I dont have a feeling for polygons and what is to much and what is ok..

[Miyo Darcy]

---

Glad to hear and that sounds much much better, hope the new PE is good

Its a really tough question on how much PE a model should have, the one thing is if you have the prims spare you can enjoy a much higher detailed model. Its something I found myself constantly thinking about when planning my builds and it felt limiting. I decided for my first 'sellable' building to just go with the creative side rather than the technical side and I expect my first luxury home to be around 1000PE which is high, yet when I compare it to other builds ingame right now I would say it will be much much better than those 2000+PE luxury mansions you see.

I may be wrong but I think the first thing a customer thinks about is the visual appeal, if they like it they will then start examining all the other aspects like PE and footprint ect and in some cases be willing to sacrifice those over looks.

If you are intending to sell it, you could always create 2 versions, a HD version and a low poly version, its actually quite easy to do that once you have created your LODs, just use the medium LOD as the high LOD and create a simple low LOD to put into the missing LOD slot while using a default 2tri for the lowest one.

---

About the PE:

I still could not find out the PE of the object. When I try to import the collada, my mesh viewer will show no reaction, and I dont know whats wrong. The mesh viewer will just break when I import an when the import preview window is there.

A simple cube export will work. But not my staircase. The viewer will crash then.

I assume that my railings are the problem as they are made out of one circuit spline and one helix spline.. and those are under a Sweep-NURBS Object. I did read anywhere that NURBS is not supported for import. To check that out I will delete just the railings and make a test import without railings. If they are the problem than Im not sure what kind of other methods I have with cinema 4d to create those railings. A sweep nurbs with splines was at least the handy and in my opinion best solution (at least in workflow). Maybe I can use a simple cylinder as well and just bend it.. but im yet not aware if I have a tool in c4d which can make it like I want.

But maybe I can also convert this finished NURBS railing to another object or so. I must try that out.

 

About your last idea.. Yea I will probably sell this and that when I start to be aware of all the import basics and when I start to feel good with all that.

I pretty much like the idea to sell low poly and HD versions. I propably would just put both in one package so that the customer can decide what to use. Thats at least how I would like it if I go on virtual shopping tour

[Miyo Darcy]

Ok its working now. Anyhow I had to load my collada files again with my 3d application.. then close appliaction without saving.. now I can upload and see my staircase in SL.

It has 34.000 PE

I guess that is still much.

Inworld the staircase looks good if I am close with my cam. But just three meters away with the cam I wont see objects but instead of that objects change to triangles. Hmm. Any idea?

[York Jessop]

Those triangles are the generated LODs (Level of detail) which are directly linked to your 'Object Detail' slider in the graphics settings, if you move the slider up it will show the highest LOD much quicker.

LODs are in 4 stages, High, Medium, Low and Lowest, High being the model you made and imported and the last to load as you approach it. The upload window then lets you either generate the rest of the LODs (which reduces triangle counts and why your model starts looking like a bunch of triangles) or you can upload your own LODs which are basically simplified models of your main model. Basically continuing what you have done already in reducing poly counts, the benefits to this is you can control the appearance of your model much better and if you do it right you can lower the PE even more.

The way it works is that your lowest LOD will count for the most in terms of PE while the highest counts for the least, the reason for this is because the lowest LOD will always render first and for the longest in most cases.

Alot of people use a 'billboard' (just a flat square) for the lowest LOD as in most cases the majority of residents will only ever see it from far far away, personally i let the upload window generate the lowest LOD with the lowest amount of tri's (you can control that too)

 

Im not sure what kind of business you run or are intending to run (do you just sell building components?) But with meshes you can control everything including your work flow and methods, to give you an example, in the house that im building im taking it in 2 parts, the first is the exterior shell, this has all 4 custom LODs but only has exterior walls and features there is nothing on the inside, the 2nd part is the interior shell, this only has the 2 highest LODs as the interior cant really be seen or needs to be seen from far away, this give me alot of control over PE counts and makes it alot easier to work on.

I guess if your working on interior features like stair cases, how far a resident is ever going to be from it until its behind a wall or obstruction should be something to consider, you could in theory reduce the 2 lowest LODs to 1 triangle which should help even more with PE.

