Environment | Mountain Mining Town (fantasy)

[Pizza]

My only crit is to make some of the window lights turned off.. they all seem to be on. Other then that very nice.

gj on the models and textures

:thumbsup:

[goodlag]

great work!

i agree. Pizza :drool:

[Annuostivix]

wow, great job. One day when Im into the industry I'll call you up and get you to make environments. Unless you're way ahead of me like you are now, and people with more money will hire you. we'll see, hehe. Iunno what Im getting at really. :D

[jfalconer]

looks pretty good, but for the prerenders it could really use some bump mapping to bring out the cobblestones. The textures on the temple look very dull and could use more detail. The flags on the temple look boring to me, since they look brand new. Remember that whenever something is brand new, it tends to lack character that comes with age and use.

One thing that is really detracting from the look of the renders is your lighting. The shadows being cast are too crisp. The lights themselves could probably be a more orangey tone compared to the bright yellow which looks more green when mixed with the ambient blue tones you've put in for that "night time" effect.

Also, take a good look at other styles for the houses. For this type of scene, there are many different ways to build the houses. Look at distorted towns from "The Nightmare Before Christmas" (Disney), and "Legend of Mana" (PSX) as a good starting point. From what you have here, it looks a lot like the town called Kalm from "Final Fantasy 7", but with less details and less variety in the houses. Also, the town from FF7 looks very outdated by today's standards.

I would say it is a good start to an environment, but needs a lot more work.

[Artician]

Hey guys. Thanks for the comments!

falconer:
I agree with everything you've brought up, thanks very much! The initial goal for this was a more standard-looking, modularized piece. I have some other work on my site that is much more stylized, but I don't have much current work that is of a more realistic look which is why I went with this.
For technical reasons, the majority of the buildings are instanced geometry. Each section had to be similar in form and dimension, so all could snap together to form as many different combinations as possible.
Also, due to some lighting issues with the engine I'm using it was recommended that I show something a bit more polished than just the real time shots. I don't usually work in pre-rendered media, so I struggled a bit with using the advanced lighting techniques in Max. I will try and touch up the lighting as much as possible, especially since that is the sole reason I went ahead and included pre-rendered images. Your input was very helpful there, thank you very much!

Aerestis: I am presently happily employed, but who knows?! ;)

Thanks very much again for all your comments!!

[Dhin]

Well I have only on crit for you. The lights when navagating though the environment slow down the navigation. The other problem if frame rates at different views. I.e. whe look at the ground frame rates and nav is much higher that when looking level and moving along one of the streets. Over all it's nice though an I think in a different real-time environment it may run better. Usually in a RTE you should keep your view limiteds to as few polies as posable.

I'm working on some RTE's for web use as well. I've been using VRML and Cortona VRML web cleint for it. It tends to suit my needs right now. I have a link to an example of some of my work in my sig, if you want to check it out.

[Artician]

Hey, thanks for responding.

Please let me know what your CPU speed is and what fps are you getting in and out of the town area?
You will always see an increase in framerate when looking away from a high density section of geometry in any game. At least from my experience. Though it's ideal to keep them as similar as possible, which makes me wonder how bad the performance is on your machine. If it's so noticable, then I will work on smoothing it out as much as possible. Thanks for letting me know!

Do you have any experience with portal culling? It's a much better way to spread out what areas the engine focuses on rendering, but applying it to outdoor areas is tricky.

Also, I don't really develop for web based content. The development environment I use is used for porting to XBox, PS2 and PC. It comes with a web based player though, so it's convienent to show work.

Thanks again!

[EvoG]

Artician...how are you doing portal culling in an outdoor environment. Be as technical as you like. :D

Thank you.

[Artician]

EvoG:

At this point in time I am not. It is something I wanted to discuss with other real-time artists so I could illuminate the subject for myself.

If I were to employ that in my current scene, I would split each path of the town into it's own section and place portals at the intersections of the area.
There would be difficulty in changes in height, but that I have handled before by segmenting more open areas, and having lower detail geometry exist outside the occluded space that is representative of accessible locations in the level. Does that make sense? Explanation by text is poor.
Dealing with the open sky would be the most problematic. I have no idea how to deal with that at the moment.

What are your experiences with this? Have you used it in personal projects? Or could you point out an existing piece of work that would be a good example?

