Weapon Shop Demo Reel Project
Hello fellow polycounters!
My name is Daniel Kaltenbach (my user name is derived from my bad luck with dice games...specifically Risk) and I will be graduating from school in the middle of October. I have from now until then to get my demo reel done and out in the industry, so I need to hit the ground running.
I'm posting here to get feedback and critiques and as a place to ask questions when I hit the inevitable roadblocks. Please be as honest and harsh as necessary.
Without further ado, here is my first update:
The next step, to be done today, will be to UV map the low poly and then create the normal and occlusion maps.
I will be doing 5-6 more weapons and placing them into a weapon shop (hopefully I'll get that 5 hour modeling time down a little lower...no idea what industry standard is on a weapon like this...anyone?).
The concept for the weapon shop is this (also from Mythic's concept art page):
Anyways, fire away with the crits, and I'll get back to work. Hopefully I'll have another update before I leave class today.
My name is Daniel Kaltenbach (my user name is derived from my bad luck with dice games...specifically Risk) and I will be graduating from school in the middle of October. I have from now until then to get my demo reel done and out in the industry, so I need to hit the ground running.
I'm posting here to get feedback and critiques and as a place to ask questions when I hit the inevitable roadblocks. Please be as honest and harsh as necessary.
Without further ado, here is my first update:
The next step, to be done today, will be to UV map the low poly and then create the normal and occlusion maps.
I will be doing 5-6 more weapons and placing them into a weapon shop (hopefully I'll get that 5 hour modeling time down a little lower...no idea what industry standard is on a weapon like this...anyone?).
The concept for the weapon shop is this (also from Mythic's concept art page):
Anyways, fire away with the crits, and I'll get back to work. Hopefully I'll have another update before I leave class today.
Replies
I will try and optimize a little more and post the update later. Thanks for the reply.
as poop said in the first tutorial i ever watched "if a vertex isn't defining a shape, you don't need it".
I have been struggling with getting clean normal maps out of this, but its made me go back and rework and optimize the low-poly, re-UV it (hopefully better, I'll post that for crits with the next update), and has overall made me rethink my approach. This will also let me make some changes that have been suggested that will make it better in the end.
Updates for will come at the end of the day. Thanks for ideas and critics everyone.
A better low poly version chopped to 200-600 polys would make more sense. Try putting a 100%selfilluminated black material on the object so that you can't see any shading or wires...thatll help you see what polygons are not pulling their weight.
anyway, thats knowledge that comes with experience. its still pretty obvious you know the tools, and if you can crank out a bunch of pieces with a similar level of quality (and kickbutt textures) it wont take you long to find a job.
This is not for a game like Titan Quest, where the axe would take up, at most, 1/10 the screen if the player zoomed in.
This is for a game like Oblivion or the likes where the axe could take up half the screen easily if the player is in first person mode. The final models and textures will be shown in the Unreal Engine 3, where they will be showcased and take up a good portion of the screen.
With that said, I am going to knock the triangle count down about a 1,000 tris or so...maybe less depending on feedback with the end goal of the piece in mind.
Thanks again for the ideas and feedback, I look forward to hearing what people's estimates for triangle counts on weapons in games like Oblivion and Unreal 3 are (texture map sizes as well).
If you look at those weapons, the polies STILL go to the silhouette--not much is wasted on inner detail that won't break the silhouette much--a little, to get shadow detail from it and depth, of course, but for the most part all those polies are used to create the form. in theory, a lot of the inner depth detail should be coming from the normal map.
A wasted poly is still a wasted poly, whether its on a 200 poly model or a 2000 poly model.
the more important thing i wanted to comment on wasnt really the polycount (perhaps the 200-600 suggestion was a little harsh now I think more about it--theres a good amount of detail of silhouette info to convey in the design)--it was more so the optimization itself--the places where you're still spending the polies that you don't need to.you seem pretty skilled--I'm sure you'll be able to get that puppy down quite a bit with some shaving, but if you want and specific feedback on the optimization just ask!
