UV Map Workflow Questions

[vasf_jeffs 0]

Guy's I've searched quite a bit, done the tutorials and even gone back to the ofp stuff. Mondkalb's tutorial was great and allowed me to get a simple object into the game with the proper texture, maps, etc. What I'm running into is an utter lack of understanding how to get UVMaps put together for more "complex" objects. So after 5 hours of trial and error (and significant forum searches) I'm throwing myself at the mercy of you the experts.

I"m working on a billboard (don't laugh it is complex enough for me). I have it broken up into the following named sections in lod 0: Support, Advertising, Pole, Catwalk and Ladder. I asked someone earlier on Skype about just uv mapping on rung of the ladder and his reply (yes do that) got me thinking.

Should I be creating a separate UV map as i create each component in a model? Mondkalb's tutorial has you select the whole model first, then do the UV. When I attempt to unwrap more complex objects it doesn't seem to be the right answer as all of the faces aren't being shown. Alternatively can this be done as each object is made (i.e. ladder rung, ladder support, main support, etc)?? If I UV map an object and move it later is the map tied to the component independent of the location?

Is it easier to do the UV maps in 3ds? Any advice on your workflow for uvmapping would be sincerely appreciated.

Thanks,

JeffS

[max power 21]

It would be 1000 times easier to do it in 3ds max once you get used to the tools imo.

[STALKERGB 6]

I Personally create the whole model before I UV anything, although I personally don't have a good reason for doing it like this, I guess you could say you know how much you'll have to cram into your final UV if the whole model is there beforehand.

Although you can UV stuff in Oxygen, it is sooooo much easier to do it in an external modelling program. I can't speak for 3ds max in particular as I use modo but what would have taken days in O2 takes hours in modo when it coes to UV mapping. For the most part, all I'm doing is selecting edges where I want the "seams" to be and modo does the rest (obviously it needs tweaking here and there but still!)

If you are creating the UV in O2, you can always hide objects that are in the way (Ctrl + H, to unhide I select all Ctrl + A then press Shift + Ctrl + H).

If I UV map an object and move it later is the map tied to the component independent of the location?

The UV map shouldn't change regardless of how you actually edit the model itself. So say you UV'd a box, and then stretched and moved the box about, the UV map will stay the same :)

Edited July 17, 2012 by STALKERGB

[soul_assassin 1723]

There are also tools which are made specifically for this:

http://www.uvlayout.com/ for example. Avoid O2 UV tools like wildfire. Maybe only to fix something that you notice waaay after import. As to do it in the end or in progress, both have their own quirks. Sometimes I do UV while I havnt finished the whole model just to avoid being bored with just UV once its done. This way once the model is done so is the UV. Either way it will not affect the final result.

\

[b00ce 160]

For those of us who don't have the wads of cash to throw at modeling programs, here's the easiest way I know how to do it in O2, though it doesn't really work for character meshes or spheres.

First off, you need to create two animation keyframes. This allows you to fillet your model, without destroying it.

Select the second of the two and start separating your model into manageable chunks via "split topology", you should bind it to a key.

Once you have the model in its pieces, you need to flatten it. Use the pin to rotate the faces so they're as flat as possible, though it doesn't have to be perfect.

For more squarish objects, split the faces off and move them away from the main body, rotate and then position them so they "make sense" from a texturing standpoint to you.

For cylinders, split it in half and use SelectObject to select the half you want for rotation.

Once you have the model splayed out, make sure they're all flat against one view port (It also helps to organize them as well), select them all (From that view port) and go into the UV editor.

Once there, hit the planer mapping button (In the middle of the row of buttons, Green box with an arrow pointing to a red mesh.), this will put the faces you selected down.

(If you have a view port that has a side view of the faces active when you "paste" the faces down, it won't work right.)

Make sure all of the faces are inside one of the boxes; once you're happy with it hit file -> Export and save it in your model folder in the P: drive, for the sake of keeping things together, it doesn't matter where... yet.

Next, open the .emf file in MS paint and save it as a .png so you can open it with the texturing program of your choice. (For some reason, Paint.net doesn't recognize EMF or WMF formats)

Once in Paint.net, Gimp, PS or whatever program you use, re-size it (The canvas, not the texture) so it is the nearest base of two (1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048), though it does not need to be square. 2048x32 works just fine, though it is a little daft to make a texture like that. :p

Make the UV map mesh the base layer and doodle on it a little (Just enough to make sure its working) Save it as whatever default save file the program has, to save the layers, and then as .TGA (Do NOT use RLE compression, it won't work.)

