Glock 18
polycounter lvl 6
Hey everyone, working on a glock 18, about to get on the road but wanted to get what I had up so I could have some critiques to look forward to when I get home in 4 hours.
So far I am calling the slide finished (before taking into zbrush) still have a few details left on the handle to model before I move on to zbrush but feel free to tell me what's wrong with this, be sure to understand this is the glock 18 (with the regular sized glock 17 clip) and not the 18c or any other model.
I love how polycount has so many people who are gun nuts and will notice every detail missing, have at me!
The wireframe is quite messy and I could optimize it a ton but that's a last step, but I didn't want to leave you guys without a shot of it just in case
Ref shot of an 18C, difference is the 18 doesn't have the vents on the top of the slide.
So far I am calling the slide finished (before taking into zbrush) still have a few details left on the handle to model before I move on to zbrush but feel free to tell me what's wrong with this, be sure to understand this is the glock 18 (with the regular sized glock 17 clip) and not the 18c or any other model.
I love how polycount has so many people who are gun nuts and will notice every detail missing, have at me!
The wireframe is quite messy and I could optimize it a ton but that's a last step, but I didn't want to leave you guys without a shot of it just in case
Ref shot of an 18C, difference is the 18 doesn't have the vents on the top of the slide.
Replies
In hopes of provoking some replies i'm going to make a bold statement - my current model is perfect to that of a glock 18!
Anyways have a look, I have finished the high poly for the handle (only new part) other than having to add the noise, labels, and the switch that releases the slide for dissembly, this is pretty much it. Let me know what you think please.
I thought the Gorilla Mat looked quite nice for a quick preview
Some shots in gray so that the detail is more easily seen for maximum crits!
Low before high? Generally this is a pretty bad way to work, you should do your high, and then build your low around that.
Details in zbrush/mudbox? This looks pretty sloppy to me honestly, and would serve you better to learn how to do these sort of details with proper sub-d. Especially if you find out that your details do not bake well in the final map, or the resolution of your geometry is less than that of your final texture, it will be a pain to go back and tweak.
As EQ said, the HP looks sloppy. The over all shape looks good, (Saying that without looking at an reference. I don't know what a G18 actually looks like in detail) but the detailing and extrusion details is super sloppy. And the shape on the handle that's for the thumb is really wobbly.
As far as you lowpoly goes, you have alot of edges that runs over flat surfaces - waste. In hopes of provoking some replies i'm going to make a bold statement - you can cut ou about 500 tris!
All the luck (:
1. Read the thread title
2. Who hasn't seen a glock!? XD
Zbrush indeed, as for building the low before the high I realized when you said it that's probably why it's taking me so long to finish this piece.
As for the sloppy look overall I could only get it to 6mil poly's on this comp, if I could get one more level (i tried - 24 mil poly's won't run) then they would probably be a lot cleaner, I would love to learn to do this in subd, can maya really do it? That's the package I know but it seems like it's really max that excels in this field of modeling, I need to learn it but have so much on my plate!
My final texture has to be 1024 x 1024, and has to leave no detail to the normal map alone, meaning the low has to have some extra geo is how I take it. Limit of 9,000 tri's but I don't think i'll be anywhere near that.
Thanks
Oh sltrOlsson you made me laugh, this is a commission work, I would never do this on my own as I know it's very overdone and boring as far as eye candy goes, that is why I was posting though, to hear this type of "it's bad" feedback with specifications as to why. I don't have time to go back and redo the high entirely anymore
As for the lowpoly i mentioned it could use a ton of optimization, that was just in the process of being built without worrying at all about polycount, I am a man who loves some optimization and i'll be sure to get rid of those extra loops trust me
Thanks everyone for your comments it's nice to hear back from the community!!
Here is the final high - (handle only, but the rest doesn't need it, im going to apply the wording within the normal maps themselves)
Wire for the handle only, currently 2803 Tri's, total tri count is 5945 Tri's, this is pre-optimization counts though.
hope the diffuse and specular are just as high quality.
