KT-s747 (ZBrush Pistol)
polycounter lvl 10
Hi, First post here on polycount. Was hoping to get some opinions and critiques on my 3d work.
This was made entirely in zbrush 4 R2, With the exception of the lettering on the grip.
(correction) 17Million points. 32 million polies.
35 subtools
3 days(on and off)
Thanks in advance.
This was made entirely in zbrush 4 R2, With the exception of the lettering on the grip.
(correction) 17Million points. 32 million polies.
35 subtools
3 days(on and off)
Thanks in advance.
Replies
how did the work flow by comparison to working in traditional high poly workflow in max or maya? do you have much experience in those channels, or are you more of a zbrush guy and this is your first foray into hard surface, so you're doing it the zb way?
i think a lot of the best guys in hard surface work are going to be very reluctant to try this methodology given their high level work in traditional methods and the likelihood that translating the desired results in zbrush is going to be quite a paradigm shift.
that said, i am staying open to it because i think it has the potential to be revolutionary in workflow methodology. more than likely it's going to provide an alternate workflow based largely on program familiarity (an artist most comfortable in zbrush rather than max would probably prefer it?)
as for the model, from what i can see it looks good, but i can't see very much of it. could you please repost with lighter levels so we can see more, also higher resolution to better gauge detail?
i think the model also suffers from being flat black on a black background, try giving it a more interesting color profile. quick photoshop to illustrate:
so anyway, it looks pretty good, post more and better shots, and let's get to talking about modeling weapons in zbrush, because i know a lot of people around here are very curious about it
As a guy who does a ton of weapons professionally I'd love any infromation you can provide on workflow. Are you using shadowbox to build the major subtool forms, or just carving them out of primitives directly?
How animatable is this? Does the slide rack back and forth? I'm not familiar enough with zbrush to know how you'd test mechanics like that quickly.
As for the gun itself, the concept is pretty awesome, although I think the rail on the bottom has its grooves too close together, both from a practical point I believe those are usually wider, and from a strictly design point they add a lot of high-frequency noise on part of the gun that doesn't really want it. The handle is a bit blobby, too, I'd have liked to see the handle have a bit more bulk in the back and a bit more defintion to the bottom.
Overall it looks aces, though (at least in gauss' lightened version).
To start from the top going down;
I started modeling out of boredom and curiosity in 3ds max. Modeling vert by vert never really clicked for me, so i never got very far into learning max. Not too long after a friend told me about Zbrush, and once i tried it, i was sold.
It was less restrictive, more intuitive, and more complementary to the kind of person that creates things on the fly. This gun, For example, I started as a blank canvas, not knowing what i was going to make, then created the grip, and worked my way up. Constantly changing things as i went. Max isn't very helpful when it comes to that kind of approach in my experience.
As for the potential shift from tradition packages like those from Autodesk to those more un traditional like zbrush, I'm optimistic. Just from zbrush 4 to 4R2 there has been some major game changing features. Such major features that if they keep coming, they will soon make 3ds max obsolete in my opinion. For example, Dynamesh. I dont know if your familiar with it, but it essentially gives you the power to add and subtract geometry from or to and existing piece of geometry. This makes the creation of complex shapes and hard surface pieces in particular extremely fast and useful. Pieces like the mount for this scope were made in a matter of minutes. One solid piece of geometry without "hard" floating geometry seams, that I have personally grown to dislike.
Now i dont know how fast it would take someone fluent in max or Maya to create that, But Id be surprised if it could be done as fast as in zbrush.
As far as Im concerned, the only thing max and traditional packages has going for them is the "Cleanness" of edges. In those programs the edges are exactly the way you made them. In zbrush, while you have a ton of control over edges, the results are often varied and inconsistent.
As computers keep becoming more powerful and more powerful, the poly restrictions are going to get less and less. I think there will be a point were the poly limits (as far as high poly modeling) will virtually disappear. When this happens do you think people with be modeling vert by vert in max? or creating extremely complex and detailed models in zbrush?
So in short, yes, one day i believe Programs like zbrush will be dominant over programs like 3ds max.