 

[Miyo Darcy]

That was great explained. I did not even notice that I can click in the level of detail section there. Thank you. I will play around with that now to get a little feeling for it.

But I also notice this will be a lot of learning about the import and a lot error and try.

I see its not enough to know how to create some simple or intermediate objects in the own 3d applications. I see we must know learn how it goes together with SL. So I am pretty much sure that I am not bored next weeks

About my SL business. Last time when we only did build with prims and later sculptys, I did build just things that came in my mind. So overall I didnt specialize in anything. From marketing view it might be crazy as I have differend objects in my shop and the customer can get confused lol. But I do all that for fun and I create things that come into my mind and if there are some incoming L$ I am happy one more time. But you can say I did build mostly club objects like DJ Booth and so. Yea you could say they are also building components at least for clubs.

When we talk about mesh I have a text file here were I drop my SL ideas.. So I had this staircase in mind.. some cool club speakers which might be a simple 3d job. DJ cases.. maybe some new DJ bothes but this time with shapes that cant be done with prims (rounded). Well tables and and and. Many ideas.

Lets summarize.. I think you are right with interior and building components. But also architecture.

So yes.. the interior stuff could be done with lower LODs then. I understand what you mean. It must only be seen in the area where they are placed.

Again thanks for the great explaining. I think that might give me tons of new question lol, but thats the reason why we all play SL or?

I assume when I played and understand the LOD things.. physics will be the next thing wich might give me little headache

EDIT:

I played witht the LODS and fixed the problem with the triangles. Thanks so much for that cool info

I start to understand it now.

I did decide that full details for high, medium and low are great for the staircase as long as I dont make own lods. But I put down the lowest lod as this is far away enough. I still can make HD versions as well.

 

 

[York Jessop]

Ah i see, the one thing I think you might find over time is that what you can create with mesh will blow its prim version out the water, in terms of looks and prim counts. Appart from modeling, texturing, LODs and physic's shapes thats all there really is to know, to be able to sucessfully import meshes, I would say your pretty much there

Physics shapes are just very simple mesh models that are invisable to you, the one important thing they have to be is generally the same dimensions of your main model (height/width/depth) as when you import it it will use the bounding box of your model for its position, you know when people make scuplts, make them phantom and then use invisable prims to make the bit you can walk on? Its the same principle with meshes.

On the physics tab, you can import your physics model and it will appear in the preview window, this lets you check if its positioned properly, or you also have the option to use any of the LODs as your physics shape too, this is where the benefits of custom LODs comes in again, I try to use the Low LOD as my physics shape.

A good work flow would be:

High LOD > Your main model

Medium LOD > Your low poly version

Low LOD > Your physics shape

Lowest LOD > Your 'billboard'

 

By using the above as a template this is how I go about making a mesh:

1. Create the high detailed model and texture it and save it as HighLOD

2. Using the HighLOD model, I then start stripping it out, reducing polygon counts and details until im satisfied i have a good low poly version, tweak textures if needed, then save it as MedLOD

3. Using the MedLOD as a guide, I start 'boxing' my model, as in making a super low poly version that only accounts for where I want to it to be solid, instead of having those curves I would square them off, the idea being to have the lowest possible poly count without losing the basic shape, textures dont need much tweaking at thos point as its a model that will either be something you see from far away or invisable as a physics shape, this would be saved as LowLODPHYS

4. At this point I would import all 3 LODs, and let the Lowest LOD be generated by the upload window, then I would click the physics tab and set the physics shape to use LOD and then in the drop down select use Low LOD. I would then click close edges and then analyse, to check it looked ok in the preview window. If it does then I would make sure textures is ticked in the last tab and then upload it ready for use.

4a. To create the low poly version, I would do everything in step 4 again except instead of importing the HighLOD, I would import MedLOD twice using them for both the high and medium LODs

5. Once imported and rezzed, I would right click > edit > click features tab and then change 'Convex Hull' to 'Prim' under Physics Shape Type. That will make sure its using the physics shape you uploaded.

 

Of course I would do all of this on the beta grid first

[Inea Wemyss]

After reading this thread, I am starting to wonder if I have always worked backwards? :smileywink: 

I always start with a blocky shape and work my way up from there. As I work up, I save the objects as the different LOD levels until I get to the highest LOD, where I am happy with the shape and number of polys/tris. But still keeping it low. At a low LOD level I also create the UV-map.