Why? With hard surface objects, I find it a good way to go about it. That way you really aren't modelling something twice, more like 1.5 times...saves me a lot of time.
A. Its best to match the low to the high, to get the most accurate bakes, doing it the other way around is restrictive and limits you to the exact shape of the low, which isn't good from an artistic point of view, as you will likely need to adjust proportions or details when you're doing the high.
B. Working from a lowpoly -> high is very messy, you end up re modeling the high from scratch as working from an optimized lowpoly mesh is never ideal, so you're not actually saving anytime, or, if you're reworking your low entirely to make a usable base for your high, you're actually spending MORE time doing it this way.
C. For complex areas, sometimes it is much faster to simply model a quick low around it, than reducing from the high. Going high-> low doesn't mean you have reduce every little bit of your high back to a low. However, if you're modeling clean, this is usually just a matter of quickly removing some edge loops anyway, you're not "modeling it twice". Generally reducing down from your cage tends to be a very accurate way to match the high as well.
Some people do blockout->high->low, and use the blockout(after re-matching, and optimizing) for the low, however I generally find it easier to remodel the complex stuff quickly, and optimize from the high, as polyflow and proportions tend to differ drastically from the blockout.
At the end of the day, do what you're comfortable with, but high->low is basically the universally recognized way to go.
Right, this is one of the bad reasons to try to do this in ZB, because it just takes a really excessive amount of geometry, its just not efficient.
Maya can do it, modo can do it, blender can do it, its just a matter of learning the tools and learning how sub-d works. Learning the concepts of sub-d modeling is generally more important than the specific toolset.
I would recommend sifting through this thread:
http://www.polycount.com/forum/showthread.php?t=56014
If i don't add these edgeloops, then go into zbrush and subdivide without smooth on, then my normal map ends up being faceted for obvious reasons.
How would you go about this process to create the same object?
I guess this is closer to what I actually do. I don't go and create a high from a mesh that has been optimised and has triangles all over, otherwise Id get all sorts of pinching if I chamfered edges bordering triangles...but my blockout is very very close to the finished low. I spend a lot of time on it. Just needs a quick optimisation, some vertex welding. I find that faster, plus you get a better idea of the finished product earlier on. It don't matter if your high is a marvel of modelling if the low that's going in-game ain't up to spec.
Dumb question - so it doesn't matter if there are ngons as long as it looks correct? I feel if i had known this early on it would save me a lot of time. Also do you soley mean sub d or just using the smooth mesh preview? I've never even messed with subd meshes and am not even sure how to start, but if that's what you mean i'll get to learning
Just to clarify on what I was talking about earlier.
This is a screen of a bushnell holosight I am working on. It will serve as the base for my high, but also the final low. I will only weld verts in the most obvious places (the grip wheel thingy, some parts of the main mesh that have cuts not neccesary for the low. I find you save a lot of time with this half-n-half method.
I also used this method while modelling this.
I find it keeps your focus on the whole project, all the way through the pipeline, the way it should be
A. I would model the geometric shapes with basic sub-d, and do the font and small scratches and such in the texturing phase, there is no need to go to zbrush for this sort of work, photoshop is going to be much faster. I'll see if I can find some time later to whip up a quick sub-d bullet mesh later. Going to zbrush for this stuff makes it needlessly complex.
B. Yes n-gons are fine, only the final result matters. As long as the n-gon isn't adding pinching or poor smoothing, then there is no issue. Just do what looks good!
C. When I say sub-d I mean general sub-division smoothing/modeling, in maya this is split up into a few things "smooth mesh" and "smooth preview, which is weird, but they're really the same thing. You'll do the actual modeling work with "smooth preview" but you'll have to "smooth mesh" at the end to get a bakable result in maya.
