That's cool about the texturing just wondered if you had used an alternative method.

Your scaling texture issue is abit tricky. All depends on the mesh and UV island %area.  I'll have a think about it - if I find anything I'll get back to ya, sorry.


Some work Ive been doing.

I've spent a little time working on trying to optimize my meshes for in game. Although I tend to just make the one mesh and make simplifications as I'm going. But to take that a bit further I had a look at Meshlab to reduce the number of polygons and I've had a bit of success after running the filter > quadratic edge decimation.  Just testing out the number of target faces against the amount of shape deformation.

Some results:

The Rosetta Stone

Left (un-optimised)  Right (Optimised)


Tris

Before  vs   After

296             148   = 50% reduction

A cog

Left (un-optimised)  Right (Optimised)

Tris

Before    vs       After

559                  400   = 28.4% reduction

Please not the texturing is very rough on this one - its not the final one.  Just felt I needed to show some work /results.

Sorry for the late response my machine need some maintenance.

@Bli!tz

Thx for the encouragement Bl!tz.  With responses like that I may have to start taking requests.

@LPS

Cheers champ. I'm encouraged you like it.

@Odedge

I think I know a possible "rough" way of getting the scaling more consistent, but I have to do some tests first.

Yeah does seem a bit tricky. Kinda need a numerical UV island to mesh density value to use as a factor to scale your texture.  The only other thing is to bring up the properties panel in the UV image editor and check the stretching. Other than eye balling your islands that's all I've can think of atm.  I wonder how the 3ds max crowd approach this issue.

I have heard of Meshlab, but since I haven't done much serious modeling before, I never bothered to look at it.  It would be nice if it could reduce the polygons.  My biggest issue is adding loop cuts and having faces that don't really need it, become more complex for no good reason.

Meshlab will take the strain out of using loop cuts. So I just use it for the Filter > Re-meshing simplification and reconstruction > Quadratic edge collapse decimation.  Then just do some trial and error with the target number of faces untill I find a good compromise. Reloading in the mesh between trials as I could'nt find the undo button.

I can't tell a difference between the Rosetta stones, so the very poly's would help.

Yeah the Rosetta stone was made using lots of loop cuts so there was alot of wasted faces on that mesh.

The cogs are a bit different, but it's mainly the lighting.  It might just be where the light is placed or the fewer vertices's (assuming you are using vertex lighting).  I am sure with a model like this, you can add a second UV map for the light map.

Nar I don't think its the lighting so much as those ones are light mapped form memory.  Having a second look at it I may have chosen a too rough a compromise.  Yeah I always create a second UV map for all my meshes for this reason.

Either way, you are stretching your modeling skills more than me right now as I am building fairly "basic" stuff.

Hopeful I'm not missing any of the necessary basic skills and end up climbing into a culdesac.

Can you post a wire frame picture of the Rosetta stone?  Even if it's just from Blender?  Have you learned anything about rendering and such?  I just helped a friend figure out some stuff, so I may start using that to create some preview pictures.

296 Faces                                                             148Faces

So the first one (296 faces) shows the top faces are equal too the bottom faces so the mesh is more complex.  The second one (148 faces) looks more complex but your seeing both the top and bottom faces at once.

On the rendering front nothing too much but hears a quick memory dump. 

-Getting to the camera view hit 0

-For composition move in the camera view by using Shift F and wasd to control it along with the scroll wheel or just reposition the cursor and the camera will track towards it.

Alt

-Find a angle you like and hit ctrl alt 0 to position the camera to where your facing.

Rendering speed

-Under world properties > Gather > choose Approximate over Ray Tracing to get a faster render of your scene when working out lighting / shadowing settings.

These are the things that seemed highlight from my awareness. Hope that help a little.  I'm only really scratching the surface at the moment so still heaps to learn.

Thanks champ

Lord_PorkSword wrote:

---

Dunno if these are useful at all for yaz, but there's some Blender tutorials here!

---

 It took me a second or so to find them as most of the page focuses around the UDK. 

While I personally appreceiate the thought, these are very basic tutorials (which can be found on the net) and they use 2.49.