I then "upload" or at least just go to the upload window to see the numbers.Then I will know if I can go higher or have to go lower on the tri count.

The less verts I have to move and work with the easier and quicker it is.

(You = the general "you")

A good practice for learning to optimize your objects is to look around you and draw, and I mean use pen and paper.. , objects you see around you, but draw them as blocky as you can. Then do the same thing in your 3D-program. If there is a change in the surface of the object, look at it from different angles, how far away will you see it in-game (SL), do you really need to add extra triangles for it or can you add the details in your texture map?

And like Tarius said, you don't need any extra polygons on a flat surface. If you're used to working with sculpties, you need to have polys on flat surfaces because of the UV-map, but with mesh you're creating your own UV-map and a flat surface might only need one poly (2tris).

Please note, it's the end result that matters, how you get there is not important, each person has his or her own work flow.

[York Jessop]

---

Inea Wemyss wrote:

After reading this thread, I am starting to wonder if I have always worked backwards? :smileywink: 

---

 

I thought the exact same thing when Gaia mentioned she started from the top and worked down heh, theres nothing in the rule book to say you need to work that way, the reason i do is because it makes it alot easier with textures, what I found was when i started with my basic shape, then move to the next LOD to get more detail I ended up going back down again to make sure the textures were ok, it was like a yo yo effect. By working from the top down you can start with a whole and start removing or simplifying your model without losing texture faces and without having to really 'add' stuff, its way easier imo to do that.

[Miyo Darcy]

---

Inea Wemyss wrote:

 

Please note, it's the end result that matters, how you get there is not important, each person has his or her own work flow.

---

That is so true. Everyone use differend ways to work things out. But it is still interesting to see how other peoples do it. Sometimes there are some mighty hints I or we can use or learn from.

Thats what I like on SL Community. Here are a lot of peoples who share teachnics and know how. I noticed it already very early when I startet with SL 4 years ago or so.

I think I learned much about creating stuff just because I uses SL and because of the peoples who share ideas. Maybe also with lots of tutorials. But all this is fun. This thread here is an great example to learn how you all work. Then we can see if our own way is ok.. or if we maybe try other methods. I agree so much to you.

My own way is really to start with high poly. Anyhow I like that. And after reading the thread I did learn its not that hard to reduce polys after that. I must just go through my objects and see where I spend to much polys and then I can think about ideas to reduce them. I like that. But as you said.. everyone has there own workflow.

 

[Gaia Clary]

It appears to me that the workflow is also somewhat dependent of the model you make and what exact goal you have:

 

• When you want to make a replica, you will probably start with a high poly model and then do reduction
• When you start with a general idea about what you want, then you might start with low poly and go up from there

My personal workflow seems to stabilize now like this:

 

• I start with a low poly model
• Then i add edge loops (mostly using subsurf at the moment)
• I tweek the low poly base until the subsurfed model looks like i want
• I take the subsurfed model and start reducing the edge loops (the more flat an area is the more reduction i can make)
• I repeat this reduction one more time
• From the resulting models i select those which fit best for the 3 higher LODS. I often end up by using the same mesh for High LOD and Medium LOD (i am only making small objects. With big objects this may be not a good idea!)
• I try to get down to a box model for LOD4. Sometimes it is the very first low poly model which i made. Sometimes i make a billboard instead.

Regarding UV-unwrapping i am still unhappy. For now i unwrap my High LOD mesh first. Then i use the result of that unwrap as template for the other 3 meshes. This is a time consuming and frustrating process because it means to redo things 3 times and fine adjust each UV-map.

Of course i could just make a low poly model, then subsurf it twice and use the results as my LOD meshes. But subsurf does not care about topology, thus it creates way too many faces on flat surfaces. So i again end up with reduction and tweeking.

Here is a result from my shoe quest tutorial:

http://www.pasteall.org/pic/show.php?id=16526

HIGH: 2144 triangles
MEDIUM: 978 triangles
LOW: 482 triangles

These results mean to me that it is possible to create smooth low poly meshes. And i even tend to add another LOD and make the 978 triangle mesh the HIGH LOD. In the final version of the tutorial i corrected one topology error on the LOW mesh which makes it even less edgy ;-)

What i am realy missing is something like a "copy UV-layout" feature where i just define the seams of each LOD and the copy function fills in the faces for me. maybe something like that exists as a standalone program ? Blender does not have it.