Once you got that saved, move it to the proper folder. (I use "P:\my_model\materials\texture_co.tga", and have my models in a models folder because I'm used to it from G-mod)

Once everything is where it should be, go back to the UV editor, select everything and copy it. This is important. You'll lose everything if you don't.

Hit Filter -> browse texture and browse for your .TGA file. It'll delete all of your UV data, don't worry, just paste it down again. ;)

Once this is done, go back into O2 and start Buldozer, it'll have to convert the .tga files but it'll work. Alternatively, you can hit "Use Direct 3D" and then "Display Faces Filled" and it'll put the .tga texture on there without the need to convert.

Once you're done and happy with the UV map, go back to keyframe 0 then delete both of them (The keyframes, not the model) and then highlight your whole model and use merge near to stitch the points back up. You may need to do some manual work as well.

For the ladder rungs, I would map 3 or 4 and mix them around going up, so they don't appear to repeat. ;)

Good luck.

Special thanks to Rock who showed this to me a while ago. :)

[PuFu 4591]

For those of us who don't have the wads of cash to throw at modeling programs, here's the easiest way I know how to do it in O2, though it doesn't really work for character meshes or spheres.

blender is free, so is max, maya, softimage (as long as the output is commercial is free).

Other than that good post

[b00ce 160]

blender is free, so is max, maya, softimage (as long as the output is commercial is free).

Other than that good post

For students, yes? Not all of us are students.

I've tried blender and couldn't wrap my head around it, no matter what tutorial I followed, and I have a hard time working with any 3D program without 4 views. I know you can get it in blender, but that's a pain too... Blender is not new user friendly. :p

Edited July 17, 2012 by b00ce

[PuFu 4591]

For students, yes? Not all of us are students.

no. it is available for anyone for some time now. it is still available via students sub-site, but it is edu version. you don't need to testify you are a student anymore, you can even choose unemployment in this area.

I've tried blender and couldn't wrap my head around it, no matter what tutorial I followed, and I have a hard time working with any 3D program without 4 views. I know you can get it in blender, but that's a pain too... Blender is not new user friendly. :p

there is no such thing as user friendly 3d software. modo is closest here, but ten again, the more capable the software, he harder to make it noob friendly.

[Messiah 2]

Booce, no need to manually unwrap cylinders with that technique, O2's unwrap tool will do it perfectly well (Surfaces -> Unwrap Structure). It's a perfectly viable way of UV unwrapping, long winded, but viable. I believe Earl (BISim) pioneered the technique, and taught it to Rock and Myself.

Also, I find i easiest to layout a single sided face in the viewport you want to flatten everything to, so you can a) use it as a template to layout the unwrapped mesh and b) to use 'background from face' to create a texture box, and then load a chequer pattern texture to apply to the unwrapped model (obviously at this stag you can just delete the square guide). In that way you avoid a chunk of the faff with the uv editor, and just have to export the EMF. You can export the EMF at set ratios and make sure to tick all three boxes, which will ensure the EMF mesh stays in ratio by adding a bounding box.

[b00ce 160]

Oh, sweet, downloading Max now. :D

I'm definitely going to give it a shot.

Booce, no need to manually unwrap cylinders with that technique, O2's unwrap tool will do it perfectly well (Surfaces -> Unwrap Structure). It's a perfectly viable way of UV unwrapping, long winded, but viable. I believe Earl (BISim) pioneered the technique, and taught it to Rock and Myself.

Also, I find i easiest to layout a single sided face in the viewport you want to flatten everything to, so you can a) use it as a template to layout the unwrapped mesh and b) to use 'background from face' to create a texture box, and then load a chequer pattern texture to apply to the unwrapped model (obviously at this stag you can just delete the square guide). In that way you avoid a chunk of the faff with the uv editor, and just have to export the EMF. You can export the EMF at set ratios and make sure to tick all three boxes, which will ensure the EMF mesh stays in ratio by adding a bounding box.

When ever I tried using the Unwrap structure tool, it gets ugly...

Once the checker is applied to the mesh, how does that translate into the UV map/texture? By that I mean how does it look when you're actually painting it? Do you still have the wire frame?

And I'll have to remember to do that with the EMF though, it'll save a little time.

Edited July 19, 2012 by b00ce

[Messiah 2]

The unwrap tool is fairly useless for anything but cylinders, give it a try, it splits the caps, rotates them flat at either end of the cylinder, and then unwraps the cylinder itself. Huge time saver. Sometimes it can be a bit pedantic, so as a rule I split the end caps myself, and then let the unwrap function flatten the barrel, which is the most time consuming part.