Thanks very much!

[EvoG]

Well I was curious with your implementation and if maybe you found a way for it work on a marginal level. If you're doing 'cities', then you can use portals to to define areas throughout that could be synonymous to hallways and rooms of an interior space. My experience lay in these examples above, and I'd be interested to see if there's a way to 'kludge' it to work for outdoors. As I understand it, Virtools uses a simple 'visibility' portal check, with hierarchies tied to the portal(which is an object itself), so if the portal isn't visible, neither are the 'children'.

Just for the sake of talk, you know that Unreal uses true portal technology. Well the most interesting yet little talked about feature the engine has is the 'anti portal', useful anywhere, but VERY useful as outdoor occluders. Whereas you place portals in doorways and windows, anti-portals are actually submerged into level geometry. Take a large hill outdoors for example. You would place an anti-portal ( these can be 3D not just sheets ) of roughly the same shape and dimensions of the hill, INSIDE the hill. What then happens is that everything behind this anti-portal, referenced in screen-space, is 'masked' away...occluded. It is VERY powerful for making wickedly detailed and very large maps. Very exciting and I wish I could see it implemented in more realtime engines.

Oh, well, keep us informed of your experiments! :thumbsup:

Later

[Dhin]

Well it seems like you already know the areas of difficulty in your model. In answer to your questions to me, my home system (where I first checked it out) is pretty standard for low to middle end pc's 1GHz Amd CPU, 512 MB RAM, and a Diamond Viper V770 Ultra vid card with 32MB RAM.

A scene (minus characters) should be less than 15,000 polies. Unless you will never see any characters in the environment then you can increase to nearly double that. This is of course dependant on rendering engine restrains (if you are using an older engine like Dark you're totals are much lower, Character limits are under 1000 polies, and scenes poly counts aren't really limited but viewing polies need to be under 500 for optimal frame rates.) Your scene right now is within the "no char" limit but the partial effect for the flames suck your frame rate down quit allot. To fix that, add "ambient" lights (e.g. lights with out actual sources in the scene to act as fillers and such,) and cut down on the torches.

One thing that always works to save someone from viewing to complex an area and thus losing frame rate (without portal culling) is to limit the view. Try to keep the number of viewing polygons as low as possible. That means a little more scene refinement as far as you poly count goes and use objects (such as walls, gateways, etc) to limit the distance the user can see. (Also instead of a long street with no viewing obstructions, short winding roads where you can only see a short distance in front of you.)

Hopefully that is helpful to you (maybe you already know all of this any way.) Good luck keep us posted on the progress.
:thumbsup:

[EvoG]

I didn't address the question of performance, so I'll add to what Dhin posted.

First off, numbers mean nothing. Don't determine the perfect solution for every machine...thats not the way to think. What you want is scalability, if indeed, a larger audience is your goal. If you need to capture the slower users, then you need to allow your system to be flexible to downgrade the quality of the objects and textures.

Now, the one most overlooked killer of framerate is fillrate. When you have lots of large objects, one in front of the other, you will see slowdown regardless of poly count and texture size. If you were looking at a wall on the side of a building up close in 1st person, and there were several very large walls and buildings behind that wall, the computer as you know, renders the furthest object first, rendering your entire scene from back to front. Each time a single pixel needs to be redrawn, that starts to cut into your framerate. So if you had 100 objects one in front of the other, quite a few of your pixels are being drawn 100 times before coming to the final color. Thats not good, hence why portal culling and any sort of occluders that limit geometry unseen behind objects are a serious benefit to performance. Just to note, what you said Dhin, about 'limiting' the polys in view is actually a misnomer...as you can't really 'limit' what the player see's, especially outdoors...or you'll come across the law of diminishing returns. If a player can position himself around a map from a variety of angles, he's bound to see more than you may want him to. If you keep removing polys, you end up with a very sterile scene. Distance has nothing to do with anything other than fillrate and the amount of objects rendered, so placing obstructions to 'occlude' the view only exacerbates the fillrate problem.

A more flexible solution to distance and viewable polys is LOD. Make sure your objects in the distance are as low a quality as needed just to represent them...that or lose them altogether. 90% of your details should be in just 10% of you view. The other 90% of the view should be 10% of the total detail. Maybe thats a dramatic percentile, but my point is more important. LOD CAN kill your framerate too, due to the processing of the LOD if its realtime, or to storing extra objects to be swapped out if they are preprocessed. Its almost like you're damned if you do and damned if you don't. Balance is the key.

