WIPs and crap. learnin and stuff

[orange13]

Trying to get started on basics etc of game prop modeling. I have only really sculpted in zbrush and used 3d for 2d crap...so time to get things working.

decided on a victorian style closet/desk thing. heres a picture of the ref:


and heres just a quick viewport grab of it. its around 7000tris now, but theres some fat to be trimmed that got shoved behind some floating geo.



From what I have read so far its find that i am floating a decent amount of geo since it makes making all the slots/drawers much much easier.

any wood work fine details i plan to do in zbrush/textures.

because this is my first time building a game prop. all crits/advice is welcome. I really am unsure of the tri count allowed for something like this though. (larger piece of furniture)
lots of tri counts listed for characters haha just not as many for props. :think:

[orange13]

ok just an update. i am sitting at 7500 tris now but almost all the needed geo is in already (i am missing one last bookshelf part in the middle). I wanna try to stay true to the reference as possible.


going to clean up the geo on the leftside door. alot of the drawers etc. i plan to just overlap in the UVs to make an efficient UV layout. should be a good learning experience haha.

on that note...when you over lap uvs what happens if your useing sculpting data from zbrush?

[MRico]

on that note...when you over lap uvs what happens if your useing sculpting data from zbrush?

Just make sure to move your UV's over before baking and you won't have any problems.

[3DVlad]

I would have probably just had planes on those doors, and just faked most of these shelves using normal maps... 7500 Tris seems excessive for a game prop... This is more of a count for a character. Also alot of things could be faked using planes with alpha maps, like those little ornaments on top of the openings... But saying that, if you are just doing this to practice your modeling skills then its completley fine... in fact, then I suggest you chamfer your edges so they do not appear razor sharp, and model all those ornamental details... It just depends on what you are going for... could be good practice for modeling. You don't really need to take it to Zbrush, everything could be done in max here... Saying that I would have done alot of things in photoshop to save time, but if you need the practice then max it out... good luck =)

[orange13]

MRico>
ok, I will probaby just do them before starting any of the sculpting data.

3DVlad>
Nah its not for high poly modeling practice otherwise I would have taken a different approach. I think the pitfall of this item is that due to the shelves it requires more geo. Normal maps wouldn't work to well to fake this I think. (perpendicular geo wont work in normals from my understanding) However I am going through trimming a lot of fat in the shelf work to get the count as low as possible.
edit: your right about a lot of the alpha maps. going to fake all the openings with that, now down to 4450tris. i think i will stop there. now time to just move on. As I am going on with this project I am realizing how much crap i didnt need to do haha. or could have done much much better. (which is why i am doing it so its all ok i guess)

[MRico]

One thing you can do to cut down the tri count drastically is to float the shelves. Don't actually have them extruding and connected, you wont be able to tell the difference and you'll save a ton of tris.

[bcottage]

I would also do something with the hinges. Seems like your wasting many polys on something so small.

[orange13]

MRico> Yea i floated all the walls inside the shelves now too and reduced it alot

bcottage> yea i noticed i was wasting a lot on the hinges and other places.

now down to 2406 tris

working on the UVmap now but trying to see what i can just...shove together (notice mess in one area) Alot of the areas only need minor paint details (shelving, metal pieces, etc) so i figure it wont matter much (or so i think...haha) let me know if thats a no no

[Guedin]

2k5 tris start to be better. There's still some place you can optimize. Try to think by silhouette and shape. If a vertex doesn't affect the shape, then there's a good chance you can get rid of it. For exemple, the vertex at the center the side wall, the wall is flat so it doesn't help to define the shape and you probably can weld to get rid of them. it's just an exemple but I can see some more vertex or edges that are no need for the shape.
However, it's possible that sometimes, vertex which doesn't help to define the shape are useful anyway. For exemple, you may need to cut a mesh, even if it's not needed for the shape, because it's make easier to unwrap.

What I'm trying to say with my poor english is, try to make a optimized low poly model is trying to find the best balance between the polycount, the uv's space, and if your mesh is animated, the way it deforms. So, I think keeping this in mind as soon as you start modeling your mesh makes thing easier.

Anyway ... you have a solid mesh so far, and your optimization of the UV space looks good. There's is maybe just some place you can optimize imo, if you want to push it further.

Anyway it's getting better and better. Good luck !

[orange13]

For exemple, the vertex at the center the side wall, the wall is flat so it doesn't help to define the shape and you probably can weld to get rid of them.

I keep never thinking about this but...are N-gons fine in game world? I am so use to them being a no no for my zbrush work I never even think to throw them in. The reason I have that vert there is just to allow me to make the curve of the door smoother (otherwise it gets really blocky). However if i can just make that an N-gon panel then its all good...I guess since it doesnt need to animate (well no more then opening and closeing) it shouldnt matter right?:think:

[omnicypher]

n-gons are turned into triangles when put into a game engine, so it wont cause too many problems, but its better to lay out those edges yourself, so you know which direction everything flows. also, this is good for a low poly, but your zbrush model will need more consistant density of square polygons to sculpt well.

