Jump to content

3DCoat Experts with workflow Advice?


Toysoldier Thor
 Share

You are about to reply to a thread that has been inactive for 4529 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

Throughout January I have been trying to figure out the best workflow to get a 3D Model frmo being sculpted to being textured to getting into SL with what was sculpted and textured.  I have been experimenting wirh many tools and different workflows.

I love Sculptris and thanks to some advice from this thread I have even been able to learn how to use Sculptris's polycount reduction tool and reduction brush tool to make some really cool artistic models to bring into SL.  But I have learned that the ZBrush UV Master has become a critical part of getting the Scultpris model ready for exporting to Blender which in turn can create the DAE file.

So  I decided to take a serious stab at 3DCoat.  It uses the same Voxel modeling that Sculptris uses and has a lot of additional modeling tools that Sculptris doesnt have.  I also have been hoping that 3DCoat could do in 1 tool what I now rely on two tool to get ready to send to Blender for DAE creation. 

Well after several emails back and forth with the staff at 3DCoat (and they have tried to be helpful), I am pulling my heair out is getting through 3DCoat's workflow steps from VOXEL to a texture painted model that can cleanly be exported to Blender.  Its a bit of a confusing tool.  Sadly the 3DCoat staff dont know much about SecondLife and their intructions keep assuming I fully understand the UI and have a deep knowledge of 3D modeling, retopologizing, UV, and their 3 types of painting.

QUESTION....

Are there any 3DCOAT SL Mesh creatiors that have some basic easy to follow workflow steps on moving a voxel to SL DAE with the textures?

Link to comment
Share on other sites

I have owned 3DCoat for about 12 months now, but thus far (to my shame) I haven't had a chance to properly explore it. However, it is a VERY powerful application, which I can see myself using a lot more of down the track once the SL materials project eventually gets released (to create realtime bump / specular maps etc, if they become feasible within SL).

Although I am primarily a polygon modeler, I can see that 3DCoat could be quite useful for organic modeling via its voxel abilities, which you are no doubt already familar with.

Voxels, by their very nature, are extremely dense in regards to geometry detail, and hence are far too heavy for viable usage in SL unaltered. This is where the retopology aspects of 3DCoat are really needed, as the 3DC staff have already mentioned to you. This is the section where you can "build" a polygon mesh around your voxel shape - which then would be a viable mesh asset to optimise for SL. Ideally, the polygon mesh would be a recognisable shape compared to your voxel sculpt, although obviously you need to allow for it to be somewhat angular/blocky due to the realtime rendering requirements of SL. It wouldn't be as super smooth as your original voxel sculpt, but should still look quite reasonable.

However, DO consider that you can use your original voxel sculpt to generate a "baked" texture from, which you can then apply to your low-polygon mesh shape equivalent. I'm not sure if 3DCoat can actually bake a texture yet (this is where the model is rendered, and the result is transferred to a flattened UV texture map). However, this can be achieved via outside applications if necessary (Blender etc).

So my suggested workflow (from what I understand so far of 3DCoat) would be this:

1: Create your voxel sculpt
2: Use the retopology tools to wrap a polygon mesh around your sculpt. CHECK THIS PAGE FOR VIDEO TUTORIALS.
3: UV-map your polygon mesh (this is where you prepare your mesh so it can have 2D textures "wrapped" onto it). 3DCoat has some pretty decent UV-mapping abilities from what I have seen. TUTORIALS HERE. (There are a couple of English tutorials near the bottom of the page). I also believe 3DCoat has an automatic UV-mapping option, which could be a time saver - but keep in mind that the resultant UV-map would probably only be usable for painting on within a 3D workspace (3DCoat etc)... (ie: It would probably be a nightmare to paint on as a 2D texture space).
3(a): IF POSSIBLE within 3DCoat, it would be worthwhile converting your quad-based mesh into triangles (for SL compatibility). Personally, I prefer doing this manually for better control of the resultant triangulated mesh, but automatic triangulation is quite viable as well (IF it divides your quads into TWO triangles each, for efficiency reasons). Generally, converting a mesh to triangles is better to be done AFTER UV-mapping, since most UV-mapping tools are designed for quad-based meshes for better results.
4: Paint your polygon mesh via 3DCoat (in the 3D viewport). TUTORIAL HERE. Keep in mind that displacement painting etc won't translate to SL - stick to diffuse texturing for now (basic texturing/colouring), UNLESS you intend to create bump maps etc for baking out in a separate application (where you use the bump/displacement/specular textures etc you create to "bake" out a flat texture map). When you have your painted texture completed in 3DCoat, you should be able to save it as a flattened texture, ready to be applied to the mesh in SL.
5: Export your mesh as an .OBJ file (I don't know from memory if 3DCoat can export in DAE format).
6: Assuming that you can't get a DAE from 3DCoat, just use a recent version of Blender to achieve this (no learning needed - simply use Blender to import the OBJ file, and then EXPORT it as a DAE file.
7: The resultant mesh should be ready for SL.

