View Full Version : FPS Prop - Gas tank

12 December 2006, 02:02 PM
Maybe a bit cliched but I'm working on a mod and a science lab calls for it. It's been awhile since I posted any work I've done here so I thought it's time for some feedback.

I just rendered with a skylight at 50% and an omni. I'm afraid it's a bit to heavy for a source mod at 915tris. What do you guys think?

12 December 2006, 02:41 PM
it's looking pretty good. just two things, that nozzle handle up on top might be a bit dense if this is just a room prop. but no real worries on it. also, the gold on the gages & the maroon of the tank are pretty saturated. they look like a pimp scientist's O2 tank lol just a suggestion but if you dull the saturation a bit on them it may look a bit more like it fits in a lab. maybe some scratches through the paint exposing the metal? but that's just a suggestion, take it or leave it! keep it up though, show us some shots in the mod!


12 December 2006, 03:31 PM
It's a good start, but a couple of things stand out to me.

First - It feels very clean. Dispite the label being ripped off as if it's been in use for a while...The rest of the texture is overall very clean. I see that you have some scuff/noise going on to kinda break it up, but add some chips, heavy scratches and other forms of wear/tear. Especially towards the bottom. People picking that thing up, and putting it back down will do a lot of damage to the bottom over time. You may just want to make the scratches a grey color in the diffuse, then give them a hotter appearance in the spec to make them really stand out when the light hits it. Make sure to get a lot of good reference for what you are working on.

Second - It looks like overall you don't have any lighting painted into the map (which is obviously what you want to do when using normal/spec) however on the nozzle (the blue piece) You seem to have really painted in some high lights that stand out. As well as a couple of the brass/yellow areas. This stuff is alright I suppose in the source engine, but since you have that nozzle fully modeled out (to represent what you have painted) and I figure you might have normal mapped the area a bit too, then let your normal/spec/geometry do the work there, and not force the lighting to always look the same. That's the whole point of spec. Sure, you want to paint in a little lighting here and there so that certain areas don't get washed out when the object is in a shadow, or a poorly lit area. But, you don't want to make it so that the lighting in your diffuse is so strong that it can negate your normal/spec map.

Seems like an overall solid model...Does feel a bit high for a prop, but so is 5,000 polygons for a first person weapon, which people do all the time in mods, so sure, why not? :)

Also, if you don't already, I would strongly urge pursuing Ambient Occlusion. It is a great way to get some lighting detail into the diffuse, without it being overwhelming. And the best thing of all, is that it's all automated in the 3D App (at least Max/Maya) :)

Best of luck, look forward to seeing your progress.

12 December 2006, 05:17 AM
Thanks for the tips guys

Ill_logic, I'll play with the saturations to see if I can get it a bit more believable. I know the blue nozzle is a bit dense, but the rest of the model doesn't call for as many polys. Everything else is just cylinders of varying radius, height and edges.

Flewda, you're right I can definitely use some scuff marks and scratchs towards the bottom. Thanks for the help on the lighting! I am always confused as to how my bump and spec maps should match up with my diffuse. Since games are allowing more and more real time lighting it seems like I shouldn't be painting so much lighting detail into the diffuse, for instance the numbers on the gas tank. Should I take that out and leave it in the bump and spec?

Just out of curiosity how long should this take a profesional? My current contract is up in April. I feel like by then my 3D skills will be ready to go professional as a junior so I can begin learning from some of these senior guys but I'm a bit worried about my speed. How long should this take for the whole process? Researching and finding concepts, modeling, laying the uv's, all the texture sheets, a few renders, saving everything and putting it in it's required location?

12 December 2006, 05:24 AM
You would most likely want some sort of presence of the numbers in the diffuse. Even a subtle one. The easiest/quickest thing to do is to take the numbers, and just make them a slightly darker color, then add some subtle (very subtle since it's the diffuse) scratching around the edges. That way it's still visible in the diffuse, but by no means contradicting what the shader is doing.