[orange13]

thanks for all the advice everyone!

Well I am moving back to the states this weekend so I have been busy as hell getting crap together. So I just slapped some really quick texture work down to get a feel for where to go with this (alot of fixes and crap to go) but i thought i would throw it up here real quick then once i got my feet back on the ground I will finish it up.

(kinda a beauty render haha...i should get marmoset toolbag to show this off. well that or import it into unreal engine...well once i am in the states its easier to get those things)

As for all the smoothing/normal work stuff. I have been doing some reading on things like xnormal program etc. I think since only a few parts will actually need the details (top pieces, legs, etc) it would just be much easier to take only those parts into zbrush to detail up or just model the details myself with high-poly modeling and then bake it. I want to add some wear and some grain to the wood all over but i will wait till the end for that. I have never worked with normal maps before so taking it slow is probably best haha. Since UVs are already laid out just doing it piece by piece might be best

thanks for the tips and all the mighty info filled in these forums. makes learning quickly easy.

[orange13]

after getting busy with other things I came back to the desk and got frustrated so I took a break and decided to do something else more to my tastes...makes it easier to finish at least.

I did the sculpture in zbrush and then decimated.

I am iffy on how I laid out the UVs though. I want to keep the pillar as one group due to how it flows in the sculpt. But its getting a bit distorted I fear and painting the texture might be a pain. I will probably do a mirror of the UVs, since I have space though I kinda want to be able to have the texture be not symmetrical. Right now the map is way to big. 2024 but the tri count is around 870. Idea would be that it would be hand held size idol. For a larger pillar version I might add more tris or just let it have the large size map (hand size only 512ish, pillar size 1024 or 2024 if it needed it)

I did the unwrap in a mix of programs, but i recently downloaded the max 2011 trial and did the baking of normal/occlusion and liked it a lot more then xnormals projection. So I might re-do the UVs. opinions welcome!

[orange13]

I started working on the texture now. I went back and painted up very loosely the original sculpt in zbrush, baked the texture quickly, then started painting it up. its fuzzy still which i don't like but my goal is a copper-ish worn metal.

i am using xoliul's shader for this.

[orange13]

well been working on everything else and then started doing things in UDK...which i got to find out is a lot more complicated then i origonaly expected.

still nailing down the material (especially for the idol itself) and the ground/vines are a WIP for texture.

all the textured assets i did just thrown into a cube with some lights.
I want to use UDK to render the shots of it but I am kinda lost at what the best method might be for some "beauty" renders. i turned AA up but it doesnt seem to be working as well. might be i need more work in UDK itself.

[orange13]

well, i was out of commission for a week due to sickness, so sadly i have not made much progress. I am running into issues with UDK as i learn more and more about it.

from what i have learned i am going to HAVE to rebuild the pedistol for the idol. Something i was thinking about doing anyways due to the fact that it became a larger part of the presentation and now i wanna add some glyphs etc to spice it up. However, that wasnt the issue as much as lightmap/normal seams were the issue.

here is an updated version at a MUCH better resolution (figured out how to do that finally)


otherwise i have been working in zbrush on a creature I will use for a game-ready creature, heres a render of the work so far from zbrush. (slight pose)

[BrendanBax]

that is not a wardrobe lol. too cool

[orange13]

BrendanBax - haha thanks

trying to get some more props done and such to really make sure i got a grip on this stuff and just to flush out my portfolio aside from zbrush goodies. so here is a crate. oh my.

[orange13]

getting on my next stuff. doing this as a zbrush pipeline method, created the base, sculpted it out, and now will polypaint it. Then decemate, uvmap (partially inside zbrush oh my), then bake and touch up a little in photoshop. my experience with Polypainting (all my textures have been done this way thus far) is that its to soft/blurry when put on the game asset.

just kinda doing a small scene. its ready for polypainting fun time...wooo....:uhh:
the bricks are just repeating of 4 of them, when i actually lay out the bricks i will flip them around more to make sure it doesnt look repeated. (both top and bottom are sculpted out)

[orange13]

more updates. the ground is kicking the polycount way higher then I want, so I will probably redo it quickly.

i used roadkill to do the uv map mostly. I dont know how I feel about decimation as the tool to use for creating the low poly. its really quick and easy, just probably would need to clean it up some to make uv-mapping easy and reduce random artifacts that seem to happen.

the tree is only 1k tris, tombstone 1.5k (could be reduced more now that i look), and each brick is only 166 tris. the decimation tool of zbrush really does make it look good with normal maps though. and easy baking with xnormal.