Alex
So I cleaned house, and then redid my low Poly, based on suggestions, and that quick mock up Sage did for me (thanks for that man, helped me quite a bit...if anyone wants to do paint-overs or do quick models as examples, I don't mind...it'll help me learn)
Here is the new low-Poly.
So here is my question going into UVing this low poly: Should I add an extra edge split to the blade and spiked handle guard right down the middle so I can mirror the UVs and save texture space?
Thanks for the comments everyone. I'll update with UV layout and some normal map tests next.
I would loose the three small extra loops on the round "bump" right under the blade, that small rim at those "turn-around" spikes, those extra loops at the very top and bottom of the staff and the vertex in the middel of the radial tris at the "grip thing".
Basicly think about how many pixels it will take even in closeup shots on the screen, and everything below the size of a pixel is wasted polygons.
AND also think about the size of the texels the normalmap is "projecting" onto the mesh, everything below that size will not really be visiable due to how the light is rendered.
Oh and I wouldn't use such long thin quads like Sage did in the grip... they usually lead to vertex lightning artifacts, even though that is probably much less important when a normal-map is rendered (uhh question to the experts: is there actually any vertex lightning rendered at all in modern normalmapped games?)
I'm going to start attempting the normal map baking now, hopefully I'll have some to show soon...and I am %100 positive I'll have questions at that point.
jkm pointed out a lot of the things you can still do to chop it ever further, particularly those rims on the shaft near the underside of the blade.
other things-- on the flat top/bottom of the handle; there looks like there's a floating radial vert in the middle (where the shaft intersects). that vert can be collapsed to one side or the other, changing the flow of the lines, but saving a few more polies--quite a few since it looks like you could make that same change in several places.
Personally I think you can get away with having the handle just be a few long polies--anytime the axe is REALLY close to the screen that's the part that will be covered by the hand/off screen.
There's also a few edges on the flat of the blade itself that you could collapse down, but that's a bit more of a crucial area, so its personal preference.
All in all, what you have now is conservative enough to keep going--its a LOT more realistic. Its nice to see you can take feedback well and work with it. :-)
"is there actually any vertex lightning rendered at all in modern normalmapped games"
I'm no expert, but yea, I think so. Depending on the engine, some will still mix vertex lighting with normal mapping. I don't think they do it on the same surface though...itll be a mix of vertex on terrain and normals on objects or whatever.
CursedDice I like your uv layout very nice. There is a lot of detail on the spikey bits, which I would model with a 3 sided cylinder instead of four like I did for my example, but I would try an normal map those bits on the flat plane with alpha. The thing is polies are easier to render than textures, so it may not be worth using alphas, in this case I think the polies are better though since they don't eat up as much of your texture. You may be able to mirror some of the spikes as well. The thing is looking much better though. Glad I was able to help.
Alex
I used xNormal for the creation of these. I've been fighting getting clean edges (the RED outlined areas) as well as getting rid of the 'squigglies' on the shaft (the green outlined area).
Any ideas on these? This is the first time I've normal mapped something 'complex' and could use any help or advice!
Jk my model is actually one water tight mesh, would it still have vertex lighting issues? The main reason I did that was to make a point, about how such a high polycount to make this model wasn't needed to model almost everything.
[/ QUOTE ]
The vertex lightning problem I ment is more or less unrelated to the "waterthightness" problem. Think about how the vertex lightning is calculated (please correct me if I am wrong on exact technical terms):
Basicly a gradient is applyed to every triangle that is calculated from the light that hits it's three vertexes, so if there is a pretty harsh light difference between the top and the bottom of you axe staff it gets averaged out over the entire length, and the longer the triangle is the more likely it is to result in a bad looking gradient.
And it's even worse in a game angine than the viewport of your application since all quads are transferred to triangles thus the benefit of quads (that they have a unified vertexlightning for both tris) is lost and vertexlightning artifacts are even more apparent.
But I think this doesn't effect normalmapped models since they aren't vertex lit.
So the final normal and ambient occlusion bakes won't be onto a grid like it is here...but this was just to show something. Anyways, let me know what you think, thanks for viewing!