And Ill fix the trigger
@Ghostscape: Thank you for the advice. I love how people dont just say "It looks good" here
For how i create the organic hard surfaces, its a combination of shadow box for shape, Sculpting and smoothing form into it, clipping brushes, and the biggest contributor is dynamesh. How I work all these together would be a bit hard to explain in text alone. I can create a walkthrough of how i did it though.
And depending on what the object Im trying to create will determine what i use to start is. The scope, I started with a cylinder, hollowed it out with dynamesh. Duplicated it, shadowbox for the piece in the center, clipped for shape edges then use deformation to "Bubble" out the ends(Gravity on the z axis, To be specific). Then use subdivided primitives for the little nobs. I masked and deformed (inflated) for the grooves in them.
I will change that lower rail asap. I want to change the grip to be more "Lending" to having a magazine within it, And like you said sharpen the bottom. Not sure when i will do this though. its going to be a lot of reworking.
As for how animatable it is, I couldn't say :P I can move the slide to the back postion and everything still looks sound. Its going to be for a game project me and buddy are working on, So i'd hope we can animate it
I need to remake the slide, The sides got way to thin somwhere along the way.
Thanks again guys. I can't tell you how refreshing it is to get some real advice.
:thumbup:
Gotta get more RAM!
I've taken a few pictures of my FNP .45cal pistol to help give you more of an idea of what I'm refereing to. Apologies for the low quality, I only have my Iphone
and Alec, I can create a smoothed edge anywhere its needed. Just run the smoothing brush over any edge and it will soften. I'm guessing the slide was one of the most noticable? :
That i just masked what i wanted to stay sharp, blured the mask, And ran "Polish crisp edge" deformation. Theres not really and edge you couldn't get from the right combo of Cliping, Masking, Dynamesh, and deformation. Just after seeing so many "Hard surface" zbrush models that were all pillow like, I prefered make very hard sharp surfaces :P Guess i can get to carried away with them.
And josh, The reason all the alignment for the internal fuction is off, Is because the gun used to look like this(Below) and then i decided to scrap that and slamm it down quick. Didn't pay much atention to fuction very much
As the OP says, I wouldn't be so quick to assume on the smoothing point. It is is undeniably a different workflow, but there are a crazy host of options for tweaking meshes (creasing/smoothing etc) in rollouts, which combined with zb's ability to handle a lot of shapes quickly that would be time consuming for even veteran high poly modelers, it's nothing to dismiss easily.
And the most consistent way I've found is to start with a dimensionally correct round of your chosen caliber, and then make sure that the rounds all fit in the magazine (don't need to get obsessively detailed here for parts unseen, but it helps make sure the logic of the weapon is intact) and that the bullet is pretty close to the diameter of the barrel.
JoshC's FNP pics are hugely helpful here, because you're looking at a big fat round (.45 ACP) laid against the pistol frame, and you get a sense of the scale of his hand relative as well.
So it might do you good just to figure out exactly what round it's firing (your first pics look like FN 5.7
http://s620ex1.deviantart.com/art/Zbrush-Pistol-grip-Tutorial-283825933
http://s620ex1.deviantart.com/#/d4ozdhp
Will work some more after errands and dinner.
MORE!!!
I think i will make at least two version of every gun. Its easy enough to have one with all the attachments, Being a gun metal black "Authority" version, Then a stripped down "locals" version with another color scheme. Then maybe a "Rebel" version that merges the two. WIth some cammo and stuff.
If neither of these look right, I can do anything in between really. Could clip then smooth. Clip very little so i get a hard edge into a small curve. Lots of options.
Edit
earthquake/racer445 talk a lot about handling edge hardness. i would take time to really browse all the gun models they did for brink, which were exaggerated in aspects but the edges/shape reads were perfect basically no matter how stylized the game it's for is. i am continually amazed at how soft some of the edges are on their high polys, but it ends up looking like it ought to as the finished product, and in the end that's the only thing that matters.
also be sure to check out the "how do you model dem shapes" and some of the related discussions in technical talk, if you haven't already.
looking good!
Other then that, great progress.
