Mesh UV Map Questions - How to use the entire map space?

[Toysoldier Thor]

thanks Code for your ideas on process within Zbrush.  I actually do follow a process similar to what you mentioned.  Most of my initial focus on creating a new model is to create the LOD0 shape to as low a polycount as is feasible and that will be the base LOD that will be exported to SL.  I use different methods to attain this LOD base model approach. 

Some times I create the model from scratch in its lowest form - like the Spool of Thread.  Other times I use the cool DynaMesh tool to created the desired advanced model and then use a couple different techniques to re-topologize back down to the lowest poly count I want to attain as my base LOD0.  Often this involves the use of the easy to use Decimation Master plugin.  The models can be manually re-topologized to a much lower polycount as well but I am far to impatient to go thru the manual process when the decimation master does an acceptable job at decimating in most cases.

I even take this LOD0 base model into SL by itself to see how it looks on the grid and then make fine-tuning adjustments to it.

At this point I divid the model so that it reaching a polycount exceeding 1 million.  Sometimes I allow the divids to smooth and other times I dont if I want to maintain strong edges.  Sometimes I mix the smooth or non-smooth between devids to control the softening of the edges.  But, as all your fellow Zbrusher's know, the high polycount is needed since textures on Zbrush models are literally painted onto the polygons of the model (outside of any UVmap structure).  This is one of the most beloved function in Zbrush for me.  To be able to hand paint and even image project colorization onto the model knowing that the UV maps for what I paint is completely independant.

I then go back to the LOD0 and use UVMaster to create a UV Map. Set the UV Maps size to 1024.  And then set the LOD to high and tell the Tool Texture to create a Texture UV map from PolyPaint.  Poof - the painted model is perfectly super imposed onto the UV Map.  So it doesnt matter how many islands the map has, the painted texture will fit perfectly onto them.

I then use the Materials system to create the desired look I want the texture to have on the model (since the look of the materials system is influenced by the Z elevations on the surface of the highest LOD, I also Z Paint elevations on the highest LOD to further accent the materials system results.  Then I use an image plane method to bake the texture and its materials system onto the texture.  I export this resulting baked UV texture and re-apply it to the model's texture UV.

Finally I set the model back to LOD0 and use Zsculpty plugin to create the Collada output (with its UV baked texture) for import to SL.

Done.

I will admit with all the latest new features in 4R5, I am re-learning several new, faster, more efficient, and more accurate modeling and texturing.  4R4 & 4R5 have been revolutionary improvements to Zbrush than the old 4R0 I was using for a long time.  Its forcing me to relearn a lot of new techniques to my processes.  But well worth the re-learning.

[Nacy Nightfire]

Toysoldier, anyone who sells their 3D art and collects money from other people, whether its on a small scale in SL or if they work full time in the industry should consider themselves a "SERIOUS MESH CREATOR".   And they should attempt to  learn from people, without argument, who have advanced skills - plus the time and interest - to train newcomers to the field.  A job worth doing is worth doing correctly and in the end it's generally the easiest and quickest solution.

However, after knowing the correct approach to this task, If you choose a limited or non-optimal solution for your workflow, that's perfectly fine, which is why I answered your question, despite my not recommending the solution be used.  It's unnecessary, however, to keep justifying your choice to go that route here on the forum.  And it's confusing to the many beginners who rely on the forum to get good information on how to approach their work on a very fundemental level.

CG is very intimidating to folks who are just starting out and they need the right direction from people with experience.  If those people  happen to have a professional background even better.  I'm so grateful those folks take to time to help us here.  And you mistakely assumed I was a professional and I'm far from it, so you can't really speak to population of mesh forum watchers and there and their skill level or needs.  I spent alot of time in Zbrush before I got serious and wanted to learn how to make things properly.  I was pounding a round peg into a square hole and although I could make my stuff look ok, I knew I was missing a big piece of the puzzle.  All which is why I decided to get serious with Blender (FREE and used by amateurs and professionals, alike).  Suddenly I understood what good topology really meant and the importance of optimizing uvs, etc.  A whole world opened up for me and all the strange scary terminology started to make sense.  And  I had a lot more fun.  You are shortchanging yourself if you dont' keep an openmind to this.