Of course, you need to consider mesh optimisation etc etc, but the workflow above is what I would probably use to create meshes and their subsequent LODs for SL.

On a side note: I currently use the 3DCoat paint room to create bump and specular maps for my meshes, which I am experimenting with to bake out texture maps, either to use directly in SL, or modify in Photoshop.
Once the materials project LL are working on is released (someday in the future), I am guessing that realtime effects created by bump/specular maps etc will probably be possible (most likely in a low res format, but would still look lovely). In this case, 3DCoat would be a perfect tool for this.

I hope my suggestions above will help give you a foothold in 3DCoat for SL - From what I can tell, it's definitely a powerful program worth coming to grips with, even if it's not perfectly suited to polygon modeling. If nothing else, it will let you create meshes which can be exported to other 3D applications for tweaking, if need be.

:matte-motes-smile:

Link to comment
Share on other sites

Maeve..... :)  thank you so much for your response and the time it took to write it.

 

Many parts of what you mentioned and suggested I already knew or discovered, BUT, it also was valuable because it told me that for the most part the limited texturing quality and lack of the quality of texture map I created in Sculptris or 3DCoat is normal for SL.  I thought the poor quality of the texture I got into SL for my model was just because I was doing something wrong.

Since I am not familiar what to do with the normal , texture, bump map images that Sculptris or 3DCoat can output (i.e. I dont know how they are embedded into the OBJ or DAE for import into SL) I simply tried to blend the 3 layers toegether in Photoshop and get them to look as good as possible as a single texture map layer then import this PS texture into SL seperately from the model import.

In one case I stumbled thru Blender enough to add this texture onto the imported OBJ mesh and exported it to a DAE.  It actually showed up in SL with that texture although not that great quality.  So I still think there is a trick on how to merge the imported OBJ and the texture,bump, normal maps in Blender to make it look a bit more high quality in SL.  I have seen some of the other SL models with texture and they look much more high quality and accurate than mine.  SO... I am still not there on this.

As for VOXEL.... ohh yes.... I love this form of modeling.  I am very much a fan of the ORGANIC models and sculpturing.  That is why I have made a lot of money on Landscape sculpties like water rocks volcanos terrains... these all fit what I love to make.  Now with me wanting to make artistic statues... Voxel modeling is AWESOME!

And in the past couple weeks I am really learning the differences between voxel and mechanical / polygon modeling.  With VOXEL modeling you are dealing with a real clump of ever expanding clay.  Unlike Polygon modeling where you are only working with the SURFACE/SKIN of the model, with voxels you are literally working with countless 3D cubes that not only make up the surface but also full the inside of the model.

I did also learn just recently and thank you for confirming my underestanding - that voxels do not have a mesh of polygons around them.  This was the fact that was missing for me and why I couldnt understand why modeling in Sculpris was so different than Zbrush... With Zbrush the model has polygons right from the start and voxels dont. SOOO I just learned you have to actually topologize the voxel to lay a skin of polygons around the shape.

I have not by any means perfected how this is done effectively in 3Dcoat although supposedly it very powerful.

QUESTION:  So about UV-Mapping.... is it ok if the UV has a lot of islands?  In Zbrush when I take a Sculptris model and UV MASTER it, I often get one UV Flattened layout.  With 3DCoat it always creates many islands.  I was told islands are bad for SL efficiency.  Is this true or does it matter if the UV has a lot of islands?

QUESTION:  Are quads or triangles the better mesh for ultimate import to SL?  I dont know where I heard this but I thought someone said avoid triangles if you can and stick with quads.  I will say that triangles are easier.  Also, is it ok if a mesh has both?

POINT FOR OTHERS:  I have talked to the 3D Coat staff (they are very quick to respond and help) and although they do not have DAE export - they are working on it but the creator said it is a complex export to develop.  So sometime in the future 3DCOAT will have DAE directly and then no more need for Blender!!

QUESTION:  I have been using Blender for exactly what you mentioned.  Its the only DAE converter I know.  I can get the OBJ into blender but I am completely lost on how to attache the textures to the mesh in Blender.  I somehow did it once - so frikken confusing - but not sure how i did it.  Is it better to figure out how to attach the texture, bump, normals in blender and export these textures into the DAE or to create the texture in Photoshop and import it independently into SL?

Suonds like this Materials Project from SL is the major missing peice to make my life easier and to let SL see the amazing texture painting i can create in Sculptris and 3DCoat.  But I suspect this is a lont time away.

The one reason I like 3DCoat over Sculptris is that sculptris cannot render images in the tool beyond the resolution of the screen.  3DCoat can.  I not only create 3D models for SL but as much to create 2D art and I need high resolution image renders from my 3D model to create my 2D art.  here is an example of my 2D art from a 3D sculptris model (that was subsequently brought into 3DCoat for rendering)...