EDIT: better showing the difference between the last two variants:
here i've recreated a section of your model with edges similar to yours, then one i thickened a fair bit, then one i really spent some effort on to get the falloff to look str8 obese. in the top right closeup you can see that the "fat" bevel has a really thick falloff, thus grabbing even more light than the "good" example. so though all of the examples look technically fine up close, the good amd fat examples catch the most light without being "soft" as people tend to suggest. this of course makes them much more attractive to look at, especially when the entire model has a consistent edge treatment.
the visual difference between the last two isn't big and doesn't matter too much, but if you can do it, why not?
you can also see that while they all look technically fine up close, as you zoom out, the specular highlights start to alias into oblivion. this is more apparent in the popular example image i made a while ago (http://dl.dropbox.com/u/1725586/crits/edgewidthmodel.jpg) which does not have any form of anti aliasing like this new one. in that image you can really see how zooming effects this.
this however doesn't stop me from being interested in this zbrush stuff you're doing. i think it's really cool, and has the potential to be insanely fast, but i'm concerned about the control you have over edges, as well as changes to shapes for clients. how is that stuff handled? is it pretty easy to make changes like traditional subd? that's really what it all comes down to, as i want changes for clients to take as little time as possible, even at the expense of a slightly longer initial creation time.
oh also you need to use simpler materials to present wips. a simple dark/midtone slightly tinted diffuse with a slightly colored bright, low gloss specular highlight is best.
Here is a little example picture of a harsh clipped edge, a "Dynamesh" edge(just happened, Wasn't really made), and an edge i masked, Blured then deformed a bit.
I'm not arguing that zbrush can control edges as well as max, cause it can't. And it probably never will. But you still have a lot of options for zbrush.
For editing shapes on the fly, It depends what shape really, and how. Zbrush can create shapes extremly fast, So in some cases it would be more effective to redo the shape entirely. If it was a proportion change, Between masking, Deformation, and dynamesh there isnt really a proportion you cant change and retain even topology.
Other than those, zbrush can still hang with changing things on the fly, Just not as cleanly as max or maya i think.
Keep in mind here, I'm 16 and have been using zbrush for about half a year. So i could know little to nothing on the subject.
Again, Great advice. I seeing that slide really makes me want to get into max
Out of curiousity, If you even come back to this thread, How long did that take you? and could i see some wires of it?
Racer, you really should check out the eat3d hard surface 2 DVD. Their method is good because you pretty much have all the controll you need over edge obesity since you usually start by concepting and blocking out the mesh with dirty/ugly hard surface brushes and dynamesh booleans, then you retopo it right inside of zbrush. now that you can use creasing in zbrush, you controll the edge obesity as you create the topology. OPs method if using smoothing brushes isn't ideal.
Oh yeah, Forgot all about retopo. Here is a thread for a zbrush mech that was part of zbrush 4 R2 beta. he goes over retop on a hard surface.
http://www.zbrushcentral.com/showthread.php?161185-ZBrush-4R2-Beta-Testing-By-Nicolas-Garilhe&p=892733&viewfull=1#post892733
and to clairify, I never use smoothing brushes to fix an edge. i hardly ever use any brushes, in fact. too in consistent. Was just saying it was a posibility
Uploaded with ImageShack.us
Only down side to this really, is zbrush crashes A LOT when retopologizing. It trained me well to save my work often. If i can keep it from crashing long enough, i can try and make some example of potential edges to get from that LP base mesh.
i think that's really cool how it just works like that! could you try and match the shape with edges in similar width to what i've done and see how long it takes? i'm interested to see what you can come up with. i'm also interested in the tricount of the model you end up with, since there's no point in shortening the modeling time if it's going to eat 16gb ram and take forever during the bake process.
computron, don't get me wrong; i'm not in any way trying to hate on this method. rather, i'm quite curious into the details about it. it seems like something that could be very useful.
you think i actually saved this model?
