Hey guys. I finally finished up everything for school (Graduated last Friday
) and I went a head and started working on this piece for my portfolio. As of now I want to get a nice foliage space with maybe some sort of buildings or old ruins.
As of now I went ahead and started working on some of the vegetation and such. I still have to make a few other plants and such to give it a more natural and diversified look. I went ahead and put this small test scene together to see how everything will work together and possibly build off of it.
Any feedback would be awesome. I want to make this to a degree of detail that I can present to a potential employer for an environmental design position. I don't have much industry experience other than iphone game development and working on mod team so any advice would be very much appreciated. I'll post updates here as well as my
artist blog.
Jarlan
Replies
My only crit is that the grass models you have vs the grassy floor texture seem to be a totally different scale. Unless that's a certain ornamental plant and not a grass, but that could just be a bit nit-picky for me.
thank you very much Makkon. haha yea that's my grass model. I'll try and see how I adjust the floor texture. The blades of grass are not huge but not small either so I have to see if making the floor texture grass bigger would make it look weird or not but I'll see how it goes Thanks for the feedback .
EDIT: I see you have a lot of information on your blog/site, so the answer will probably be there! I will check.
Thank you as of now I have one master shader for the plants that I then drop into a material Instance Constant to change the diffuse and normals. That way they all have very similar styled propertied but I can still change what I need. For the plants each one has its own Diffuse and normal. I can post some up if you'd like.
I'd suggest (in addition to what's already been said) modeling some clumps of dirt with moss on them and scatter them around to make the ground look like less of a plane. You could also have some upturned terracotta pots or a garden tap(i think americans call it faucet or spigot?) or something
They don't always do that anymore. If you check out the march udk update map (the one with the ruins), you ll see they use separate meshes for the grass blades (or sometimes 2 or tree per mesh).
Thanks for sharing.
Yeah, I know, and Far Cry 2 also uses mesh for its grass. But other than that... can't think of anyone else :]
Looks ace! Nice work!
@SHEPEIRO I'm on that right now. I'm also creating more smaller plants to add in as well, Thanks for the feed back
@Snefer I'm designing it more a portfolio piece but I've seen as others have said a few games that do a per blade or per couple of blades mesh. Thanks, I'll see how it goes with performance but for now I want to make something that looks real good
@Harry I'm working on those It wont be as flat as it looks now. Trees are going to be a pain later on haha
@pliang Thanks you very much. Yea as of now this is just my test scene that I put together to test out how my assets would work together but trust me there will be a lot more and a focal point
@Vig hah yes I agree. thank you
@Jeremy Thanks for the feedback. Alot of them are place holders, especially the ones that are standing. I should have some other type of barriers there
Yea sure I used a program called speedtree to generate the trees. It's pretty awesome how you can just draw out the shape and branches of your tree to pretty much whatever you like. For the leaves the program gives you the option of placing leaf plane on the branches you create and it also has a bunch of randomizing feature to help customize them.
For the leaves I put together a small cluster like this (making the alpha took forever ) and just drop it on the leaf planes
This was the first time i used the program but so far I really like it a lot and want to see what I can push with it. I hope I answered what you wanted to know
That is true in theory but in this piece there is no loss of opacity sorting as each blade of grass still has an alpha.
If you look at the polys of the grass here you can see judging from the wireframe shot that the blades of grass are not modeled out completely. Here its just each blade of grass has its only polygon plane with an alpha on it.
The difference with modeled out grass is that it should actually be fully modeled out. Having the blades of grass made entirely out of polys, they are modeled to have a pointed tip at the top.
Right now you are getting the worst of both worlds. Higher vert counts and opacity sorting.
You want this to be a professional piece but I am pretty positive anyone who sees this will ask the question "why do they have each blade of grass on its own plane?" And when someone is looking at your stuff online you wont be able to say "well the UDK GDC demo and Far Cry 2 do it" Not to mention the way you have it now isnt even like those too examples.