Really as far as your process, as long as you're not doing your high from an optimized/and/or triangulated low, you're probably fine. I dont work exactly like you, I prefer to be more loose with my blockouts as I often change up details a lot on the high, and it would mean going back and making those same changes on the blockout when I do the low, I prefer not to do that. So my blockouts aren't as detailed, and just serve to get the main proportions. But the base concept is the same.
Often times the sort of topology I want for a blockout, high, and low are totally different, so it really doesn't save me time to get too detailed with my blockout in that case, as I'll end up remodeling either the low or the high anyway.
Dude! Didn't you see the sign outside? DON'T FEED THE TROLLS!
Yeah, i guess my crit was pretty bad on the "this is what you could have done" part. Sorry about that. Provoke and you will get bad critics
I think you latest HP looks better. The only thing that's actually bothers me is the clip release. The back part of it. You really can tell it's the same pieces. I think you should go back to ZB and try to make just that detail a bit sharper.
And yeah, as far as SubD (or in maya it's pretty much poly modeling with smooth preview. Push "3" on the keyboard) modeling, maya is as good as it as max is. You got some modeling tools in max that's cool, and especially the modifiers. But allot of it can be done in maya pretty much the same way as in max. If you looking for a program that handles modeling in first hand, you should look to modo.
The lowpoly mesh looks much better to! Is it going to be used for anything or is it just a show piece? I didn't really get it. Course if this is going ingame you got some extreme detailing in the handle that will never show up.
[EDIT] I missed out on about 7 post up, alot of what i wrote were already mentioned. Sorry about that (:
Hahaha he had to feed the troll! I didn't bother dignifying him with a response haha. As for the same piece comment, do you mean that the clip release button itself looks like it's part of the same mesh as the handle? It's a seperate subtool, but I thought perhaps you meant the slight glare of smoothing on the back side of it. I tried to point out what I think you are talking about
I'm not entirely sure what this is going to be used for, I was not informed, I am just providing a model as asked haha, I believe it's going to be in game as they want it to be fully animate-able, which I have done with a clip and bullet and all of the right parts internally that are visible - well at least i tried to
As for me I am going to use it as a show piece since I have done so much work on it, I am allowed to show it off since I have not been told the usage for it I am not under an NDA, it's sort of a cool position to be in!
Currently at 6567 Tri's, limit is 9000, i'm very happy with this
Laying out the UV's, baking and texturing up next!
Thanks again everyone for the feedback, good and bad, sltrOlsson I am happy I provoked because it got you talking rather than just skimming and closing
I mean on the other side of the handle, were the clip release "sticks out" when you press it. Not the clip release it self.
Haha 9k is a bit crazy, not really you concern though :P Maybe add some bevels and shit on what's going to be up in you face? Like the aim and stuff? I'm asking since, as i said, the handle have some details that wont even be visible to the player.
Anyways yeah i'm going to clean up it's normal map as it's pretty lumpy now, got a little clean up and a lot of add on to do to it now, sure i'll throw some bevels onto the sights, couldn't hurt right
Rendered in Marmoset
Some wires
My Maps, half size
It seams like you gun is the 3th generation and that it was made in both black and green for the handle. And from my quick search i found that the surface of the handle was made in rubber or plastic (or whatever that super awesome sci-fiy plastic is called).
So, if it's rubber, yo should go with a very broad highlight/gloss (dark) and if it's plastic you should have a fairly tight highlight (lighter), For the metal, as far as gloss goes, you want something in between, kinda.
If it's plastic, i think you want a pretty black and with spec map. For rubber i guess you want something closer to the diffuse color. And for the metal i would go with a more blueish color.
Since you are rendering in marmoset (though the engine you are making it for will handle it differently) it works good to put the scratches in the specular alone. And make them pretty light to, obviously
I wouldn't put that bullet in my gun. Would most likely explode or some shit in the gun. Remember, a bullet is only used once
The makes and shit came out good, nice cleanup of the normal map to!
As for the rest you have me slightly confused, are you suggesting I create a gloss map for it? If so I'll have to do some experimenting as I have only heard of them within the last few months.