 

Swimmers-TheArrival-900x675.jpg

thank you again so much for your invaluable input!!

Link to comment
Share on other sites

Firstly, I'm glad my information was helpful to you - and yah, 3DCoat is a seriously awesome program (which I seriously need to learn how to use properly!)

Now... to answer your questions....  :matte-motes-smile:

QUESTION:  So about UV-Mapping.... is it ok if the UV has a lot of islands?  In Zbrush when I take a Sculptris model and UV MASTER it, I often get one UV Flattened layout.  With 3DCoat it always creates many islands.  I was told islands are bad for SL efficiency.  Is this true or does it matter if the UV has a lot of islands?

Hmm... I guess it would be preferable to manually create your UV-map, in order to have logical 2D islands (assuming you are using the auto UV generator). UV-maps by their nature are an artform in their own right, and every model will have a different approach to create them. Separated UV islands are most times necessary, depending on the complexity of the mesh. I would guess that the more detailed your UV-map is (in regards to the number of islands), the more data is needed to describe them (I am only guessing here; if your UV map has a huge number of tiny islands, I would assume the data involved would be larger than a simpler UV). Not really sure how MUCH this would affect SL performance though - someone with greater experience in that area would need to answer that aspect. But personally, I would prefer to manually create my UV-maps by hand (creating the islands in a logical way), which will then give you greater flexibility to texture them in 2D via Photoshop etc.

QUESTION:  Are quads or triangles the better mesh for ultimate import to SL?  I dont know where I heard this but I thought someone said avoid triangles if you can and stick with quads.  I will say that triangles are easier.  Also, is it ok if a mesh has both?

In regards to quads... it's DEFINITELY better to model WITH them for sure - most polygon modelers' toolsets are designed for quad-based mesh making. The only reason you need to convert your mesh to triangles is due to how SL (and I assume the majority of game engines) render them in realtime. Generally I will leave the triangulation step until I have finished the modeling and UV-mapping steps, to make life easier. Blender has a handy triangulate option in its OBJ exporter... so I often use this function to convert my quad-based meshes into triangles.
Generally, for 3D RENDERING (non-realtime / non-SL), most programs create far cleaner renders with quad-based meshes; triangles will often create render artefacting etc.
But for meshes you are creating for SL, a mixture of both quads and triangles won't really be of consequence - the eventual upload into SL will be automatically converted to triangles. I just prefer to create my own triangulated meshes beforehand, so I can tweak them to my own preferences if necesary.

QUESTION:  I have been using Blender for exactly what you mentioned.  Its the only DAE converter I know.  I can get the OBJ into blender but I am completely lost on how to attache the textures to the mesh in Blender.  I somehow did it once - so frikken confusing - but not sure how i did it.  Is it better to figure out how to attach the texture, bump, normals in blender and export these textures into the DAE or to create the texture in Photoshop and import it independently into SL?

As far as attaching textures - I wouldn't worry about that prior to uploading. If your model is UV-mapped, and you have created textures which match the UV-map, simply upload them into SL separately as you would a standard image. Upload your mesh without textures. Once in SL, simply apply your texture to the blank mesh via the texture section of the inworld build menu - the mesh's UV will automatically wrap the designated texture as you intended it (IF it looks unusual for some reason (rarely), try tweaking the texture rotation options. (For me, sometimes Hexagon will rotate the texture by 90-degrees by default)). Nice and easy, plus the texture you upload separately will probably be of better quality (I THINK the uploader might apply compression to images which you include with meshes, from what I sorta remember reading).

..........

The eventual LL materials project sounds very promising. If you are familiar with any reasonably modern games, you have probably already seen realtime bump-mapping and specular mapping (not sure of the exact terminology (forgive me Chosen One if you are reading this LOL)). This is where the various texture channels on the game models can add extra detail to relatively low-polygon models. Bump/Normal maps, although only flat textures, can describe height information, and the game engine can create shadows with them on the models (the textures themselves can self-shadow). The same with textures in the Specular channel - these can create fake shiny "bumps", even though there isn't any real 3D geometry at that fine detail.
I am guessing (and this is just an assumption on my part) that eventually, the LL materials project will incorporate these for SL... so say, for example, your voxel sculpts... if you baked out Normal maps and Specular maps from them and had them applied to your lower polygon models, in theory, they could APPEAR to have much more detail than the geometry actually has. Of course, there would be limitations to this (I am guessing the extra channels would need to be lower resolution than the maximum 1024x1024 diffuse textures currently available (the basic textures we use in SL now), however, they would go a LONG WAY to giving extra lovely detail to our low poly meshes.
(Chosen One etc, feel free to correct me here LOL).

:matte-motes-smile:

.......