Other foliage is looking pretty sweet. Tree leaves seem way to light green. Maybe try and harmonize all the green colors by bringing them closer together. They dont have to be the same shade of green but just a little closer I think so you dont have such a harsh contrast.
You cant become a better artist by letting a program do all the work for you
Thanks
About the trees I usually model them all out, I was just giving this speedtree program a try to see how it is. o and the green is pretty much a very similar green for everything except one plant. It looks darker because that area falls under the trees shadow.
Again thanks a lot for all your feedback. I just recently finished school and haven't broke into the industry yet to pick up a lot of the tips and trick of it so your being super helpful
@snow It's a mix of both gathering and web. The web ones I got from cgtextures
If you remodeled the bricks to be a unified/modular mesh without the deep crevices you'd get basically the same amount of depth but you wouldn't be wasting the geometry that goes in the cracks that the player can't see anyway because it's blackened out by your AO.
I'm sorry if I seem like an asshole, I'm just really confused by how much praise you've gotten because what I see is a really unoptimized scene whos key focal point is a tree you generated and didn't make.
I can understand wanting to cheat a bit to get your portfolio shot to look good, maybe use some inefficient shaders, maybe skimp on geometry that doesn't face the camera, but no one is looking through your portfolio because they want you to demo SpeedTree for them.
Cool thanks for all the feedback. I'll go ahead and try some of that out You don't seem like an asshole. You actually seem like someone who is trying to help. Trust me I don't get my feelings hurt to easy. This is all a learning experience and sometimes things have to get told how they are. I'm here to learn as much as I can and pick up whatever industry trick I can so I can break in.
That speedtree is not going to be the focus of my scene I just made a few to try out but there is a lot more coming. This is still in it's baby stages hehe
Anyways thanks again. I appreciate it
Right now though, your leaf texture is what needs work. Your leaves look nice, but only if you're looking at them from above. Real time leaves need to be viewed from a side/downward perspective. We always make our leaf textures by creating a cluster of individual leaves and rendering them out to get a proper perspective. Also, you could use some branch textures. They add a lot to the tree without wasting too many polys. Be careful with too many transparent pixels though. Keep it up.
As was pointed out to me in the past, that's not really a valid statement at all. SpeedTree needs to be licensed for anything commercial and if you don't have access to it, or another tree generator then you have to make them by hand.
After learning to use SpeedTree, and making decent looking trees by hand, I certainly know which one is the harder of the two.
I think knowing how to make foliage by hand is a great skill, even if you always use SpeedTree for every tree you make in future.
I fully agree; when people think of Speed Tree or other tree modelling apps they automatically think of a magical Make Art button. This couldn't be further from truth, since trees that are automatically generated by Speed Tree tend to either look like crap or be useless for real-time rendering.
Getting nice results requires a lot hand tweaking and though many seem not to be aware of it, ST does allow you to edit each element separately. Don't think of it as a generator but a dedicated foliage modelling app.
It's not like modelling trees is a difficult, technical task either. Modelling a trunk and a bunch of branches follows the same basic principles of poly modelling. If you can model a gun, you can model trees.
What makes the foliage tricky is the canopy; distributing it by hand takes a lot of time and effort, whereas ST offers a more streamlined approach. However, don't be fooled into thinking that simply connecting nodes will do the trick; to get nice results you would have to play around with distribution curves, tweak some of the planes by hand and most importantly, pay attention to the references.
To be frank, creating foliage in Max/Maya/etc. is one of the most tedious and boring tasks. Not without a reason, trees are one of the most complex creations of nature. If feel it's safe to state, that if you can take any shortcuts to speed up their creation then for the love of god you should do so. Whether you are going to use, Tree[d], Speed Tree, Onyx, or some handy scripts, as long as you can come up with good results, you should be able to replicate it in any chosen app. In some, it might take you more time than in others. But as long as you have a good understanding of how a tree is supposed to look like, you should be fine.