[Inea Wemyss]

It's interesting to hear about the different ways of making mesh for SL. Making all these LODs makes it quite different than what I am used to from previous work, it's well worth thinking about and trying going the other way, from hi-res to lo-res. Or maybe something in between.

Gaia, I have watched your video tutorials and I must say that they are really good, you've made an excellent job on them!

And I have to ask, is Subsurf the same as "smoothing"?

[Gaia Clary]

I explain for blender:

Subsurf is a modifier which "adds vertices inbetween". Imagine a quad face (polygon made of 4 vertices). The subsurf modifier will place one vertex in the center of this quad, then draw edges from that center vertex towards the centers of the edges of the original quad. Thus it creates 4 subquads out of each quad. This is one subdivision level.

For each additional subdivision level of Subsurf the modifier acts recursively. Hence each level of subdivision means 4 times more quads. Subsurf acts on the entire object, hence it adds unnecessary subdivisions on flat faces as well as it smoothes out hard edges (which can be avoided by adding "crease" to the edges).

Smooth on the contrary does not create new vertices, but it changes the way how light is reflected from the polygon surfaces. When you enable smooth then the light reflection(?) is smoothed near the edges. Thus smoothing tends to hide any sharp edge transitions. While flat will make a polygon appear realy flat, thus you can see all polygon edges.

If you are interested in the technical details, here is my explanation how smooth actually works: http://blog.machinimatrix.org/3d-creation/vertex-lighting-in-sl/

 

[Inea Wemyss]

As I am not a Blender user, I had never heard the term SubSurf before. According to your description it is like smoothing in Maya. And I understand from your description that "smooth" in Blender is something different. That's the way it is with different applications, same term but different meanings. First time I heard of SubSurf, I was thinking of SubSurfaceScattering, which of course is something totally different again...

[Nacy Nightfire]

Actually it would be nice if someone could complile a reference list of terms as they vary across the various popular programs like Blender, 3d Max, Maya, Zbrush, etc.  Even the basic components: faces/polys, verts/points etc. differ from program to program.  I would volunteer for this project, but sadly I have no access to Maya or 3d Max and my overall experience in 3d modeling is limited to what I learned for SL purposes.  Maybe it could be a group effort and someone can start with the terms they think are most "changeable" and a beginner should grasp.  The most scary and off-putting aspect for a raw beginner is the terminology.  When I had to start thinking of UV's I got a bad case of virtual hives.  I thought I might push a button blow up my computer.  Ahhh the early days.

[Tarius Auxarmes]

Ya know, it might be nice if this thread were to get stickied because it contains a ton of information. Or perhaps someone condense the info into another thread. I know there is the wiki, but alot of people still come by the forums to ask things.

[Miyo Darcy]

---

York Jessop wrote:

I noticed your railings are probably taking up the majority of your poly count, it is possible to create a cylinderical appearance with a radius of 6 segments, just use soft edges. You could probably drop the poly count by 60% by doing that whilst keeping an acceptable visual look, Also I would consider squaring off the ends of your steps, its not ideal but again you could shave off another 2/3rd from the steps poly count.

---

After reading again through this thread as I found it interesting, I have again to quote you York.

I want exspecialy point in the quote to "creating a cylinderical appearance with a radius of 6 segments, just use soft edges".

 

I did try to find methods in Cinema 4d to make 6sided cylinder visual and round appearing. I did not find anything about "Soft Edges". I assume that word comes from blender as well? I now try to find out, what possibilitys I have with cinema 4d.

Also after reading google I did not find soft edges related to cinema 4d.

Can anyone one explain soft edges so that I might get some hints and what I need to find in my 3d application? Maybe if I understand how soft edges works, I can better ask google related to my 3d app and how this technique is calles in cinema 4d. Defenetly nothing in the help files about soft edges.

 EDIT:

I mean I do understand the word soft edges. But not how it could be done in c4d.

I also ask my self why soft edges would not increase polys? The word soft edges would tell me "better and round edges = more polys". So how is soft edges working in your application? Is it something like a fake-method or so?

