MS type AI Progress

[Bullethead 12]

Sometimes I get strange artifacts on screen, like past remnants of what I've just moved/deleted/hidden. Zooming in & out or panning the view a touch will remove them to show the present condition.

 

Yup, but this apparent z-buffer issue in "User" never goes away there, unless you have nothing overlapping at all.

 

"Perspective" and "User" do things quite differently. In "Perspective", when you rotate the view, the pivot point is like 1/2way between you and the object so when you're up close, it appears to move the object left-right more than rotate it. And when you zoom in , everything closer than "arm's length" to your POV disappears.

 

In "User" OTOH, rotating the view happens about the center of the selected object and when you zoom in, you can never get closer than its outer surface, and nothing disappears.

 

It also appears that "User" and "Perspective" do the z-buffer differently. "User" has horrible problems with it when you have polys that overlap or are even just close together. I suppose this is helpful for finding problems, but because only "User" has the useful pan and zoom features, this can be a real problem when you stick 1 object through another to make a cut.

[ConradB 0]

Also of note on the user view, it does some odd things to the plane perspectively so that may also play into the oddities noted. It appears in some respects to distort the shape, like shrinking things that are closer and enlrging things further away. Kinda the same way photographs of planes will distort depending on how close you are to the subject you're photographing.

 

I like the regular perspective view for after I'm done with a part of the plane. It is a good way to get a proper view perspectively of the plane with the additions.

[Bullethead 12]

A bit more progress.... Wing and strutus. <BR><BR>For a monoplane, this thing sure has a confusing tangle of struts. I had to scrounge a bunch of photos of the real thing and RC and plastic models to figure it out; 3-views just can't protray them in all their intricacy. I haven't trimmed any of them to length yet, so there's some Gmax overlapping in this pic that makes things even more confusing.<BR><BR>In case you can't tell, this thing's got a pair of long, angled struts from the lower fuselage to far out on the wing. At their midpoints they are connected by a horizontal rod and from there smaller struts angle back toward the center. Their top ends meet the tops of the outer N-shaped cabane struts. There are are inner cabane struts that lean inboard a little. In front is an inverted V and in the rear is just a lone vertical post. The blue thing threaded through the cabanes is a Nieuport-type aileron pushrod.

Edited February 7, 2010 by Bullethead

[hairyspin 1]

And a nicely undercurved wing. I hope you haven't gone overboard on polys for that, but it's looking good!

[Bullethead 12]

And a nicely undercurved wing. I hope you haven't gone overboard on polys for that, but it's looking good!

 

Thanks :).

 

As to polys, I think it's considerably leaner than the more recent OBD planes. The struts are all just 6-sided with 5 segments, except for the front long one, which has a few more due to a notch for the crossbar. The main part of the wing is only 20 or so polys: 14 forming the airfoil shape that run from centerline to the outboard end, then 3 or 4 more each in the cut-outs for the aileron and over the cockpit. Smoothing groups are cool . The outer ends of the aileron and wingtip objects naturaly have more due to having to curve around the tip and taper in thickness while following the airfoil, but it's not that many. A recent OBD plane would have 3D ribs molded into the wing, all sorts of fittings and clevises where struts join wing, etc. I left all those off.

 

I decided not to do 3D ribs more for aesthetic reasons than for polys. IMHO, while 3D ribs look good in the LOD100 model, they're not in other LODs and that causes a problem. The problem is, all LODs use the same texture, or at least a MIP of it. Anyway, if the LOD100 model has built-in ribs, then skinning it with 2D fake ribs on the texture looks like crap because it unrealisticaly exaggerates the ribs. But if you don't put rib shadows on the skin to make the LOD100 look good, then the wing is utterly flat and without ribs just a few yards away. So, IMHO, the better thing is to make the wing flat and do the ribs with the skin, so they look good at all LODs.

[ConradB 0]

Hey BH, I don't worry much about making 3D ribs on the planes, especially if the leading edge had any plywood sheeting up forwards anyways. Looking at pics of planes and seeing them in museums, the fabric didn't sag like the scalloped effect between the ribs anyways. And the ones that did like the SE5, weren't that strong anyways. Most of the time, even the JN4 that used to be at the old Stapelton had the fabric flat between the ribs, and the only way you knew exactly where the ribs were, was from the stitching, and or rib-tapes anyways.