There is actually more tutorials on the net than I care to watch/read!  For me, I just have to go through enough of them and learn as I go.

Time to resurect this thread and my attempts at learning Blender...

Got some inspiration recently to make a pillar after seeing this:

So I thought I'd have a go. 

This really a WIP >

Top view showing off some of the BreakingThrough level

Full profile view

Stats:

800 Tris

648 Verts

I know not very close on the texture side, but I'm still learning Gimp.

Still to do:

correct UV alingment on top face

re-layout grove faces for better texture support. (get rid of scaling issues).

That's looking very solid there Achernar. It's nice to see some more rounded objects being made as they can't be done with BSP (very well). How final is the lighting in your map as this should effect the look for the better.

I have been in a mapping slump for a bit, which typically happens once a year. I am getting refocused now and watched 2 tutorials that are very inspiring and instructive.

Create a Grungy Brick Wall

Create a Sci-Fi Panel

You should check them out.

Keep up the good work.

I gave a shot at the Sci-Fi Panel tutorial I just linked.  Here is my normal map.  The bolts need some work, but you get the idea!

I am still using 2.57b and for some reason I had to place my low poly mesh above the high poly meshes. 

I sort of made a cube of it in the editor and applied the "color" map the tutorial suggests to make, just to give it some contrast.  After part 2, I will make an effort to do some honest work in GIMP to make it look better.

Achernar wrote:

---

Only thing I'd point out is the 'relief' or grove on the handles - as the grating touches the grove. Seems tooclose - the grove is thinner in some places.

---

 This is what I sort of meant by the bolt work needs some work.  I will fix when I do the textures.  I was having issue with the bolts showing up correctly and getting the whole thing to work, so I sort of rushed it.

I gave a semi-honest work at creating a decent looking texture. I am a bit rusty with GIMP, so be kind. Here is the the same sci-fi panel with a diffuse and emessive texture applies.

I know, I know... I didn't fix the bolts by the lights. 

Achernar wrote:

---

I like the blended scratches on the grating and I see you managed to transition in some variegation on the green metal, subtle but nice. The radiation symbol just sold it for me. Great work!

Yikes I'd better get my imagination fired up to use blender. Anyone wanna challenge me to make something - famous last words...

---

 If you want to do a friendly "competition", just let me know.  I can always use more motivation to practice.  We chose an object or such to make and we do our best to make it.  We would need some room for creativitiy though so we aren't really making the exact same thing.

This is why I like this type of tutorial (sci-fi panel).  It shows you what to do, but you can always do your own version of it and learn as well.

mAlkAv wrote:

---

I like both, texturing and modeling but I have problems with uv unwrapping, espacially with big and complex meshes like lanscapes for example.

---

 A  quick (and Dirty) Way to UV-Map Landscapes would be an averaged planar map.

This is good for prototyping and even final mapping.

Also this allows to use tileable textures instead of a unique skin

HERE is a link to a step by step tutorial ;)

Achernar wrote:

---

@Bl!tz to a UT4 but I think it will take me a decade with out substantial help ...  Epic

Blendercookie just shown a light on a little add-on that generates castle walls - so versatile!

Addon: add_mesh_masonry.py

Now just have to figure out how to texture this thing...

---

 I watched that and it looked very interesting.  There might be a problem with it as it may created way too many faces that wouldn't never get seen.

It didn't seem like it created the "grout" between the walls, but I may have missed that part.  If it doesn't, that will be a problem.

You could possible use this to make a high ploy mesh and bake the normals to a texture, then just apply it to smooth wall, which would be easy to make.

I have found a new ability in Blender that might help us out. Remember this is a "proof of concept" and I don't know how useful it can be.

Blender has the ability to bake stuff onto a texture. You can use this for creating normal maps and such. You can also do this for textures, which I have tested a little bit.

In my current Temple map, I am using stock Epic materials. On 1 mesh, I am using 4 different "base" textures, which makes the static mesh have 4 material slots. This isn't the ideal way of doing things as it's more efficient to use 1 texture.