You don't have to defend Zbrush.  The professionals do use it, and you get no argument there.  But it's not enough and if you can only learn one program, Blender (again FREE which Zbrush is not) is a better choice for all those folks you are defending who have limited resources. Using Blender the user has complete control of his/her geometry and UVS and for making streamlined low poly game assets it is superior to Zbrush.

 Edited to correct spelling

 

[Codewarrior Congrejo]

Ok I feel in need to add something again.

As I mentioned and obviously also Nancy felt that way: you don't need to defend ZBrush. It is a versatile tool. And a rockz within it's purpose. And if you choose willingly to work with less optimal solutions then stick to it, and say I know its not optimal but I'm doing it that way.

My corrections about what Zbrush is good for and what it is not, are not only for you but also for other new comers reading this and being caught in the jungle of CG and 3D not knowing what to aim for and often running into the wrong direction based on sayings of others that this is the optimum.

That being said let me explain some general things.
I am one of 'those' you keep speaking of, I am doing this for a living. And when spending some of my free time to share my knowledge, thoughts, or the will to help others I am not doing this to lecture you. Nor do I believe any of the others here wanted to lecture you. So relax about that, we are here to help each other and to learn or teach. : )

I am however doing it for one purpose - I will quote Nacy Nightfire here:

..I was pounding a round peg into a square hole and although I could make my stuff look OK, I knew I was missing a big piece of the puzzle.  All which is why I decided to get serious with Blender (FREE and used by amateurs and professionals, alike).  Suddenly I understood what good topology really meant and the importance of optimizing uvs, etc.  A whole world opened up for me and all the strange scary terminology started to make sense.  And  I had a lot more fun.  You are shortchanging yourself if you don't' keep an open mind to this...

Sentences like these are what I am doing it for! For the slightest chance of someone being open and wanting to grow and proceed and to understand things. And the lucky moment when they reach the status, and things start to unfold for them, enabling them to be creative in a way and quality they always wanted to reach. If I can even distribute just one of the many drops that brought someone into this direction I am happy.

And even with the risk of starting a longer conversation with you, let me drop you some of the points where I think you're thinking may be mislead - at least in my humble opinion:

When I first moved to Zbrush in the 4R0 era, I will admit that even Zbrush had a lot of limitations and complexities that frustrated me. 
- I am working with it for several years now. And I can tell about its limitations, and it may seem it still has a lot, but that is only happening when you don't know what it is for and what not. There is just no need to add certain functionalities into a tool that is simply not meant for those.

Could the UV maps technically more accurate and deemed more clean and efficient, or the decimation / retopologizing of the model be even lower and more cleanly structured in the eyes of an advanced professional 3D model creator?  I AM VERY SURE THEY CAN.  

- Not only for professional targets. And do consider this: you are 'selling' products. This means you have left the sector of "doing it just for the fun". If you woulnd't care you could just hand it out for free, and people would surely not worry too much about quality - as we learned as kids: never judge a free gift. But the better they become the more they will be loved.

But, do any of my customers that love my mesh art creations care that the UV maps is or is not 100%?  ABSOLUTELY NOT.  
- customers are happy folks when things are nice and even better then before. Thus advancement can never be a wrong thing. But they easily fly away when your rival has nicer stuff. Simple marketing principle...

- Will Linden Lab be thankful when mesh is finally used to do what sculpties never could: create engine conform low weight and optimized game content? I am sure they will ! And I can hear already the mechanical sound of relieve of their servers and the calming down cooling fans of the user's computer; )

For what mesh content I create for SL, ZBrush (and PhotoShop) are the only two tools I needed to invest in that serves my way of creating mesh / textured art for SL very quickly and intuitively and with more than acceptable results that is BUYABLE to the residents and customers of SL.

Okay, again - regarding the 'buyable' and as reference for other newcomers reading this:

- To make products 'buyable' you don 't even need to invest any cent other than your electricity and internet bill.
- Gimp = free (offers the same functionality as Photoshop, just has different attempts and workflows)
- Blender = open source / free and comes in so many compilations for every possible need as their are stars in the sky.