As far as BAKING textures go... this is something I am still exploring the workflow for. I am going to have to learn this in Blender, since my preferred programs (Hexagon and Carrara) don't have baking options built-in.

Below I have a few images of a boot mesh I am slowly working on (so many distractions in RL are keeping me from finishing it LOL), which show the potential of Normal mapping via 3DCoat, and how it can be used for texture baking for SL (when i make that final step via Blender):

MB Granny Boot LOD1 mesh example.jpg

ABOVE: This here is my basic boot mesh, triangulated and UV-mapped. I created this in Hexagon, however you could achieve the same via 3DCoat's retopology tools (with mayhaps a bit of triangulation via Blender and reimported etc if necessary).

3DCoat Normal map example.png

ABOVE: This is the same mesh, imported into 3DCoat's paint room. Here I am using 3DCoat's tools to create a Normal map (this is a texture which simulates bumps/depressions in the mesh, which aren't really there in the geometry). This adds a lot more potential detail than is feasible for realtime meshes. (This is still a work in progress, by the way, and nowhere near completion).

Normal map generated via 3DCoat for mesh usage.png

ABOVE: Now this purple mess is the bump mapping on the boot in the 3D viewport, EXPORTED from 3DCoat as a 2D Normal map. You can see the stitching and fine leather pebbling details. This map matches the mesh's UV-map, so will automatically wrap onto it.

Manually created 2D texture UV-map.jpg

ABOVE: This is a manually created texture map to match the boot's UV-map (with a pre-baked ambient occlusion texture (created with SMAK) merged via a multiply layer). This is a standard texture which could be used as it is in SL, which would be fine, but I want to take it a step further...

Mesh render with Normal Map & Diffuse Texture applied.png

ABOVE: Now this is the same boot mesh, rendered in Carrara, with the 3DCoat created Normal Map AND my manually created texture applied. I have applied a basic shine to the mesh via Carrara materials (but nothing special). See how the rendered light plays across the Normal Map leather grain and stitching? Definitely adds some extra quality to the look of the model. Now with baking, you can actually capture this kind of look and create a flattened texture from it (keep in mind that with SL texture limitations, the resultant bake probably wouldn't have the same amount of detail here in these renders, but should still be very acceptable). Obviously, the lighting won't be dynamic, since the texture itself is static, but definitely easier than trying to paint this look by hand. Since Carrara can't bake out textures (especially with the shine etc), I will have to teach myself Blender for this part (I know of others in SL who can do this, so I will be prodding them to show me LOL).
Also, 3DCoat can create Specular maps as well - so you can actually control how the shine effects work, and how strongly etc - the resultant Specular maps would just be applied to the mesh in another material setting, which could then be baked out.

Phew... I think I'm all typed out LOL... Hopefully I've given you some food for thought in regards to 3DCoat.

Have fun! :matte-motes-smile:

Link to comment
Share on other sites

lol WOW Maeve!!  Thanks again for all your knowledge and input!

and your boots look frikken awesome!!  I wanna see that texture in sl even if i have to fake it with PS.  So you got those boots in SL with that level of textue look??

As for 3DCoat... just as I think i was starting to figure it out a bit the damn trial ended.  I am gonna now see if i can do as much with Sculptris.  So far even with all the cool tools with 3Dcoat... I have had more success with the hugely ugly workflow of:  Sculptris to shape and create the reduced polygon mesh (sculptris's hand brushed mesh reduction tool is cool), then export it to Zbrush by OBJ, Use Zbrush UV Master to apply a real simple UV Map, export OBJ back to Sculptris and paint it.  Export mesh and all the textures out.  Use photoshop and blend the normal, bump, texture, shadow maps together to make it look somewhat close to the image in sculptris, the import to blender and export to DAE and into SL.

:)

I might re-install 3DCoat again on my laptop to get another 30 days and hopefully feel comfortable with it.  I dont wanna buy 3Dcoat until i feel that it is the tool I want to use.

 

But all you mentioned really is helping my sink the terms concepts flow into my lil brain.  I will admit some of what you said was a "whaa" lol but i will read it again.  Most of what you wrote was perfect as u even explained what most of the terms meant.  A lot of the mesh experts dont.

 

You r awesome!  thanks!

 

Link to comment
Share on other sites

Aw thank you, I'm glad you like my boots thus far... that final image is only a RENDER from within Carrara. However, my goal is to get that kind of look (WITH the shine effect) converted into a single flat texture - the final result should look similar to that in SL, even if it's a lower resolution texture overall.

