My goal here is to build this rifle in its entirety. I'm building every piece within a minimum accuracy of 1/100th of an inch, which may sound absurd but when everything needs to fit perfectly, minor mistakes are very large. The visual goal here is to have a 100% accurate hi-res model of a complete rifle assembly which I can eventually bake down to a game res model to begin building a game portfolio, while the personal goal is to build some hard surface pieces that I would not normally come across in most scenarios. I am not leaving Any grooves, nicks, screw threading, or any notable (and specifically, intentional) plane changes out of the modeling. Things that would just be silly to model like caliber readings on the barrel will be disp maps because I have to draw the line somewhere.
My project can also be followed at both the following links:
http://eat3d.com/forum/work-progress/model-mp15-sport-riflehttp://forums.luxology.com/topic.aspx?f=9&t=69174
I've already learned a great deal and improved upon many of my hard surface techniques. I'm very excited to see the knowledge I acquire in the game res and texturing areas later on.
Progress to this is very slow going as it is a side project that I pick up when I have the availability. Please forgive the random downtimes.
To help improve my skills here I set myself a few rules to follow just for this project. The first is that this weapon system is modeled within a minimum of 1/100th of an inch of accuracy, but for the most part 1/1000th. The second rule is that I will not fake any geometry by having two floating/intersecting meshes give the illusion of a single mesh; that is to say, one piece in the physical world will be represented by one mesh in 3D. The last rule is that the HP model will be made with all quads, and whenever possible, evenly spaced.
Any and all comments and crits are very much welcome, especially when I get to the game res areas as I am someone unfamiliar with that territory.
3
Replies
Thanks Hang10,
The reason I originally began this project was simply to improve my HiPoly hard surface skills, and I needed more complex geometry to practice with. A gun was the logical choice to me because it is easy to study in person and there is plentiful online reference. So, you are correct, most of these details will not be seen, ever (although I considered doing a simpler breakdown akin to what the most recent Ghost Recon title did with the weapon custimization. Perhaps it could be a cool "hey, look how good gunsmith could be on the next console!" sort of a thing.
So, that explains that. heh. It was actually my friend Amadreaus (his alias on this site) who convinced me to do a game res model later. There will also be film quality firing cycle animations with cutaway views and full exploded view assembly animations (but to not bore the game-res-centric website here, I'll just be posting those on the eat3d and luxology forums).
Hey, thanks very much Havoc! I appreciate that. It's been a really fun and challenging project. Can't wait to have it behind me tho!
Also your rear backup sight is on backwards...
(LMT L8A Tactical Rear Sight)
Wow, thanks very much everyone. This is all very helpful. This will be my first detailed piece that will be baked down into a game res model, so as I have limited understanding of this process, this has all been very eye-opening. I appreciate everyone's feedback!
Hopefully the majority of the edge loops will be easy enough to loosen up without any major reworking, but overall this seems like a simple enough fix, and I'm glad I got this advice now rather than later.
Thanks kaburan! Yes, I agree, of course in a production pipeline, this would be an incredible waste of time. But, for the purpose of the project (simply to improve modeling skills and techniques), I have already seen many great improvements in my modeling since I began, so it's been a good ride. Really appreciate the feedback!
Thanks very much for this! I found this visual you added the most helpful. Much appreciated
Haha! I can't believe that, that's funny to me that I have so many pieces lined up perfectly and am paying such close attention to detail, and yet I'll still just flip an entire piece around carelessly. heh. Thanks for showing me this.
Thanks for pointing me to this thread, this was a really impressive example! Once I get to it I'll start loosening edges and trying various bakes until I can get that right look. Good call.
hahaha! Awesome quote. Thank you.
hah, indeed, indeed. I wish I had a 3D printer so bad!
Note: I also included an image with a cutaway view to show just how precise both the screw and the inverse itself are. They are perfectly matched with this new method.
The receiver continues to be very rude, but I'm slowly chipping away at it. The accuracy and all quads and no faked geometry rule is very tempting (not enough tho) to break on this piece, because the topology is quite frustrating. Anyway, as you can see, it's somewhere near the 40-45% mark. What IS done tho is the magazine box, the mag-floor plate, and soon the mag-spring.
Thanks so much, Sir Great! Well, I Woouuld like to put it in a game one day (only the surface pieces, of course) but I learned here that I'll need to rework the model a lot before that's possible. So, after I've finished the film-res stuff and the animations, I'll come back through and loosen up the appropriate edges to see if I can't learn how to make a proper game model. :-P Time will tell! haha
heh, well, I sure hope you're right about that! I appreciate the encouraging words tho, buddy. I was slightly disheartened to hear about the switch to game models I wasn't expecting. But, it's all just more to learn!!and that part I love, so it's all gravy. =P
I'm trying the difference between different transparencies and cutaway views to get more of a feel for how to present this. Anyone have any opinions between any of the following? Better suggestions in general? All thoughts and crits are very welcome.
To Fridock: sure! It's currently comprised of 81 pieces that total to an (almost) all quads polycount of 299,311 verts or 298,225 polys. I say "almost" all quads just because I haven't cleaned up the mesh yet so I know there will be some tris and ngons hiding out in a couple places (mostly the earlier meshes of the project, I haven't allowed for that for awhile now). Now, of course, it's clearly a subD model, so after it divides just one time it's already looking at approximately 1,200,000 or so. I think for these test/progress renders, I've had the rendering subdividing two times (all at render-time, of course).
Hope that helps answer the question.