I coulnd't do three variations in 10 minutes, Thats for sure ! I'm having a hell of time making an example out of a square. And to clarify, when i said "you can treat it just like a model in max", I meant you can add edge loops. The way you add them is much, Much, Slower. But i should be able to come up with a few examples. :thumbup:
and just had a thought. If you made a concept model in zbrush, ( like the original slide i made) then retopologize it, Like the pictures i just posted, Then went to max for edge control, That would probably be a pretty nice work flow. I think i'm going to try that sometime very soon
...Even with this square its taking me a while to make the loops, and zbrush keeps crashing
Instead of making accturate loops you could make rough loops where you want edge control, Sub divide a few times without smooth on , Then sub divide with smooth, and probably get a decent result. again taking a while to get something so close to what you already had.
Noob max question here, What needs to happen for me to apply normal maps to a plain shader? ...
Getting back to the point, Retopologizing a mesh then adding really general edge loops would probably be possible and even effective given a stable zbrush(which i dont have).
Anywho For what relevance it has:
Uploaded with ImageShack.us
Anyway, This was made in about 20 minutes. Thats mesh variations, Unrwrap(in max), and then subdivide and bake in zbrush. Screen grabbed from UDK, Sense i had no other way of getting normals on the model.
Original HP:
Uploaded with ImageShack.us
Retopologized(Left side was finished, Then mirrored to the right)
Uploaded with ImageShack.us
Edge vartions. Keep in mind at this point, The subdivide depend on how i retopologized. So if i wanted sighter edges with less polies, I could have made the edge loops closer to edges. and less loops.
Uploaded with ImageShack.us
LP In UDK with normal variation
Uploaded with ImageShack.us
NONONO!!! The point was to re-topo it into a new sub-d (all quads) mesh and use creasing instead of control loops. not make a lo poly with super hard edges.
I didn't take it that way.:)
I am just as curious about adding zBrush to my hard surface arsenal as you.
OP was just doing it in a way that requires a lot of manual work and ends up with varying edge-width. the retopo method from the eat3d dvd is very different from what's being shown, I recommend you take a look at that dvd or any ryan kingslien hard surface DVD. If you know your way around zBrush you can very quickly pick it up.
Not sure if you missed the post right above your own, But i beleive thats what your talking about?
and normals, if they are relevant.
Uploaded with ImageShack.us
Also, My way requires a lot less work than making everything i make, Then an aditional Retopologization and edge control pass. Though, Given consistent results, I may add that into my workflow :thumbup:
Also, I've yet to find out what the creasing fucntion is used for in zbrush. If you could infrom me, that would be great. If it helps with hard surface modeling, I would love to learn it
What you got is pretty good for the result, but it can be easier to retopo certain pieces into true sub-d meshes berofre dynameshing and polishing.
Uploaded with ImageShack.us
Uploaded with ImageShack.us
Also, i totally forgot about zbrushs edge loop functions. You could make a retopologize version in quads, Then Just hide everything but your edges and add edge loops with the click of a button. I might try an example or two on that slide. If things go well, I'll try and match racers 3 and see how long it takes me.
By s620 at 2012-02-12
By s620 at 2012-02-12
By s620 at 2012-02-12
By s620 at 2012-02-12
One of these days I really need to get into Zbrush.
Here is an example of the sights:
Uploaded with ImageShack.us
Gun question: Does the red dot appear to hover in front of the scope? Is it usually adjustable? How far ahead can it appear?
For the bake, it may be terrible the first few times
computron: the dot is invisible until you look through the sight within a few degrees of having your eye directly behind the sight. this is a tremendous advantage to old timey laser pointers, which are visible out in the world; a red dot sight is invisible to everyone else.
weirdboy: generally speaking, faster target acquisition, and they're just plain easier to shoot for most people. instead of lining up your irons with the target, which might involve actually covering the target with the sights depending on range and how the sights have been calibrated. some guns are set such that you should "float the bull," which is to say, your target and where the bullet goes is actually above the sights, not right on top of/behind the front sight.
a red dot gives a more open "sight picture" as it's called, which is simply how you see your target relative to your gun.
Since red dots and holographic sights have an "open" sight picture they make for great rapid target engagements especially in close quarters and even mid range distances.
Acogs and other 4x scopes are more for the mid range and long distance shooter but can be used in the short range area with much practice.
By s620 at 2012-02-14
By s620 at 2012-02-14
By s620 at 2012-02-14