:matte-motes-smile:

It's a pity that your 3DCoat trial has expired. Keep an eye on their website... occasionally, they have special sales with discounts to the software. Of course, it also depends on whether you are comfortable with using the software itself, as to whether its worthwhile purchasing it.
Keep in mind that the workflow you described is perfectly adequate, if it gets the job done for you. My own workflow is all over the place LOL - I jump between a lot of different 3D programs in my collection (gradually acquired over the past few years as I have gotten more and more into 3D modeling). Quite often, a specific program such as 3DCoat will have a very specialised task that it excels at, when compared to other programs - and for me, things like that can make them worthy additions to my toolkit.
Obviously, it also depends on your personal budget as to how much you want (or can afford) to spend - 3D can be an expensive addiction LOL! But the freedom of 3D creativity is hard to resist!

:matte-motes-bashful-cute-2:

.......................

ohh i got one more question....

Says mesh can have multiple sides/surfaces that in SL can be independently applied with texture?  How is that stated on a model?  or is it that if the corner is sharp enough - its a new surface?  I am sure its much more complicated then that.

Each mesh object can have up to EIGHT different texture faces - these work in the same way as texture faces on normal SL prims (as in, how you can change textures in the build menu etc). BUT with mesh, you can define these texture faces in any way you see fit - and these texture faces don't necessarily need to be in one spot on the mesh - they can be split up into segments across the mesh, and still act as the same texture space (I hope that is sorta clear LOL).

To define these texture faces, you need to define MATERIAL zones on your mesh (this is nothing to do with the LL materials project, by the way - that is a totally different thing with a similar name (just to make things confusing!)).

Think of MATERIAL ZONES as defined sections on your mesh - selections of triangles etc. You basically work through your mesh object, assigning batches of triangles to different material zones (up to eight in maximum). The process will vary from program to program, but it's a fairly standard thing in 3D overall. So you work until ALL the triangles in your mesh are assigned to one material zone or another. (You will need to assign ALL triangles in a mesh object to a material, otherwise that will cause problems with the uploader - it doesn't like unassigned triangles in regards to materials).

(The material zones still use the exact same UV-map as your mesh does).

Each of these material zones can then take a totally different texture if need be (these would all utilise the same UV-map your mesh has), or these material zones can just be used to change up the colour, glow etc via the SL build menu (just like with prim faces). Very handy! To give an idea - my boot from my previous post has a number of material zones - the heel, laces, toe cap etc (eight in total) are defined as different material zones (they are all still part of the same mesh object though). So if I chose to, I could tweak the colours of these sections to mix up the look of the boot, or even have a totally different texture applied to these zones etc.

Definitely worth learning how to create them - materials will add extra versatility to your mesh texturing options.

:matte-motes-smile:

Link to comment
Share on other sites

Pretty complete as usual.. I got one small thing to add at this point...

It's not the number of UV islands that determines the load in SL, it's the number of vertices.

An example... If you make a sphere you can make a single island, with one stitch or UV edge from pole to pole. The UV map will be a square where the pole vertices are repeated a good number of times and all vertices on the line from pole to pole aswell. This is pretty efficient, most vertices are represented once. You can also make all triangles making up the sphere single islands (for example purposes, this is ofcourse stupid for any model). Then every vertex is represented six times (if all triangles are orientated the same way) on your UV map and SL will split the vertices accordingly. SL calculates the weights by vertex, not face, so you can imagine the extra impact.

The best thing you can do is use as little edges as possible and let the seams match seams of smoothing groups. SL splits the vertices on smoothing edges aswell.

I'm not entirely sure if the actual load on the system is represented well this way, but for Landimpact this is very important.

Link to comment
Share on other sites

To put a quantitative gloss on that ...

For a simple 24x24 quad torus (like the SL prim), accounting for just the triangle lists (three 16bit indexes per triangle) and the vertex list (eight 16 bit values per unique vertex*) gives uncompressed values of 16912 bytes for the smooth torus with maximally connected UV map, and 62208 bytes for the same mesh with all triangles separated in the UV map. That is a ratio of 3.68 that should approximate the ratio of download weights.

Uploading these at 2x2x0.5m, using the same mesh at all LODs gave a ratio of 3.24 (LIs 107.625/33.180). Uploading at the same size with all autoLODs gave a ratio of 3.68 (LIs 4.19/1.14). So the effects of compression on the ratio of download weights for a fragmented UV map are small, and the theoretical prediction is fairly accurate.

Note that this is the maximum effect. Where vertices are already increased by different normals (sharp shaded edges) or by lying on material boundaries, the effect will be smaller. Also, complete fragmentation of the UV map is very unlikely to be approached (although there is one program that does this always .... I can't remember which).

Take-home message - worst case fragmented UV maps can nearly quadruple land impacts.

*Unique vertex = distinct values for any position, normal, uv coordinate or material.

 

Link to comment
Share on other sites

OK Drongle and Kwakkekde

:) I read both your posts a couple times and you might have well talked greek to me.  I am glad you both have the intimate technical depth understanding of what drives / impact LI.... but its conversations that this depth that scares away a lot of new SL Mesh Creators - even ones that were successful as Sculpty creators.