It's a hard plastic, I don't believe there is any rubber in it, just a single plastic mold for the handle. The top of course is supposed to be metal though
Also here is what is probably the best ref for materials that I was going for on the weapon, and here is the ref for the clip material
And I toned down the bullet some
You might also try to lighten up all exposed egdes a bit. That especially goes for the specular, which is probably the map that needs most work in this piece. Just stain it lightly on the open surfaces, especially on the metal as thats a sponge for greasy fingers. You can also experiment with slight "rainbow" effects on those to get an oily look in places..
Apart from that the grain in the polymer is too fine, find higher resolution reference and be anal about nailing it :P
Racer 445 had this great tutorial on metal and guns in general. But hes site is still down. This is the general metal tutorial, but the gun tutorials were really super. One of them can be found over at game-artist.
You should really get in to the function of a gloss map. It will help you alot! The wiki tells you this. All you need to know, dark is broad highlight, light is tight higlight. For example, rubber would have a 60-70% black gloss map and a super shiny metal button (or you bullet) would have a close to white colored gloss map.
You specular needs allot of work to. The specular you have up feels like it's just a tinted diffuse with the AO. Take the handle as an example, a used gun would have the material slightly worn. But you grip is just genericly shaded. And so on
I'm far from saying this is perfect or good, just wanted to show you how i did my gun.
great work though, nice bake and this will look awesome textured- it has the perfect mix of matte painted, scratched and textured surfaces. Cant wait to see the end result
Thanks for the further detailed crits sltr, however i mentioned I was thinking it looked worse with the blue in my last shot for the slide, I think the bullet works just fine though. Either way I submitted to my client and they said it looks "excellent" (this was before I made my post here) so i am going to move on to the flashlight attachment, I will make a few adjustments that you have mentioned before I do but I can't spend forever on it just yet. I'll give gloss maps a shot once I finish the flashlight attachment, I don't think they (client) want a gloss map included though. The specular definitely is a tinted/tweaked diffuse texture with AO haha, from working on TF2 assets I learned that's how they created their specular's so I was giving that a shot. I'm not really sure this gun is supposed to be a highly used weapon or if it's supposed to be new, I wasn't given the circumstances. Still though like I said i'll make some changes and then I just have to move on.
EDIT: after watching that video I think it helped me learn a lot about spec, however your other links for tut's are all down including the one that is working, but his links are down and noone has re-upped, also thanks for letting me have a look at your gun, i was wanting to see your spec map so thanks a lot for that!
This post has confused me quite a bit, first off none of these shots are orthographic, at least not the finals - perhaps you mean the zbrush screenshots? in which case I can see why you would say that, I like them because they create a cleaner presentation layout haha. Most of all you have confused me as to why you are saying it will look cool when textured, it's almost as if you were only on the first page of this post and didn't see that there was a second page, and after realizing that's the most likely occurance while typing this, i'm going to assume that's what you have done haha.
This I can really take in
Do you have any tips on how to generate those rainbow effects? I'm guessing you mean like those iridescent purples and greens that show up from oil on that type of metal?
Finally the grain on the polymer, after being away for a while and having a second look i fully agree, it is too small, do you know of a way for photoshop to generate larger sized kernals when creating noise? My current technique is to fill the layer with black, then filter-> add noise, the size of it is always teeny even if you slide the slider up it just becomes more opaque but doesn't get much larger. I'll figure it out one way or another,
Thanks everyone for looking and commenting as always!
Also I enlarged the noise size to try to match the polymer better, after staring for so long i'm having a hard time telling if it's right or not.
Finally I realized what bbob meant about my clip and slide release needing some AO work so I tried to work that out as well and I think it looks quite nice.
Let me what you think, thanks!
I would do something more in this direction. Ofc this textures are far from good. They are actually pretty retarded, but still. And oh, it dosn't make any sense to have shading in you textures. Like the slide. Why is it light on top?