 

Now in flight, you will notice from time to time the fabric ballooning or sucking in on the ribs, but that is from pressure changes. And until we get uber 'puters that can handle that kind of graphics, it's like building a stock plastic model out of the box, with no enhanced details. You can see it on some of the videos where there is a camera mounted on a strut, and you can see the wing fabric billow, and flatten, especially in turns where wing deflection and air pressure cause it to happen. It would be like one of the effects of diving to fast, you should see your wing and tail fabric balloon from the high speed dive.

[Bullethead 12]

Looking at pics of planes and seeing them in museums, the fabric didn't sag like the scalloped effect between the ribs anyways. And the ones that did like the SE5, weren't that strong anyways. Most of the time, even the JN4 that used to be at the old Stapelton had the fabric flat between the ribs, and the only way you knew exactly where the ribs were, was from the stitching, and or rib-tapes anyways.

 

I agree. They doped the fabric specifically to stretch it taut, so it wouldn't flail around and disintegrate. Thus, there was just a very subtle depression between the ribs and mostly all you saw was the tapes, unless the light was nearly parelle to the surface. That's another reason I decided not to do 3D ribs.

 

Now on a fuselage like this one, with fabric stretched over stringers, you do get prominent lines there. That's why I didn't put smoothing groups on my fuselage, except longitudinally between the stringers.

 

One of these days, I'll do a plane with a scalloped trailing edge. Then there's no choice but to slice up the wing.

[ConradB 0]

I agree. They doped the fabric specifically to stretch it taut, so it wouldn't flail around and disintegrate. Thus, there was just a very subtle depression between the ribs and mostly all you saw was the tapes, unless the light was nearly parelle to the surface. That's another reason I decided not to do 3D ribs.

 

Now on a fuselage like this one, with fabric stretched over stringers, you do get prominent lines there. That's why I didn't put smoothing groups on my fuselage, except longitudinally between the stringers.

 

One of these days, I'll do a plane with a scalloped trailing edge. Then there's no choice but to slice up the wing.

 

 

Just make a cookie cutter, and do lots of slicing. Then close the gaps. I suppose that would be the best way to get that done.

 

I was thinking of messin' around with the Juckers CL monoplane. That would be interesting to simulate the corregated duralumin that was a trademark of Junkers planes until the end WWII. There were a couple attached to most 2 seater squadrons.

 

There is adifference though in the way the fabric goes over the wings, than it does over the stringers on a tubular fuselage. Same thing for the Hawker Hurricane. The fabric over the rear fuselage wasn't so much scalloped, but created flat areas between the stringers. The same thing wold have happenned to the wings if they had stringers running spanwise acroos the top and bottom of the wing. But since the ribs have a gentle curving to them, the fabric is going to follow the contour by itself.

Edited February 8, 2010 by ConradB

[Bullethead 12]

Just make a cookie cutter, and do lots of slicing. Then close the gaps. I suppose that would be the best way to get that done.

 

That would make the trailing edge too thick between the ribs. The reason some planes had scalloped trailing edges was because the trailing edge wasn't rigid--it was a length of cable. The scallops were created when the fabric shrank from the doping and pulled the cable inboard between the ends of the ribs.

 

So, in real life, the trailing edge was the same minimal thickness all along the wing, but portions of it are moved forward. To model that, I think you'd have to build the wing with a straight trailing edge, then divide at least the TE into 3 or 4 segments between each pair of ribs. Then you'd move the TE vertices forward (and no doubt upwards) different amounts to get the curve of the scallop. This will of course change the slope of the rest of the wing surface. I'd expect that unless you had the TE cuts run the full cord, you'd have an ugly mess of shadows on the surface, and these would naturally lead you to make 3D ribs by default, to get a better looking surface. Hmmm........ Maybe the best way would be to make 1 section of wing between 2 ribs and clone and weld it all down the wing, assuming constant cord.

 

I was thinking of messin' around with the Juckers CL monoplane. That would be interesting to simulate the corregated duralumin that was a trademark of Junkers planes until the end WWII.

 

That's something I'd do only with the texture. WAY too many polys doing it 3D I'd think.

[ConradB 0]

That was my thinking too. Doing the contoured metal with the texture. Don't have powerful enough rigs to run that kind of monster.