I sure do not mind that you say all you said in there to explain what you were trying to explain.... but since the forums (especially this thread where the OP... me... mentioned that I am trying to grasp and understand the basics) is to learn and educate those noobs of mesh like me.... it would be a great idea to at least finish off your posting to make a 1 line summary of what the heck you were saying.

What I really loved about Maeve's postings were that even though she got somewhat technical she tried hard to translate to layman's understanding of the 3D modeling industry.

So let me ask you both... related to the question I posed and that you both responded to...

IN GENERAL... to the question initially asked:

1) Does a model with a lot of UV Islands significantly impact the LI of a SL MESH?  (Generally YES or NO)

2) Does a model being developed outside SL significantly impact the SL MESH LI if it quads or triangles were used?  (Generally YES or NO?)

I dont really care if an SL mesh is 2-10% less efficient on LI if I go one way or another on UV Islands & Triangles if its easier for me to develop the way I understand it.  If the bad ways ad one or two additional prim counts to LI but its far easier for me to develop and understand then thats ok.  But if having multiple UV Islands will surely increase LI by 50-200% then I better spend time to understand what the heck you both were trying to say.

 

But basically, what you both posted - I had no clue what you said and its these kind of posts that really takes the wind out of my sails in thinking "maybe mesh is just not for anyone but the real tech geeks in this world".... and I am a tech geek in RL which really frustrates me.

 

Link to comment
Share on other sites

1) Generally less islands is better.

It's not the islands themselves though, it's the "shoreline" that makes the difference, make them as compacts as possible.

2) Are you asking if quads are better than triangles for LI? If that's the case..no, quads are two triangles, so there's no difference at all. All models are made out of triangles, quads are not something your compter understands.

Link to comment
Share on other sites

MY goal is to learn enough about the advanced yet minimum required 3D modeling technologies and techniques to be able to express my creative talents in creating great 3D organic models (land terrains, art statues, etc.) that I can already create inside the Sculptris and 3DCoat and get them somehow into SL is a form that looks somewhate close to what I see in these tools.  If I can do this in a way that the SL MESH is somewhat efficient when rezzed in SL - all the better.

But at the end of the day... if my SL MESH MODEL is efficient or not in sl when rezzed - it will still sell to the SL community because unlike building a prefab home or furniture or even clothing.... an art piece is unique and has no direct competition.

As such, if my SL MESH Art Statue has a LI of 2 or 20 really will not be a major factor to the buyer.  If they like it - they will buy it regardless of the LI.   As long as I dont create an art statue that could with very simple topology reduction have an LI of 5 and I dont take these steps and I upload it with an LI of 40.

That is why I so far like the Sculptris mesh reduction tool and mesh reduction brush.  The reduction of the polygons in Sculptris is as simple as hand brushing the poly count under the brush in areas where the count needs to be less (like the thighs and back of the human model) and dont brush in areas where there are a lot of corners and curves.

THAT MAKES SENSE TO ME and its very simple to understand.

Link to comment
Share on other sites


Kwakkelde Kwak wrote:

1) Generally less islands is better.

It's not the islands themselves though, it's the "shoreline" that makes the difference, make them as compacts as possible.

2) Are you asking if quads are better than triangles for LI? If that's the case..no, quads are two triangles, so there's no difference at all. All models are made out of triangles, quads are not something your compter understands.

Thank you! :)

Related to answer #1.... how would one make a "shoreline compact" in the 3Dcoat or Zbrush tool?  Is there a button to press "Make shores compact"?

Link to comment
Share on other sites

Thank you Kwak and Drongle for your clarification. 

Yes Drongle that one line I missed on how it could severely impact LI if there are way too many islands.

The few tests I did on 3DCoat to auto-UV created a UV map with a few large chunks and then like 20 smaller islands.  Looked like a real ugly UV with flattened compared to when ZBrush does the UV mapping via UV Master.  Basically it was a ball with a few snakelike tenticals pulled out from it - just as a test.

 

PS.... sorry if my postings sounded "angered".  I guess i get a bit frustrated when so much of the 3D modeling world is so hyper geekoid and technical when I am just trying to grasp the more basic needed components.  After a month of trying to wrap my head around how to make a reasonable textured mesh for SL...your super tech postings me feel like I get into a lake and start my swim across a large lake .... and then in the middle of the swim people in a boat near me inform me that I am not swimming in a lake ... its an ocean!!

Not your fault..... So sorry for sounding harsh.

Link to comment
Share on other sites

Don't worry about it, there's just a lot that you can consider if you want to optimise your models. Not caring about your objects being 20 prims if they can be 2 is not a wise choice I think, if it's 2 prims, people might buy 10 instead of 1 and enjoy then 10 times more.