As i said, obviously this is a bit retarded, for example the bullet. And about that, when you do copper or what ever that is you should have a very dark color in the diffuse and "the color of the metal" in the specular. It will get you a very shiny effect with good highlights.
And yeah, you asked ho to make oil stains and i think you have a good start, just pump it up a bit. This is how i make my generic stainsines/hue shift's in my textures. I use this for pretty much everything.. Put it to overlay and rutn it down to 3-9%
Don't rush your textures! Just course the client says it's done and fine, it doesn't mean you should stop there. This is way to generic looking atm to be a good show piece, ofc my opinion. It's not interesting to look at and the material definition is poor.
Of course there are various parts that didn't get covered due to them being away from the main slide chunk but that's normal
Thanks so much for spending all this time helping me out, i'm not stopping with the textures, just got to meet the deadlines ya know, can't work on something forever. I'll definitely go back to it whenever I get the chance, as of now I have to make this deadline and then redo my entire portfolio website as I currently hate it and with good reason, I got an in depth critique on it outlining the problems I already had with it and I believe it's keeping employers far far away from me!
The slide was just light on top because that's how the ambient occlusion came out, I had the base as a brown because it was giving me results in the render that were similar to my ref shot which was a brown toned ref shot, I couldn't figure out how to make something black without it actually coming out as pitch black. For the gloss map (i'm assuming this is the third one) marmoset only has slots for specular or gloss maps but not both, and since that's what i'm using for my renders I figured I shouldn't stray from it.
For the oil stains that's a very cool psd you let me have and I can't thank you enough for it, I will certainly incorporate this in my texture once I get the chance to take another crack at it, which will be once I finish up the flashlight, i'll be working on the textures then because I'll have to make sure they match and since I want them both to look good, that's when i'll have a moment to really work on it!
Thanks again for all of your input and time, let me know if you ever need anything, i'm here for ya man!
But yeah, you can see what i would be aiming for. I think you should take it easy with the graininess to.
It takes time to get into the diffrent maps and textures/shaders and so on. That gun was the real serius use of gloss and reflection for a model for me, and it toke me a whole lot'a hours to get it right. I want to get back to it and change the textures though, but as you said. You need to move on
Marmoset uses sepculars alpha channel as gloss. That toke me awhile to figure out
Maybe you should try to do a small texure and apply it to a sphere in marmoset so that you can get a feel for what kinda color the diffrent things should have? (Fuck punctions!)
DDuckworth: For the bullet in your reference the reflections are very strong and glossy. in your low poly the relfections are weak and very diffused. I would darken the diffuse and make the reflections (both spec and environment) glossier.
Generally speaking high reflection needs to be balanced by lowering your diffuse.
@fletch: I'm a huge fan of Dara O'Briain and Dylan Moran, should explain it :P
Marmosets specular is using cubemap by default (:
I think its just a case of the cube being so blurred that its lost all detail. people seem to drop the gloss to tiny levels to try and get the specular blob to cover the surface and fill in the reflections, perhaps thats whats happening.
also you have great taste in comedy
Anyway its looking good keep up the good work
Thank you, I think that must be the solution to my problem, how do i make them "glossier" when not using a gloss map though? I'll throw it into the alpha channel of the spec map though when I get the chance thanks to what sltr told me. As for cubemap I didn't mess with any of those setting in marmoset, I just upped the brightness slightly of the enviro and used the default dawn environment lighting.
Haha no worries
Finished up the high for the flashlight attachment, trying out Earthquake's solution so hopefully he'll hop in here and let me know his opinion, realized I left out one screw on the right side of it, will add that in, the rest of the details I am going to do in photoshop using the nVidia plugin.
C&C Encouraged
Some ref pics
I think you should try to merge the rail part and the main flashlight part together to. It's all about the details
Basically just add some loops where it's suppose to extrude (intrude?
They are currently merged, perhaps I made the gap too large?
Though, some of the edges needs to be slightly looser. Won't bake very nice otherwise..
EDIT: blergh, olsson beat me to it..