 

 

 

 

[Drongle McMahon]

The "soft edges" are produced entirely by shading, not by adding geometry. The brightness of light from a point on the surface depends on the reflective properties of thye surface, its color, and the direction of the normal, perpendicular to the surface. At hard edges, the normal changes abruptly from that of one face to that of the other. This appears as a sharp line separating areas of different light intensity. When the edge is soft, the normal is varied evenly (interpolated) across each face, so that there is no sudden change along the edge. Thus there is a smooth gradient of shading around the edge instead of a sharp line. This gives the appearance of a rounded shape. This is all done by opengl and the graphics processor, so the interpolated normals don't have to be stored. A side effect of the way data is arranged in the internal SL format means that the smooth-edged version actually reduces the data used to store the shape. This means that smoothing this way reduces the download weight rather than increasing it.

The picture shows identical hexagonal cylinders with soft-edged (smooth, left) and hard-edged (solid, right) shading in Blender.

 

[guY Ralior]

In my workflow I start with the kinda blocky version to get the general shape that I am looking for then take the model into a sculpting program to increase quads and get the final super high rez model I am looking for (maybe 4 - 6 million quads). If for example I am working on a avatar while I am doing this it will change the basic shape of my model some I want to do this so that when I create my actual high_LOD mesh it follow as close to the contoures of the super high rez mesh from the sculpting program, this makes better AO and normal maps.

I will then take my model into a re-topologizing program and create my high_LOD mesh with as few quads as possible and still keeping the general shape of my model and having proper edge loops where it need bends. From here I will then create my UV map layout and then finally create my lower LODs by removing edge loops to end up with acceptable lower meshes and then finally a sprite for the lowest_LOD, and do same for UV map for each.

Using the super high rez mesh and the high_LOD mesh I will generate my maps. This work flow for me is good because I am not working on something that is not a finished piece for long. If for example I don't want to create the lower LODs and just use auto generated I can and I still have a usable model. If I did it the other way starting with the lowest_LOD and going up until I create the highest I don't have anything I can use if I wanted to, until I get to the highest_LOD.

As far as the smoothing, in your program it will have to do with setting and then smoothing your normal angles. Drongle has mad a great explanation of what is exactly happening when you smooth them. Hope this helps

[York Jessop]

As Drongle mentioned, its just a shading effect so it doesnt add any more polys or make any difference to your PE. I personally use Maya so im not sure about cinema 4D or if it has an equivalent.

Are you able to export .obj files from cinema 4D? If it doesnt have soft or hard edge tools, you could always import it into blender and do it from then, then reexport it as collada.

EDIT: Just reading about cinema 4D, I understand it uses an FBX exporter for collada? Hate to say this but you may experience alot of problems with that, Maya has one too and I had no end of problems with it, I eventually removed it and installed the OPENCollada plug in which works flawlessly.

[Drongle McMahon]

Here is a picture with more detail. The two shapes on the left use smooth (soft-edge) shading and the two on the right use sharp (hard-edge) shading. Each pair shows a low-triangle and a high-triangle version. They are shown in Blender at the top and inworld below. The middle row has the lighting and shadows setting on, with shiny set to medium, and the bottom row has lighting and shadows off, with shiny on low.

Notice that the smooth shading still has an angular outline in the low-triangle versions. Only increased triangles will improve that. Then look at the PE values. These are all download weights as the physics was kept simple.  The sharp shaded versions cost 2 to 3 times as much as the equivalent smooth shaded ones. This is because the vertices have to be stored four times each, for the four different normals used by the faces that meet at them. In the smooth shaded versions, the normals are the same for all the faces meeting at a vertex, so it only has to be stored once. (The UV mapping is essentially the same for all versions.) You are always better off using smooth shading except where the model requires sharp edges.

Whoops - should have said; these use default auto-LODs.

[Miyo Darcy]

---

York Jessop wrote:

As Drongle mentioned, its just a shading effect so it doesnt add any more polys or make any difference to your PE. I personally use Maya so im not sure about cinema 4D or if it has an equivalent.

Are you able to export .obj files from cinema 4D? If it doesnt have soft or hard edge tools, you could always import it into blender and do it from then, then reexport it as collada.