As soon as you apply a texture (in this case a chequer map) to the flattened surfaces, you're giving the faces a UV map. From there as soon as you open the UV editor you'll see everything laid out as in the viewport. Exporting the EMF gives you the wireframe as per what you mapped in the viewport, and if you were to replace the chequer pattern with the EMF output from the uveditor, you'd havee a perfectly aligned wireframe texture. Has never failed me yet.

Saying all that, I can't seem to get my UVeditor to open in O2 today... odd.

Edited July 19, 2012 by Messiah

[b00ce 160]

Duh! the EMF does do that huh... FPDR

Does the UV editor show up on the task bar?

[Messiah 2]

yeah, was just a case of it being minimized off screen, all solved now.

[jagheterjan 10]

For how much padding do you set up your UV maps? I'm currently working with 2 texels for every 1024 texels of file dimension where needed and try to stack and mirror UV islands whereever possible, but having that much padding makes the finished UV map rather ineffective imo.

Mods, please give me a good slapping if the question's better suited for a separate thread.

[max power 21]

I think it's fine here.

When you're outputting UVmaps from o2, IIRC you want no padding.

[Messiah 2]

For someone who just learnt by doing, explain padding? At first glance I read it as space between the various UV elements, but the modding world is full of non-sensical titles. If we're talking about leaving space around UV elements, especially ones that will have different diffuse colours, to avoid 'bleeding' across UVs, then that sort of depends on the texture panel size in my experience - I leave a certain amount of space as a matter of good course (well, where objects aren't meant to be connected of course), but that's personal.

Edited July 23, 2012 by Messiah

[jagheterjan 10]

Edge padding is applied to an UV to avoid bleeding between UV islands when the engine mipmaps your textures. In a nutshell - if you have one UV shell containing only black texels and one containing only white texels, and they are directly bordering each other with no padding applied, the colours will bleed into each other along the border, creating a grey seam there because of aliasing artifacts introduced by the renderer. These artifacts get more pronounced the lower the resolution of the texture gets because as texture resolution decreases, texels in the texture 'grow' and cover more screen space. Same applies to UV borders that are not perfectly perpendicular to the texel borders. To reduce/avoid the aliasing artifacts and colour bleed, UV islands need to be moved away from each other in order to have the bleeding happen where there is no 'useful' texture information stored, as far away as possible from any UV border.

I came here to ask the more seasoned content creators for their experiences as I'm still fairly new to the whole low-poly modelling thing and want to do it right the first time.

[Messiah 2]

Well I'm about as seasoned in ArmA content creating as anyone else around here, I'm just a bit poor on the terminology as I just dived into O2 and didn't really care what I was doing right or wrong.

Bleeding and horrible mipmapping was certainly an issue before when we were using lower res textures and previous versions of the game (ironically, in OFP, green hues seemed to have the greatest issues), but I've not seen any mipmapping issues in the later ArmA and UVs, and thats down to 2048x2048 texture resolutions. That being said, and this may of course be a personal quirk, I still pad my UV's to some extent, but I'm certainly not consistant in doing so (other than choosing one main hue for the entire panel), and I've never seen any bleeding around differently coloured parts of the UV. Example of the jackal below, all O2 uv unwrapped for good meassure (Still a fairly rough UV back then, have become a little better in my later years):

http://downloads.rkslstudios.info/RKSL/Messiah/Ref/jackal_example.jpg

Edited July 24, 2012 by Messiah

[jagheterjan 10]

Cheers Theo & Messiah, that's what I was hoping for :)

[Messiah 2]

Anytime Jan. Like we said on the RKSL forums, I think we're all rather excited about seeing your Eagle come to fruition :)

[Meatball0311 79]

First off, you need to create two animation keyframes. This allows you to fillet your model, without destroying it.

I create two keyframes 0.000000 and 0.100000. I then use keyframe 0.100000 to copy and paste the model into pieces for UV Mapping. When I am done, I delete keyframe 0.100000 and 0.000000 and hit ctrl-A to select the model. I then merge near to the 0.0001 level. Everything works fine, but the problem that I am having is now the face count has doubled. How do I remove or merge the extra faces? or does it not matter?

What is the difference between exporting into .emf or .wmf formats?

Make sure all of the faces are inside one of the boxes; once you're happy with it hit file -> Export and save it in your model folder in the P: drive, for the sake of keeping things together, it doesn't matter where... yet.