- Not even to mention Blender has its own Sculpting Tools and also was one of the first to have plugins and mods for the good old sculpties. One of the most used mods for SL is Avastar. For only around 22 Dollars as fee for the great work people put into it. Handing you all tools you could possibly need as beginner to animate, rigg and do more things and yet not only for SL alone.

Conclusion: we are now still on Zero Dollars (and with avastar on below 30) and yet we are basically already capable of making everything from  film, animations, game models and props, perfect lowpoly to absolute highpoly, sculpting, projection painting (yes blender has that too) up to texturing, image editing, even programming with an existing game engine, make models for unity and other engines and so much more.. 

And hell, i started my first professional projects on free software..i didn't have a dime when i was done with studying  ; )

Any need for several other 3D modeling tools in my model creation process would simply be a waste of my time and money as they only add advanced functions that do not need.

- Functions you don't need - I would contemplate that thought again...
Of course we can live with given limitations. But if we want to be able to create all sorts of things and even more to 'make them right' we need the proper tools. (free or for money doesn't make the difference here). And to make an optimized model can never be an 'un-needed' target to aim for.
- Waste of time - Maybe it's just me (but I don't think so) but it's never a waste of time to learn more, and to understand, and thus becoming capable of so much more.
- Waste of money - I think we have cleared at several occasions now that you don't need money in order to achieve this.

And it doesn't always have to be the expensive stuff. I've seen 14 year old blender nerds taking the job of a grown up senior artist or modeler or animator. Simply because of one fact: because they were damn good at what they were doing. And they knew to do things that fulfill the requirements. And we are not only talking here about several million dollars heavy productions. Also in lowpoly and low cost production.. and there even more, it all comes down to how good you are, and how much you understand of that matter.

If you are happy with your decision that is absolutely fine. but for complicity's sake, some of the statements have been wrong and need to be clarified for other readers. I don't want to see anyone being dis-couraged by thinking they need tons of money in order to achieve their hobby or even work related dreams.

Its all about being good, knowing your work, knowing the in's and outs, and have the strong will to proceed and become better every day. And to never stand still.. The latter is the dead of every creator or artist =)

[Toysoldier Thor]

Thanks for all your thoughts and opinions.

Thanks again for the advice of breaking model into more polygroups to use up more of the UV map,

[Toysoldier Thor]

So an important update and discovery I had today related to my own thread OP topic of how in Zbrush to take advantage of as much of the UV Texture Map as possible.  EXCITING FIND for all you Zbrush users making content and texturing it in Zbrush for SecondLife.

Today I accidently discovered a feature that makes creating extremely efficient UV Maps for models and that has been a part of Zbrush for some time but I was not sure how to use it.  This feature was also not mentioned by others in the thread as a way to get most efficiency of texture maps.  This feature as well as the process to use the feature dramatically reduces the process of creating UV Maps for Zbrush Models while at the same time creates UV Maps with greater than 90% efficiency vs the complex use of creating UV ISLANDs with polygroups!!

It is called "CREATING PUVTILES"  (PUVTILES = "Packed UV Tiling")

How do you use it?  Well its very flexible and very easy to use.  Any the time in your model development in Zbrush when you want to generate a UV Texture from your painted model (or you can even perform this step before you paint as it doesn't matter), you simply perform the following steps:

 

1. Set your model to its lowest LOD
2. Call up the TOOL--> UV MAP subtool menu (I inserted an image for you to see it)
3. Click "1024" to set the UV Map size to a 1024x1024 texture map
4. Click the "PUVTiles" button under the CREATE subsection

DONE !  Your model has a UV Map that leverages 90+ % of the Texture map.  No UV Islands to create via polygroups.  No needing to use UV Master. etc.

 

When you finish polypainting, you simply go to the TEXTURE sub menu under Tool and create texture map from PolyPaint.

I did a few tests quickly and brought into SL.  I could not see ANY seams at all.  Here is an image of the test:  It shows the model in SL as well as what the UV Map looks like for the model.

I would think that a possible drawback if this is an issue for what you might be creating models for is that the UV MAP is extremely efficient but it is not an intuitive UV Map that any human could understand if they decided to bring the map into an external program like photoshop to touch up the map further.  But if you are happy with the painting, texturing, material baking, etc. you created in ZBrush and have no need to use the map outside of what was generated in Zbrush for SL import, then this function is a dream.