EDIT: Just reading about cinema 4D, I understand it uses an FBX exporter for collada? Hate to say this but you may experience alot of problems with that, Maya has one too and I had no end of problems with it, I eventually removed it and installed the OPENCollada plug in which works flawlessly.

---

First of all thank you Drongle that you greatly explained how it works. I think I understand that now.

At York... I think this works with any option in the texture settings in Cinema 4d then. I can setup shaders there. How its done with smooth edges in C4d I dont know but at least I have now tons of informations here so that I can discuss this in a german c4d forum. There are some professionals who propably can tell me how its done in Cinema 4d. Thanks for that to you all.

About the Cinema 4d collada exporter... No I think FBX exporter plugin is for 3ds Studio. in the manual I see Cinema 4d collada is from http://www.khronos.org/ . So it is not FBX.

Your right... if I dont find a solution in Cinema 4d with help from the C4d forums.. then I still could export to blender as a workaround. But the main problem is that I dont used Blender much. In fact I would have to learn the blender workflow then.

Im still sure that it can be done by shader settings in C4d, I just dont know exactly how. Atm. I read manual and help in the application to find that out. But I also will ask for that in a C4d User Forum. If I find out anything, I will drop this in the thread here then so that this maybe helps other SL and C4d Users as well.

All that is pretty much interesting and Im excited what those tools can do today

At Drongle.. is that ok for you if I use your smooth edge reference picture in a german C4d forum so that I can better explain and discuss what can be done with Blender and to get help how it can be done in C4d?

 

[Drongle McMahon]

"is that ok for you if I use your smooth edge reference picture in a german C4d forum...?"

Yes, it is ok.

[York Jessop]

Ahh I see, I must have read about an FBX plug in for cinema 4d then. I hope you find the tools you need in cinema 4D, I know what its like trying to use a 3D program your not used too, I had used maya a long time ago but decided to try blender as it had more SL support but I just couldnt get used to it so I went back to Maya.

Also didnt know about soft/hard edges making a difference to PE (albeit not very much) but the point was the same, soft/smoothing lets you use a simplified shape to get a similiar appearance.

[Miyo Darcy]

---

York Jessop wrote:

Ahh I see, I must have read about an FBX plug in for cinema 4d then. I hope you find the tools you need in cinema 4D, I know what its like trying to use a 3D program your not used too,

---

I did make a find via google on this topic now.

I think there is something in Cinema 4d which is similar to smooth edges in blender. In Cinema 4d it is a TAG which can be applys to any object via right clicking the object in the object browser and then go to "Cinema 4d Tags >> Phong".

I found it here descriped: http://www.c4dcafe.com/ipb/files/file/107-the-phong-tag/

For those who use Cinema 4d and read over this thread here.. you can also get help when you type "phong" into cinema 4d help. In german it is called "glätten".

I made a quik test with a 6 sided cylinder:

Without Cinema 4d Phong Tag

With Phong Tag on the object:

 

If I click the phon tag themselfe I can edit:

- angle-constraint [activate/deactivate]

- smooth to [VALUE in DEGREES]

- broken edge analyzing [activate/deactivate]

I think that must be the equvalent to the smooth edges in Blender

I will play a bit with that now.

 

 

[guY Ralior]

In C4D locate where you can show normals. This will put a small arrow or similar pointed in the direction the normals are at the moment. Most likely somewhere maybe in that same tab you should have set normal angle, smooth and harden normal angle. When you click smooth normals for the first time you should notice that hard edges now go away much the same way they did with the phong tag. SL uses Lambert not phong so I am not sure if the phong shading tag will show in SL, but setting and smoothing your normal angle will show in SL.

In Maya if you do a show smooth mesh preview by hitting the number 3 on the keyboard you are showing what the model would look like if it was subdivided, so smooth and normal angle are different things, and I would think that would be the same in C4D. In Blender smooth is showing you what your normal angles are and a subsurface modifier is subdividing the object with more geometry, so the two terms are not interchangeable between Blender and Maya and I would think it would be the same for C4D. Hope this helps

[Drongle McMahon]

"...so I am not sure if the phong shading tag will show in SL..."

I guess what matters is what the Collada exporter does when it sees the tag. Collada uses vertex normals, so as long as it uses the same normal for all triangles meeting at a vertex, it should work as smoothed in SL. I doesn't need to say which shader it uses.