Next, open the .emf file in MS paint and save it as a .png so you can open it with the texturing program of your choice. (For some reason, Paint.net doesn't recognize EMF or WMF formats)

[b00ce 160]

You're not supposed to copy the faces, you're supposed to use the "Split" function. Basically this splits the selected points/faces from the overall object. You need to bind a custom key for it.

As for the difference between EMF and WMF, I have no clue.

[Meatball0311 79]

You're not supposed to copy the faces, you're supposed to use the "Split" function. Basically this splits the selected points/faces from the overall object. You need to bind a custom key for it.

As for the difference between EMF and WMF, I have no clue.

Thanks for the clarity and I will definitely try that out.

---------- Post added at 11:47 AM ---------- Previous post was at 10:09 AM ----------

Well I learned something new that would help me out, but it still doesn't solve the issue that I am having. It seems that when you use this method (Split Topology), it leaves creases or lines in the sections that have been split, and those lines show up in the textured model. It is just annoying having those lines and the other method that I was using made it look like one whole texture with no lines. However, it increases the amount of faces and aren't faces polygons?

So, my question is are faces polygons? and if so how do I cut them down?

According to the O2 Manual:

Polygon – planar surface with any number of edges

Face – in O2 faces can be triangles or quads

[PuFu 4591]

So, my question is are faces polygons? and if so how do I cut them down?

According to the O2 Manual:

Polygon – planar surface with any number of edges

Face – in O2 faces can be triangles or quads

While i will not comment on O2 workflow and tools, i will put my technical 2 cents:

1. faces are polygons. No matter what are the number of edges that define those.

In general, in the modelling area, faces are called:

a. quads - 4 edged polygon

b. tris - 3 edged polygon

c. ngons - N faced polygon, where N > 4

more on the subject:

DX only uses triangles. Reason is, it is easier to calculate and thus handle. All quads are made of 2 triangles. Even if you want to keep faces a quads, the hidden edge dividing the quad into tris is still gonna be taken into account (hence for most RT engines - games included - it doesn't matter the number of quads, but the number of tris).

Some advise to optimize your model prior to game usage into triagles. I personally would say you can keep quads as long as you take care of the way the tris making those quads are setup. (most 3d packages are allowing you to set the way the triangles are set within the the quads).

2. Quads are not planar surfaces. Or not mandatory.

Example (click for bigger):

The above shows the same duplicated quad (same vertices etc).

As you can see top left corner, there are 2 polygons, and only 8 edges, and not 10 as it should be.

That said, you can see there are 4 tris. The only difference is how the dividing edge is set up (the dotted edge in the above image, shown here for this very purpose - usually not).

I hope the above clears some things, and some O2 user can help you out with the tools available for modelling there.

Edited August 21, 2012 by PuFu

[Meatball0311 79]

While i will not comment on O2 workflow and tools, i will put my technical 2 cents:

1. faces are polygons. No matter what are the number of edges that define those.

In general, in the modelling area, faces are called:

a. quads - 4 edged polygon

b. tris - 3 edged polygon

c. ngons - N faced polygon, where N > 4

more on the subject:

DX only uses triangles. Reason is, it is easier to calculate and thus handle. All quads are made of 2 triangles. Even if you want to keep faces a quads, the hidden edge dividing the quad into tris is still gonna be taken into account (hence for most RT engines - games included - it doesn't matter the number of quads, but the number of tris).

Some advise to optimize your model prior to game usage into triagles. I personally would say you can keep quads as long as you take care of the way the tris making those quads are setup. (most 3d packages are allowing you to set the way the triangles are set within the the quads).

2. Quads are not planar surfaces. Or not mandatory.

Example (click for bigger):

The above shows the same duplicated quad (same vertices etc).

As you can see top left corner, there are 2 polygons, and only 8 edges, and not 10 as it should be.

That said, you can see there are 4 tris. The only difference is how the dividing edge is set up (the dotted edge in the above image, shown here for this very purpose - usually not).

I hope the above clears some things, and some O2 user can help you out with the tools available for modelling there.

1. faces are polygons. No matter what are the number of edges that define those.

In general, in the modelling area, faces are called:

a. quads - 4 edged polygon

b. tris - 3 edged polygon

c. ngons - N faced polygon, where N > 4

Thanks for explaining that I have a better understanding now. It still leaves me with the question what is the safe area of how many faces/polygons can I have in the model? Is there a limit I need to stay under?

Some advise to optimize your model prior to game usage into triagles.

Are you suggesting that I should "Triangulate" the model to improve performance issues?

PuFu thanks for your time!