 

I also discovered another interesting side-benefit related to another topic in this thread as well as another thread on how to create UV Maps for each subtool so that you can get multiple 1024x1024 maps for a model in SL.

I created a 3 SUBTOOL model and quickly high-rez textured each of the subtools and then used the PUVTILES method to create a map for each subtool.  Quick and Dirty...

I then made ZBrush create 3 DAE Exports for SL (base, male, female).  Each had its own full 1024x1024 map.

Then I told Zbrush to model the 3 Subtools and created a merged DAE for the entire model for SL. (this only tool a few seconds to create each).  You can see the difference in the image below...

I went into SL Beta Grid and I imported the 3 subtool models as well as the merged model using the same import steps.  For the merged model I rezzed it and set it to a general size. 

I then rezzed each of the three subtool models and match each of their component sizes to that of the merged model (so they were almost identical in actual rez size in SL.  I manually positioned and aligned the three to match the merged model.  I then LINKED the three to become one linked set.

The obvious results right away was that the texture quality / clarity of the linked model was much better than the merged model.  That's obvious because the linked model has 3 x 1024x1024 textures vs the merged model must provide all the texture information from 1 x 1024x1024 map.  You can see that in the maps above.

What I found shocking was that although the LINKED model uses 3 large texture maps and is 3 physical prims vs the merged model, the Land Impact for the Linked model was only 1/2 that of the Merged model !!   I have provided an image that compares the two as well as both their metrics:

 

So by linking the subtools in SL, not only do you get a model that has better texture resolution and quality, it also does this at a much lower Land Impact.  I am not sure why.  I suspect is because of how the sim interprets the size of the object.  Even though both models are effectively identical in size in front of me, you can see the reported object size of the Merged model is larger than the Linked model.

Anyway.... I hope this helps some of you ZBRUSH 3D creators.

 

Cheers!

[arton Rotaru]

About the land impact. If you merge it all into one big model, all the triangles will be counted in the Hi LOD ( and all the lower LODs as well) at a rather big size. If you load it in pieces, not only the largest mesh is most likely smaller then, but the small ones are much smaller. So they switch to lower LODs at a shorter cam distance already. This makes it over all lower in LI.

[Codewarrior Congrejo]

Yeah it is a neat feature,

just has several drawbacks. Some of them you already discovered. Like the incapability to edit  these textures later on for drawing further or different details into it.

Another downside is the quadrupled amount of Vertices due to all the UV seams. This is nothing you can i.e.. Measure within Zbrush or blender, Maya etc. It's the impact it has on the calculations that have to be done in the render-pipeline. (there is so far only one script for 3DSMax I came across that can calculate an estimated amount of the vertice count that has to be calculated in-game / insides an engine)

As you might know every vertice on a seam is double existent ( so called 'split'). once for the face and once for the UV seam, and again doubled where material, smoothing groups, etc come in as well.

There are many reasons why you barely ever find any game assets or maps being done this way. Some of them include the above mentioned.

If you are interested why so many UV islands / UV seams and the resulting vertex-splitting in the render-pipeline is something to consider as disadvantageous:

http://www.polycount.com/forum/showthread.php?t=106108

And for the straights facts: 
http://www.ericchadwick.com/examples/provost/byf2.html

Facit: it's a neat feature, and won't do too much harm if it's just a single model having this, and the model has already a very low polycount and optimized geometry / topology. But if this would be used for a full asset, or more objects in a scene, this can have a pretty deep impact in terms of rendering cost.

Cheers! Code.

[Toysoldier Thor]

That surely could be one of the factors.  I also read back a few posts and it could also be a factor that Drongle pointed out...

 

"  ETA: Maybe also worth reminding people in this context that increasing the number of UV islands increases the download weight.  "

The single larger model had at minimum - 3 UV Islands from each of the three subtools that merged.

[arton Rotaru]

Yep, indeed, like Code already explained, too. UV Splits have an impact also. But I doubt it would that huge of a difference at this particular model. But sure, it all adds up in the end.