The problem is I can't "legally" create 1 texture from Epic's, even though I have the actual texture on my hard drive (I just use these for aligning the UV maps). If you are creating your own textures and are reusing some of them (walls, floors, trims, etc), it would be tedious to re size them for each static mesh that would use them.

I thought it would be great if you could create a mesh using multiple materials, UV map it so it looks correct. Then you could duplicate the mesh, redo the UV maps (using Smart Project method), then bake the textures from the original multi-textured mesh to the new single-texture mesh.

Technically, you can!

The left mesh has 3 textures applied to it.  The one on the left has 1 texture (see below) applied to it.

The baked textuers are all in the correct place and orientation.

I tested this on a more complex mesh (with angled surfaces) and it had some texture bleeding.  So that is one issue to be concerned about.

The other issue is the quality of the textures.  Most of the textures in the picture above started out as 1024x1024 or 1024x512.  On the original mesh, they will look great. 

But when you bake them, the image has to be resized to a smaller resolution, which will effect the quality.  The other issue is that if you are using the same texture over and over again, there is no point and having it twice on your texture.

Blender and Unreal allows normal UV maps to overlap and in the case of my example, it would better to have only 3 samples of the original images and not 6.

Time will tell if this will actually be useful, but if you have any thoughts on the matter, pleas share.

Here is a copy of blend file with the textures packed, if you want to check it out.

Achernar wrote:

---

I haven't tried baking onto a single texture like you describe. However if the UV islands are the same shape and use the same texture you can place them on top of one another - saving UV space. The texture will be projected onto the both faces / UV islands. I'll have to try this to see if it works in practice.

I wished it work that way, but as far as the technique I am using, it doesn't.  Each faces that shares the same texture is sharing the same UV coordinates.

The way the process works is it basically takes an "picture" of what it sees on one mesh and copies it to the other mesh.  The bigger the image you are copying to, the lbetter the texture should look, but it seems 1024x1024 is the biggest size you should use for most meshes.

Do you have any control over where the UV islands are placed - that might get around those bleeding issues you were telling.

You have to redo the UV islands for the second mesh, so they don't overlap to begin with.  I used the Smart Project, which usually doesn't overlap UV islands.

You could resize some of the islands that are going to be seen by the player more, so the quality is better?

The bleeding wasn't an issue with UV placement though as there was enough space between the islands.

I'll have to make some test and get back to you.

---

 Another quick tip of the day regarding UV mapping (which is still the hardest part in my opinion).

You can select an edge, press the "W" key, and select "Align Augto", "Align X", or "Algin Y" and it will "snap it to the axis.  Align Auto works well in most cases.

I tried selecting all of the edges and selecting "Align Auto", but it just snaped them all to the X axis and destroyed them all.  I wish there was a way to quickly snap them all to the nearest axis.

Also, if you are aligning an edge that there is actually 2 there (the faces next to each other don't share an edge), you should be able to press the "Alt" button while selecting them and it will automatically select both.

Also also, MCampagnini says he should be able to update the exporter so it works with 2.59 correctly.  He says he has plans for version 2.0, which he might get done this weekend.

I don't think he will be adding any new features (don't think it's possible with an ASE file), but I think he will finalize how he wants Smooth Groups to be assigned, if I remember correctly.

Yes, I am on a little Blender kick at the moment!

Achernar wrote:

---

Arh thanks for your description of the baking process I think I get where it coming form. So it seems you'll still have to work around that bleeding issue.

The funny thing is it didn't happen in other similar areas (same angle and such).  I will test a bit mroe to see what kind of results I get.

About aligning UV islands I think one can also select a row use Scale 'S' Axis eg 'Y' '0' This will straighten them along one edge. Thx for your little tip that looks like it will come in handy.

This is correct, but the way I mentioned is a bit faster.  I would like to really understand how to control UVs more.

I saw some preview videos of the changes they are making in the Google SOC and it looks promising, but I don't know when it will be added.

Wow that's great new about the exporter - Mike is legend! I think exporter is pretty good for my purposes but extra work on smoothing groups sounds very promising.

Thx for sharing your progress Odedge!

---

MCampignini just released ASE Export v1.3: http://campagnini.net/2011/04/16/blender-to-udk-ase-export/