I truely understand you don't want to bother with the tool and want the result right away. Unfortunately that's not how it works in any field. It really pays off to try out things here and there and look into it a bit more.

However..the best way in my experience is by doing that at your own pace, your own way. So please just build as many nice things as you can and the rest will come later......

Link to comment
Share on other sites

Yah, I can relate to your frustrations in regards to the technical jargon - When I was first getting into general 3D modeling a few years ago, it took me a while to get my head around everything (especially with all the new terms / procedures / steps that I kept stumbling across, which were often considered as common knowledge by other experienced creators). Information overload, basically... too much to figure out all at one time, and without a general overview of HOW things were made, it was pretty scary for me LOL.

My best advice - Just make things. Just take it a step at a time, and gradually get your head around each of the procedures involved. As you go along, you will develop your own preferred workflow and preferred software tools. The more that you learn about mesh making in general, the easier it will get - trust me on that. At first it's pretty daunting, especially coming at it from a completely fresh start (it's a big subject, after all)... but as you build your general mesh knowledge and grasp the various steps involved, everything will become logical. Once you have the basic general steps figured out, then you can concentrate on refining your skills - especially in regards to optimising your meshes for SL.

Getting low Land Impacts and low render costs for your meshes can be fairly technical... so for now, I would just concentrate on getting your mesh making workflow figured out and keeping your meshes relatively low in polygon count (as in, using as few polygons as possible to adequately describe your geometry (it's a balancing act)). If you can keep your polygon/triangle counts at this kind of low level, it will go a long way to helping reduce your Land Impact. After that, once you have gained confidence in general with mesh making, then would be a good time to get into the technical side of SL mesh optimisation for low impact.

Also (you probably already know this, but I'll mention it regardless), don't forget that the Aditi test grid sandboxes are really useful for when you are tweaking meshes - you can upload there without it costing from your own pocket - which is especially helpful when you are trying to optimise your Land Impacts. Upload there over and over if need be, until you get your tweaking done, and THEN upload the finalised result to the main grid. It's a real life saver in regards to costs.

Oh, and be sure to ENJOY making meshes - that's the MOST important part! Don't let it frustrate you, just ENJOY being creative!

And feel free to ask questions at any point - I am always happy to provide answers if I have the relevant knowledge!

:matte-motes-smile:

Link to comment
Share on other sites

Thanks Maeve for the support and telling me the ocean isnt that large a body of water to cross :)

And yes you are right... Instead of boiling the ocean... I am just gonna practice going through the topology, UV, DAE conversion process now.  hopefully as i keep doing it - something will click each time.

So here is an interesting puzzling question for you and any other SL Mesh Import Experts or maybe even LL Staff themselves....

I have been able to do this a few times and successfully with no errors and the Rezzed model looks clean.

As I write this posting, I am in the SL Beta Grid and floating over a natively rezzed 9.9m x 8.6m x 8.6m simple deformed sphere with 308,384 vertices & 154,192 Triangles (as reported in the Upload Model import tool).  It has a rezzed LI of 428 and 2 prims.

I was told its impossible for me to SL Import a model greater than 64000 vertices.  Well clearly it seems 64000 is not a limit - not by a long shot.  And in the beta grid it did not even take long to rez cleanly.  Actually it looks smooth as silk :)

 

Link to comment
Share on other sites

The limit is 65,536 vertex table entries per material. That's 524,288 total, unless there is another lower limit I don't know about.  That has to be reduced a couple of thousand because of the extra vertex table entries needed along the edges of material boundaries. This is fixed by the internal data format which uses a 16 bit index for the vertex table, so that it is impossible to index any more vertices. If the surface is smooth shaded, the vertex table entries will be one per vertex in the drawing program. So a sphere of 500,000 vertices is theoretically possible. Of course, if it's made of several linked objects, it could be many times that. The LI would depend on the low LOD meshes, whch could all be just one triangle.

I just uploaded a sphere with 64,512 vertices (uploader count), with two triangles for all lower LODs. It has LI of 2 at 2x2x2 and 33 at 8x8x. So the LI you give for your more complex sphere is consistent with that. What is the physics shape of your sphere? I would expect small triangles to make a prohibitively large physics weight, unless you provide a simple one.

Link to comment
Share on other sites

Dont know if any of you 3D folks have heard but there is an AMAZING (hard to even believe) promotion right now from DAZ3D...

For the month of February...  DAZ 3D is offering a fully licensed version of their

DAZ STUDIO PRO

BRYCE 7 PRO

HEXAGON 2.5

FREE !

Hard to believe but this week I registered and logged in and bought through their shopping cart and downloaded all three of these packages with their provided licensing.  I installed all three products.

Here is the link if any of you were interested in DAZ 3D products...

http://www.daz3d.com/i/3d/free-3d-software-overview?home_5_btn=start

 

Link to comment
Share on other sites

Actually I briefly tried working with zbrush model texturing a while ago but it was no intuitive so i abandoned trying to figure it out further.  I only used zbrush to create PLANE type models for my SL Sculpty terrain packs.