Oh well, I just ran out of 'thought juice'. I 'thought' wanted to say more but now I forget. :shrug:

Later fellas

[Artician]

Hey guys.

Dhin:
I more or less had an idea where I should have spent more time on it, but I wouldn't have put it online unless I wanted others opinions. It's good to have outside input. I really appreciate you taking the extra time to download the real time files since that's what it's all about. Your input means a lot!
I agree with your input on the limited visibility. The only way that really helps though is when there are separate areas not loaded at the time (very similar to portal culling). I think EvoG commented on the way scene geometry is sorted.
Also, I tried checking out your site, but was unable to get the plugin to operate correctly so I couldn't see your work. :( I'm interesting in checking out what you've done though so hopefully will get to see it soon.
As for polygonal limits, I usually aim for anywhere between 50k - 75k polys per environment. That depends greatly on the target platform and what kind of game it is though. Current PCs today push a lot more than that, so it's best to aim high. That's why I was curious about your machine specs. As it is at the moment, the torches in the level are probably WAY too detailed for their own good. I could probably lower them by 5 or so times and still get an acceptable look out of them, but I didn't drop below 30fps on my machine, so I didn't bother at the time.

EvoG:
I wish I had the time to script a more flexible front end for these projects. Unfortunately I'd rather spend my time working on the artist aspect than the gameplay/scripting aspect at the moment. In a more preferable scenario, I would probably add the ability to alter texture settings, LOD toggle and the like.
I have used manual LOD before, but it's time consuming to create. File size goes up, but processing time goes down (as opposed to on-the-fly LOD). I prefer that to real time tessellation by the engine because I have yet to find an engine that accurately redraws a model in lower detail.

Lastly, what you said about the Unreal2 engine really gets me curious. I could probably ask this on a more fan oriented site, but since we're on the subject: can the Unreal2k3 Tournament editor be used to create Unreal2 content (and vice-versa)? I know the engines have basic similarities, but I prefer some interactive single player environments, (with possible multiplayer capabilities) than to creating just straight multiplayer "deathmatch" style maps. At the very least, it sounds interesting.

Thanks for all the input and info!

[EvoG]

As you may or may not know, the engine ( Unreal Warfare ) is at its core, the same for UT2K as it is for U2. Legend of course added a bunch of code to Warfare, but its essentially the same beast. You not only can create lush, dense foliage, but the sheer size of a level is unprecedented. One of the levels we've worked on takes 5 -10 mins to walk to the 'end' of, and thats in a straight line. Its an open space area, with very little linearity to how you traverse the landscape. Very impressive.

You will have nothing but joy working with UED. Go get a copy of UT2K and be amazed. True terrain, decos, antiportals, static meshes, physics engine, material shader editor. Its quite sophisticated.

The downside is that you need to do real coding to create a more immersive level of interaction than the basic 'deathmatch' gaming UT2K offers. You can do leagues more intricate 'code' with Virtools than with UC, with basic understanding of logic and organization, but, if you are a competent C++ coder ( or C period ), you can do quite a lot with Unreal as well, if not more. Just see Thief 3, Deus Ex 2, Unreal 2 for starters. :buttrock:

Cheers

[Dhin]

Artician - Umm, what browser are you using? I believe Cortona is able to run in Opera, Mozilla, Netscape, and IE.

I may have to remove a little HTML code on the page to make it more browser friendly though if you're having problems (I haven't done a lot of checking it on browsers other than IE, sorry.)

If you clicked the Cortona icon and downloaded the plug in it should just up and run? But here are a few links that might help you to install another VRML Client or reinstall Cortona:

Cortona - http://www.parallelgraphics.com/products/cortona/download

Blaxxun - http://www.blaxxun.com/services/support/download/install.shtml

Cosmo Player (Development Halted so a bit out dated now) http://www.cai.com/cosmo/

Each should automatically install or offer a download that you can install manually. But after they are installed you'll need to reboot in all likelihood. If you're using an OS other than Windows and Mac OS I don't think it will work. (I haven't seen anything about Linux support in any of these clients, but I admit I haven't looked into it to much.) I hope that helps you a little on that problem.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.