interface can be used to implement this, plus a flag or something on the object that would denote its able to utilize a texture atlas. Or perhaps it can have one uploaded (or created) with the object in mesh uploader? Most 3d apps can already output a perfect texture atlas, I know blender does. But BOM + a shader could allow for creation of them out of disparate textures. For landscapes and things reused a lot in a SIM, wouldn't these really add a considerable boost to performance?

Time > Money Charge enough to ensure your time is valued, and enough to detour casual lookie-loos, unless you want to spend dozens of hours earning 50 cents lol. Sometimes thats is quite fun and the right thing to do in a given situation, but generally speaking, let price be a fairly crafted barrier to protect your time and sanity. There is something to be said about psychology of price, too low and many will perceive the value as lower somehow. Too high and many people cant/wont afford it, which may be fine, if at least its enough customers to keep you active lol. Being a RL biz developer I find SL's economy fascinating. I have been looking and making a study of it for quite some time, its brilliant, and correlates to RL business in a lot of ways. You can clearly see "best business practices" always pay off better, customer service, quality goods, etcetera. The more professional you handle things the more people will expect (and want) to pay a bit more for the added luxuries and pro attention. Ultimately it depends on your product/service, and the micro-market for that product/service, make a study of the areas you're interested in, you'll see a set price range, and established expectations, etc. There you can see where you fit in, how you can find parity with existing norms, and add your new creative inputs too.

Yes indeed, I have a great collection of mega prims, those are priceless! I loves em!

hehhee, actual prims? Doesnt that waste li and add a ton of unnecessary tri's too?

Maybe his issue was more the height of the wall, if its too short you just go over, but taller things dont have as much an issue in the normals fallback calculations.

Yes, I never use the auto-generated lod or physics, unless im just testing something and its not important. I've found its always much better to make them from scratch in all my testing. For things I dont want visible far away I can take that to one tri per material, so small it cannot be seen. On some models thats just one tri, so at minimum its reducing viewer strain to the degree I desire. Seems like it would be smart to keep clean, especially since they're doing a lot of modifications to the system. People have to be approved, and want to post though, which is really the issue. You're right, a good setup would be a great undertaking to pull off, definitely at least one full time job for some given amount of time. So true, I am getting faster and faster every time I sit down with blender, and I am have mega creative flow states, so thats always fun, now if I could just complete some more things lol! Its so much fun I cant stop! ?

Its not that difficult to keep the system caveats in mind when making things here, ive gotten into a rhythm with it somewhat already, every game will be that way, no matter what technology/platform etc. Once you know all these tweaks and tricks to building they'll find their way into every build of yours one way or another. One of the big things for li that I didn't figure out until recently is zeroing out lod for things that have no business being seen from afar, interior walls/spaces, fine detail meshes that should only appear somewhat near to the camera, etcetera. I've been obsessing over this for a while and still have tons to learn! Plus there's more coming with the new windlight/animesh/bakesOnMesh stuff, and yet more, the less discussed things like bigger textures than 1024, I watched Oz say it himself its coming! along with a much peppier cache/server and rendering, then there's physics which ive heard rumblings about some future enhancement. The possibilities increase exponentially as time goes on, with the huge investments they've been making its wild to imagine what amazing things will be built in the coming years. I heard the wiki is open to anyone who wants to edit, but I think you've got to get an account pass for it. Building is complicated though, that's never going to change, see any other games that allow modding, its really technical stuff and highly engine dependent. I think thats fine though, it makes impressive builds all the more striking when you know how much time/skill and creativity must have gone into it.

would it reduce the physics li any if he merged the top vertices in the two sides? So a pyramidal shape giving two sides, single verts at top and double at the bottom. It wont reduce tri's but maybe less verts helps too? I cant recall if i've tested that yet, but I will soon im sure.