My mind seems to wrap much more naturally to the voxel type tools like Sculptris and 3DCoat - and in Sculptris the texturing of the model was very simple and common sense.  Are you suggesting I export my texture, bump, normal maps from Sculptris and import them and the model into Zbrush to merge them together?

My understanding is that in order to get the textures i painted onto a model into SL , they have to be integrated together with the tool that will be executing the Collada export (i.e. Blender).  Zbrush currently has not export to .DAE yet (wished they did - wished Sculptris and 3DCoat did too).

Please help me understand what you are suggesting.

Link to comment
Share on other sites

i'm not an expert in regards to all those types of maps. that's actual what i'm currently interested in understanding better. but i can give you some help in understanding zbrush better and how you can do a LOT with just that program in regards to sl and mesh over all. truly an amazing program. don't get thrown by its interface and jargon, its actually very easy to do things and now that the new version has "dyna mesh" its really tremendous. you want to do things for sl right. ok. there is a sl plugin you can get from their site that prepares the model and exports the correct dae file for sl including the texture. so export is taken care of. modeling in zbrush speaks for itself so i will mainly talk about painting and baking. import or create your model in zbrush, lets say we import the standard avatar as an obj. we now have a low poly model. if we start to paint on it we cannot get good detail because the poly count is low and there is no "clear", fixed resolution texture(like 1024, 2048,4096 px) wrapped around the model that we can paint on. so we go to "geometry" menu of zbrush (zbrush calls this a "sub palette") and we click on "sub divide". we sub divide the model until we have about a million or more polygons. if you know before hand the size of the texture you want to end up with, say 1024 x 1024, then you should sub divide the model to little over a million polys since 1024 x 1024 equals a little over a million pixels, you want a 2048 x 2048 texture then you should sub divide to a little over 2 million polys and so on. the idea is the each poly serves as a pixel. now that you have enough polys when you go to paint you can now paint as fine details as you desire. now after you've painted your model and you have some sort of materiel applied you want to bake that materiel and highlights into the texture. for this there are two ways i know of in zbrush and they produce slightly different results, one is using the image plane plugin. i won't get into that here but there is a good video on youtube for those interested. the other way, the one i will explain, is using "zaplink" from within zbrush. zaplink works with and external editor, like photoshop, it allows you to export and re import snapshot of your textured model. on the top menu of zbrush click on "document, then click on "zaplink" it will prompt you to select a external program to use. browse to photoshop and select it, click ok or whatever. now zaplink knows which program to send the shots to. now in the same "document" menu, at the bottom you will another "zaplink" click on that and a menu opens up the says "front" "back" "right" "left" and so on. position your model squarely for a front view and then click on "front" in the zaplink menu. do this for each of the six views, front/back, left/right, up/down. you only have to do 3 views since the opposite side gets set automatically. now have zaplink all set up. click on the same "zaplink" you did to pick your editor and zbrush will begin to snap views of model and then launch photoshop where an image file with 12 layers will be waiting for you. 12 layers, 6 layers are the different views and the other six layers are "shader" layers for each view. in photoshop the "shader" layer will be locked. what you have to do is duplicate each shader layer and then merge it down on to is corresponding view layer. do this for each view, you should end up with 12 layers again.(people who use paintshop can just merge the shader layers down witout duping them). click save and return to zbrush, you will be prompted to except the edits, click "accept". you have to do that for all six sides. you now have a baked texture on your model. note the model may look darker than before you accepted the baked textures. that because you like still have the same shader materiel active in zbrush, switch to something neutral like a skin shader and the model will reflect how it will look in world. ok, you painted and baked your texture, now you have to save it. go to zbrush's "UV" sub palette and choose the size texture resolution you want (sl only renders 1024 x 1024 regardless of size you upload). after you selected the size go to the "texture" sub palette and click on "new from polypaint", then right next to that click on "clone texture". the texture(with baked highlights) will appear in the texture window usually found on the left of the work area. from there you can export it as a psd, bmp or tiff file. if you model did now have UVs you would have to first give it UVs using UV Master plugin. ok, now you painted, baked, and exported a texture. if you also want to export the model for use in sl. then this is the time to reduce the poly count, AFTER you already created and saved the high res texture. you can do this by either going back to "geometry" palette and reduce the sub divisions you made back down again or you can use zbrush's decimation master plugin, which does a great job a reducing polycount greatly without losing detail. ok now you have your texture and model ready to go. under "zplugs" you have the sl plugin you got and installed from their site. click on "sculpty" and you'll see mesh menu. just follow that and in a couple of clicks you have a perfect dae file ready for sl. just upload and you are done. hope i've helped you find your way a bit.

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 4529 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...