Hey there!
This is my first attempt on subdivision modeling a weapon.
I know that an ak-47 is the most boring asset ever, but I had to do it. So, I threw a bunny to make up for it!
Since I'm not really experienced in this, and also happen to not be a gun nut, I'm looking for crits everywhere!
When I'm done fixing this high poly, I'll do a low around 5000tris and bake it down on xnormal.
Replies
this is really good for a first shot, 99% of the fps crowd would not be able to spot the errors.
-as said before, the charging handle is super thin
http://www.fototime.com/7C512E65F8DEFFA/orig.jpg - top view
http://www.surplusrifle.com/shooting2005/souptonuts2/graphics/l/19.jpg -front view
http://xssights.com/sights/pix/AK-Big-Dot-Express2.jpg - oblique
Having trouble getting you a view from the back but it's the same as front, except the curve in the surface is only 1d.
The forestock looks a bit off as well, but im not sure what refs youre working to..
But either way, this might just be something which annoys only me, but when people model the forestocks as solid with flat surfaces on the inside i find it really easy to tell once the bake is done, unless you artificially black it out in the texture which imo looks shoddy as well.
The real thing is hollow and you should be able to get a tiny glimpse of the barrel from some angles:
http://browningmgs.com/T2T3/Photos/Gastubes_T.jpg
The milled style reciever you have is unusual. The edges of the lightening cut should be parallel as in the below pics...
http://www.avtomats-in-action.com/media3/mfak4708.jpg
http://world.guns.ru/userfiles/images/assault/as01/ak53_2.jpg
http://www.avtomats-in-action.com/media3/mfak4702.jpg
http://www.avtomats-in-action.com/media3/mfak4707.jpg
And you might be able to notice that the milled recievers typically have a small bump where the magazine release housing is mounted.
The takedown lever looks a little bit too wide, so that itll be a bit too visible from FPS. They're quite thin and made from a single bit of bent sheet steel, as you can see here.
http://tantal.kalashnikov.guns.ru/variant/RSB_old.jpg
There are some details in the rear trunnion which are annoying to model but give quite a bit of extra detail from FPV.
http://www.3dcontentcentral.com/showmodels/CONTENTCENTRAL%5Crom_rear_trun%5Crom_rear_trun.JPG
http://picturearchive.auctionarms.com/9846113025/9052025/trunion4.jpg_thumbnail1.jpg (this one has a modern stock installed but the trunnion stays the same, modern stocks are built to work to the original trunnion)
Unfortunately, there's a fair bit of variation in how these look when installed, and youll need to pick one appropriate to what kind of AK you wanna make. Take a look at angles showing the upper surface of the stock where it connects to the reciever and youll see how much of the trunnion is visible from FPV.
Annnnnnnd front of the gas tube (though it wont be seen apart from third person) should have a consistent curve. A large number of models also have a curve running in the other direction:
http://browningmgs.com/T2T3/Photos/T2T3_barrels.jpg
http://browningmgs.com/T2T3/Photos/T2T3_GasBlocks.jpg
In general, if you want accuracy, it helps to identify a vague date you want it to fit into.
This site will help you do that. You can either go by which you think looks coolest or which is most appropriate for whatever game project it's meant to be put into. It might seem a bit crazy and autistic to pay this much attention to detail but there are many many variants of AK out there and picking a year (regardless of what year that is) will help you narrow down which refs are useful and which you should ignore.
http://browningmgs.com/T2T3/Kalashnikov.htm
I wish I was this good.. I mean wow.. seriously wow O_O
I'm almost done with the low poly, I think. Sits now at 6750 tris. Still don't know how I'm supposed to set smoothing groups on this... And I don't know how I'm going to properly capture the bevels in the normal map. I beveled the whole front trunnion but it makes the polycount skyrocket, there should be another way. But all I got in my tests were bad shading, weird edges and artifacts
This is with all of it set to 1 smoothing group - Okay bevels but larger surfaces look stupid:
And this, using smoothing groups - Large surfaces are fine but bevels look stupid:
And a close shot on the front trunnion, so you can see how crappy the bevels are:
So, any ideas? I guess I'm back to reading the normal mapping thread all over again, but i'd love if anyone could shed a light on this
(5533 tris)
and more bunnies because I know you guys like it!
There's a few AO problems, but I'll leave that for the end. Might not be apparent on the final thing, so no use bothering about it now.
Crit: Bunny needs more attitude...kinda like this
[ame]http://www.youtube.com/watch?v=9wncS6tO59M&feature=relmfu[/ame]
another view:
Also, one other thing, the front of the bit that diverts gas up just behiond the front sight, it has a rounded front, wheras yous looks flat, or even slightly indented. Here's what i mean:
Coundn't fine any better pics...
Loving the bake, wondering the same thing.
And here's a wireframe of it:
I did that because I noticed that, with everything set to 1 smoothing group, I'd get gradients all over the flat surfaces. Which makes sense, actually, but didn't render well in realtime.
Then, had to redo a bunch of the UVs. All these parts that were on the same UV island but with a different smoothing group had to be broken apart, to avoid normal map seams (it looks like crap)
Then, I exploded both the low and high poly, combined them into a single object (don't wanna be rendering a million normal maps and manually merging them in photoshop), made a copy of the low poly and set it all to 1 smoothing group. That copy was my baking cage.
Then, with the high poly model visible but frozen, I pushed the vertices from the cage outwards until it completely enveloped the high poly. Had to tweak manually in some spots to avoid the cage collapsing into itself. After that, RTT to a normal map in max.
BUT
All the floating details (rivets etc) got skewed projections. Since I didn't want to add supporting edges to the low poly and cages just to fix that, I made a second bake, but without the cage this time, using ray distances instead. Deleted everything but the floaters on the high and baked it. Then, baked an AO map in xNormal, from the unexploded models. I just pulled the magazine outside for a couple of inches so it wouldn't turn pitch black on the top part of it.
@mr_ace: You're right about the fire selector. There's a bunch of different models but mine doesn't quite match any of them. I'll try to tweak it without having to rework too much. Thanks!
About the gas block, there are many different shapes to it, my model is some kind of hybrid monster using parts of different models, because I had better reference of some part, or just quicker to model. And there's one that looks like mine, maybe a bit flatter, but anyway. Oh, and you caught my attention to the fire selector insets, they should be rounder and in a circular layout. Will do that as well.
I started blocking out some colors, also made a blockout spec and gloss, does this work? Wood will be varnished and chipped off in some parts btw, that's why it's shiny!
Oh, I started texturing by the dark metal. I did some edge wear but I dunno...
How about this wood color? It's supposed to be varnished, chipping off in some parts
(that's just a quick color test, I know it looks like crap)
I'm planning on making the wooden parts out of laminate, using this method:
Should give me some cool color variation on them. What do you guys think?
(shown here without specularity, haven't touched that ever since)
good bake man
First thing that comes to mind is the metal all looks the same.
It's high q don't get me wrong but it's just all the same looking, no deviation
not too late for some rails or maybe sling racks? add a tactical flavor maybe