Sorry for this rant and coming off as an asshole, but this has been really bothering me for a while.
That said, I would avoid putting only Speed Trees into portfolio; it's pretty damn expensive and not all studios are using it. You could model some trees in Max/Maya just to prove that you possess the ability to adapt to different tools.
Also fully modeling the grass out can create WAY too many verts which will kill the vert shader and the shadow pass.
I think what makes foliage difficult, at least from my experience is the vertex normal tweaking you have to look at, and the alpha considerations.
This is why I think SpeedTree takes all of that ballache out of foliage making. As you say, modeling a trunk / branches is easy. Distributing the planes etc for the foliage is not particularly hard if you have ref. But then getting it to look good in your realtime scene is not easy, whether in Max with a DX shader, or in engine.
You can spend all the time in the world making wonderful looking trees in SpeedTree, but then when you're done, you hit export, check a few settings and you're done, pretty trees ready for realtime.
Finishing the model/texture on a hand-made tree is only half the process.
The way I see it is that if I was to get a job, or even an art test and I was required to make a bunch of trees, if all I knew was SpeedTree, then I'd be screwed if said company weren't licensing it, or specifically wanted hand made trees.
I dunno, I'm not experienced or employed so I can't make a real judgement, I still think it's kinda silly to rely on SpeedTree unless you are 100% confident you can nail trees by hand.
You make SpeedTree sound a lot more practical than it actually is, in reality it's incredibly expensive and the cost can only be justified if the game has a large number of trees. It's a powerful application but that doesn't make it cost effective in every situation and because of that fact, most studios DO NOT USE IT.
That being said, nice looking work man. : ) I would however modify the scale of your grass base texture to more closely reflect the 3d grass you have on top, it kind of makes the grass seem 'giant'. I'd echo what others said about the bricks. The tree assets also look excellent, althrough it's a bit strange the leaves all kinda pointing upwards. The leaves towards the bottom would be drooping downwards more since their branches don't have the same support as the ones above.
If you've ever modeled anything other than a generic Oak, you'll realize there is much more to speedtree than tweaking values.
In the end it's what gives you the product you want faster.
I was stating on the importance of learning to model trees by hand as for learning purposes. If your company uses Speed Tree hey thats awesome what ever is the most cost effective way to make awesome work is great.
Chances are you wont be at a company that uses Speed Tree and would have to model them by hand unless a ton of companys are now using speed tree. Which I doubt but cant say for sure.
As for the grass again JarlanPerezsaid he wants this to be a piece that he can use to apply for a potential environmental position so using todays standard techniques that are used in an actual production environment would be best IMO.
For grass that would be using alpha cards with multiple blades of grass on 1 poly. Its very easy to give that "thicker" look by just rotating cards around not to mention giving you a nice 360 grass effect.
Think its better to show off that you know what is common place in the industry today and can create great looking work using those techniques instead of basing it off a demo, which is after all a demo and not a production game.
again totally dependant on player size...if six foot fine...if, as im thinking in this case it would be cool if he was six inches tall...this is a way better way of doing the grass...
I don't even understand why people would see the 1 poly card as efficient? Yeah cutting closely will add more verts but if the scene isn't vert shader bound or slowed down by the shadow pass who cares? Most of the time the scene will be slow in the pixel shader and the cause of that will be overdraw with foliage so why not reduce it?
I know the scene is still very early on and I'm sure it can be assumed that the scene is going to be mostly speedtree trees and other foliage I make but it isn't. There is going to be a lot more to the scene where the trees and such are going to be slight additions. I have started blocking out some of architectural assets that I will be producing for this scene. As soon as I get a chance I'll post up new images to see what you all think. It may be a few days tho (I have to finish up some work for a client) but don't worry I'll have something.
I really truly appreciate all the feedback and interest that you all have given and shown to me. It helps tremendously. You all have given me a lot to work with and learn from and its cool to see everyone's views on a topic.
P.S thanks, oh and some variations on the texture of the tree, darker greens, would help