(Art test) Watercooler
polycounter lvl 13
Hi all,
I've been working on an old Splash Damage art test (http://www.splashdamage.com/images/content/arttest_concept01.jpg).
I'm gonna use it for my game art course too.
Here's the specifications:
Model from concept art
* 1000 Triangles
* 512 X 512 Diffuse/Specular/Normal Maps
* Textured in a realistic weathered style
I've reach 989 tris, finally!
Textures are presently 1024 x 1024, but I'm gonna reduce it to keep the quality. Also, my specular map isn't done yet.
I would like to receive some feedback on the modeling and the texturing.
Thanks everyone!
I've been working on an old Splash Damage art test (http://www.splashdamage.com/images/content/arttest_concept01.jpg).
I'm gonna use it for my game art course too.
Here's the specifications:
Model from concept art
* 1000 Triangles
* 512 X 512 Diffuse/Specular/Normal Maps
* Textured in a realistic weathered style
I've reach 989 tris, finally!
Textures are presently 1024 x 1024, but I'm gonna reduce it to keep the quality. Also, my specular map isn't done yet.
I would like to receive some feedback on the modeling and the texturing.
Thanks everyone!
Replies
Oh no, alert the wavyness police!!! This actually looks fine to me here, I would add a few more sides cylindrically(16 instead of 12) before messing about with cage tweaks, as that would improve the mesh as well as the wavyness.
The mesh itself is really messy, you need to look into merging all(or most) of these chunks together that you can, right now you just have a lot of tris and UV space totally covered up, and its very wasteful. You're not saving any performance by needing to render a bunch of stuff that is unseen, just to save some tris by doing proper intersections.
If merging everything together makes the triangle count too high, you need to look at simplifying the design, move details in the high res so they can be more easily represented in the low res. High and lowpoly modeling aren't things that happen without regard to each other, you've gotta make sure your model is going to work well in the high, low and uvs.
The half-cylinder cup holders(?) on the side and water dispensers are a lot more dense than the rest of the mesh and have some unneeded edge loops. Optimize some of them to allow for a more rounded main cylinder of the mesh.
But one thing really bothers me. I qouldn't have guessed that this is supposed to be a water cooler for getting a glass of water. when I saw your model I was thinking of some kind of industrial water tank thats used for cooling down water during production or what ever. Only after I checked the concept I knew what it was supposed to be. Anyway maybe this impression came from the watertank that is not transparent. for me its blue metal. So I would definately work on that part.
Besides this the asset itself is ok I think. Your materials need more work. You have large areas that are really bland and boring. You don't need to cover everything with scratches from hell and stuff, but some subtle surface detail on the larger areas will improve the asset. Also you should think of adding a cubemap to your material. Although looking at the given specs this might be not allowed. (I would do it anyway)
Looking at the style of the concept it looks Fallout-ish for me. It gives me this 50's 60's scifi impression. Style wise your model is more like quake4 or doom. So I think if this was an actual art test this would be a serious problem. It maybe ok to give the asset some personal note, changing the concept here and there. But the style you chose is just taking away and not adding in my opinion. By changing the style you're asset is missing all of those funky beveled and curved panels, little dents and those details were making the concept so special. They made difference between the average watercooler that nobody will notice in a hallway and the well designed watercooler that some players might notice and have a closer look, just because its so cool.
So I know the things I've pointed out are not so helpfull because adressing those issues would require major overhaul of the highpoly and stuff. But in my opinion, if someone were given this neat concept, I wouldn't accept those obvious differences in style, detail and perceptibility.
A few very important points to hammer home here:
1. Do not model the surface detail in your low, let your normals handle that. You could add a few more loops to get some extra detail on some of the bevels, but the way you have everything modeled in "panel" type shapes is inefficient, messy, and doesn't really suit the concept. The concept here is excellent in that it has nice clear main shapes, and then some surface detail thrown on top, making it very suitable to use normals to grab a lot of this detail.
2. Consistency in detailing, on your mesh the big main shape is noticeable jagged and low poly looking, but these two little things sticking out on the sides of the mesh have the same roundness(!!!) ie: started from a 12 sided cylinder. And then seemingly in a random fashion, more sides are cut into the mesh as it goes down.
My paintover is based on keeping 16 sides for most of the mesh, and using a little more on the bottom where it is wider. Keeping your mesh density consistent goes a long way towards making an asset that isn't noticeably "lowpoly" looking. Not only consistent, but you should always strive to have enough roundness to match the look of the highpoly, when viewed from a reasonable distance. Extra sides here(like mentioned in my previous post) will also help reduce and wavyness in the bake.
3. Merge, merge, MERGE! I know it will read like a broken record at this point, but you have wayyyy too many intersection chunks, resulting in a lot of useless geometry and wasted uv space. Keeping a nice solid mesh ensures very little waste uv space because of occluded areas, but also much better unwraps that give you more detail, and make the asset easier to texture. Modeled like my paintover, you should only have a few unique uv islands. The airtight aspect of this mesh will also ensure a much more painless bake, as you dont have to worry about any intersecting.
4. Model in a way that suits a low resolution model. This can not be overstated, do not over complicate your model by adding/changing details in the concept, like the "ear" bits hanging off the top/size of the mesh, keep this simple and solid(closed off) to reduce complexity and cause less headaches. Also, model your highpoly in such a way that most surfaces end up flush with thier neighboring surface, and do your detail with the bevels, dont "stack" all your pieces and make life harder with your lowpoly.
5. In addition to all the technical issues, by seeing these side to side we notice a lot of aesthetic issues as well, shapes that are totally different in the mesh than the concept. When working from a concept it is always very import, if you're making changes to it it better:
A. Be for a good technical reason, ie: reducing needless complexity in the mesh(there isn't a good example here with this concept, as the level of detail is very suitable to a real world asset)
B. Make it better, and more interesting
Never change details in a good concept, if the end result is the asset looks worse, there are a lot of areas here where this asset falls under that category, where proportions, shapes and details are completely different, but not in a good way.
if i could thumb up posts i would give you two
[]
[]
,
\ \----]
----/
[]
[]
,
\ \----]
----/
This is looking good so far, and can't wait to see the final product
Thanks EQ
megalm2000, while your model looks ok from an outside point of view, without knowing what it is or what its meant to be, it should be reading like a water cooler.
if I hadnt noticed the little spouts to show its scale, I would assume it is a canteen or plasma container or such. You did a lot of right things with the texture, but with the shapes messed up like Earthquake mentioned, and with the textures not showing the scale or function, you lose the entire model.
I hope you redo this, you will learn alot.
GL!
I realize that I am getting long-winded and not really narrowing down on a specific crit, but overall what I am saying is make this prop identify with your style, and also really consider whether or not you are making something that the player will just walk on by. keep puttin in work and good luck!