Hmmm, as far as time goes, there are a lot of factors. Genre for example is a big factor. If it's a 3rd person game, there's a good chance you'll need less detail than you might with a first, so you would probably be expected to get it done fairly quickly. Also depends if you are building a high res model for it to build your normal map with. Every company has different expectations, but with a prop as basic as this, and with so many reference images you could quickly find, I would think it would be a prop that should be built in less than 3-4 hours, model, texture, and in game. Just my opinion.

12 December 2006, 02:39 PM
4 hours?! I really got to push it! I spend a lot of time baking out maps though. I haven't found any good settings I like. If I bake a map with a sky light I can never get padding to work. If I bake a map with Ambient Occlosion I get serious ladder effects at all my seams. I think maybe I don't have my ambient occlusion settings right in the render to texture dialogue. What are some good numbers?

12 December 2006, 03:53 PM
Yeah, the 4 hours is kind of low if you are baking a high deail ambient occlusion. But shouldn't be much more than that...

I typically use skylight, light tracer, with a white material on my object that's being projected. That seems to do the trick for me. If you are using a high res object, make sure your low res object is completely encased in your high res (so that no pieces are extruding out of the high res). This can cause it to miss the high res geometry.

01 January 2007, 04:58 PM
I worked on this baby and finished it up now. Here is the final version. I lit the scene real quick with 3 omni's.

01 January 2007, 05:29 PM
Vast improvement! Looking good! Kinda hard to tell from the render/lighting the little details, but still leaps and bounds better! Good work!

01 January 2007, 01:54 AM
Just out of curiosity how long should this take a profesional?

3-4 hours

Yeah, i'd say maybe an afternoon for something like this.
research for this would take minutes. modelling very quick. I'm fussy about how i lay my UVs out and would probably try and get this on a 512x256 (512x512 is pretty high for something like this).
texturing. i'd probably add some dirt, metal overlays on this too. Flewda had it right about the adding of painted lighting.
for additional lighting help, i add vertex color lighting with a radiosity sky light and an omni.

01 January 2007, 12:58 PM
3-4 hours? Hmmm I got some practicing to do. I'd never be expected to complete more than 2 assets of this complexity in a work day, right? I didnt go 512x256 because I haven't found the process to be made easy yet in max. My mod team hasn't been specifying texture or polygon limits so I just take my best guess.

I'm suprised you guys can't see my baked lighting. It is a skylight. I think if I turned it off you would definitely notice the difference. Also I didn't want to use too many distinct details since this will be used more than once though the mod.

01 January 2007, 01:49 PM
You'd be surprised how many assets you would be expected to create in a day (especially a crunch day). As you go on, you'll get faster and faster, it won't be a problem.

It is to be expected that you see a prop used very often through out a game. Making new prop assets for every single level in a game is just nonsense. With that said, don't use that as a reason to not put some unique detail in it. The only time you may not want to do that is with a tiling texture. If you have a wall panel that tiles 10 times across a certain wall, and on one of the panels it has a small Warning Sign or some number on it, then yes, that would be bad. But if you put a label, stickers, some stenciled numbers/letters, unique scratches on one a prop, no one is going to notice (or care). There are some exceptions to this, so use your judgement. For example, you may not want to have a really unique and large stain/rust spot on the tank, because then if you have 3 or 4 tanks next to eachother, it will be very obvious. Again, people won't really care about props being the same over and over, but aesthetically speaking, it won't be very pretty to see 4 props tiling. So use your discretion, and then when throwing them into a level, make sure your twist and turn them around in different ways so that they aren't all lined up perfectly. This will not only give a real apperance of randomness that you would get in the real world, but also hides some of the reused "unique" detail that's showing through all of them. Hope this makes sense. Just woke up, so not entirely sure what I am writting :)

CGTalk Moderation
01 January 2007, 01:49 PM
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.