 

With scalloped trailing edges, don't over do it. Look at some of the real planes. They aren't that prominent. And yes the TE doing that was from the use of "wire" as they called it for the trailing edge. My Albatros DVa model used very thin piano wire epoxied in for the trailing edge, and the doped silk actually pulled the scallope in, but it isn't very thick at that point. you would have vertices so close together, it wouldn't be very noticable. Even if you were to tackle something like the SPAD 2 seater that came up in a post a while ago. Also, because of the undercamber most of the planes had in those days, the top vertice would only need to be moved down to the bottom one. The tough part would be if the wing incorporated washout at the tips, or at the wingroot, to flair the wing into the fuselage.

 

On the DVa model, you could cheat a little, after the silk was attached to the frames with dope, you would use a spray bottle with water to lightly dampen the fabric, then the best thing was to set in direct sunlight to dry it, as it shrunk the silk tight as a drum. Apply 2 coats of clear dope thinned 50 / 50 with thinner, then apply coats of sanding sealer, wet dry sanding between coats until the pores were filled. Then the color coats would be applied.

 

Kinda sucks as it is difficult to find the old materials to do it that way anymore.

[hairyspin 1]

I was thinking of messin' around with the Juckers CL monoplane. That would be interesting to simulate the corregated duralumin that was a trademark of Junkers planes until the end WWII. There were a couple attached to most 2 seater squadrons.

 

Someone (I think Greg Law) built a Ju52 transport with the corrugate fully modelled, but when turning it into a CFS3 model he used the original to generate the skin textures - too many polys for the game otherwise - and mapped them on to a smooth model. It's one of the AvH downloads.

[ConradB 0]

Stop it Hairyspin! You're giving me ideas that will probably wind up putting me in the booby hatch!

 

But I do like the idea. I suppose all one would have to do, is make a large single panel, then clone and turn where needed, then cut to shape. The wings could be done by slicing the panel, then just wrap the vertices accordongly. Same for the tail feathers.

 

OOOOOOOOOOOOOOOOOOOOOOO is that an idea.

 

But I have to finish the Halberstadt CLII first.

Edited February 9, 2010 by ConradB

[Bullethead 12]

Tail feathers....

 

There's a small ventral fin but I'm going to do that as part of the tailskid when I get around to doing the landing gear.

[ConradB 0]

Lookin' good there Bh! Lookin' good!

[Bullethead 12]

Lookin' good there Bh! Lookin' good!

 

Thanks. You been doing anything lately?

 

I'm currently taking a break from pushing polys to learn how to texture things. I want to texture all the damn struts before I trim them to length. Texturing is painfully unintuitive but I have succeeded, after a fashion, in putting 1 piece of skin on 1 part. Now I need to figure out how to get that same piece of skin on a bunch of identical parts, which is something the tutorials don't seem to go into.

 

BTW, thanks to all the lurkers for their interest. I can't believe a thread with barely 30 posts about an obscure plane being built by a complete n00b has garnered over 1500 views. I hope to repay your interest someday with something you might want to fly.

[ConradB 0]

Been busy trying to assimilate the SDK. Hairyspin tutored me well, like a Jedi Apprentice. And soon, I will be the master. Sorry, couldn't help myself. I went through the Halb CLII, and really reduced the polys, as I was without a compass on what a good average should be, but HS put me right on it. Though it was fun seeing just how realistic I could make it look, and then also some issues with the wingroot and how the polys would get all snarled up there causing smoothing issues. Thankfully, the correction was quite simple. So with the plane as she sits, and this is without all the details of course, is right around 2000 polys. That's 2 wings, fuselage, spinner, tail feathers, wheels, and the interior walls for the cockpit and engine bay. But I still have to build the cockpit innerds, and pieces parts, and the guns, and a low poly Mercedes DIII. But I am going to make a base block with cylinders and save that as is before detailing it out, so I have a base for the BMW 185hp. inline, as it is close enough that I can make a few adjustments and have the basics worked out, and just deal with the plumbing and all for the details.

 

One thing I am trying to figure out, is how to represent the elastic cord that was used for shock absorbing for the axle. So I'm wondering how to get that part sorted out, as I really would like to avoid just putting a blob or rectangle on the gear. I want to make a second variation of the plane as it did get issued with a coil spring setup when rubber got scarce. It's an easy mod, as the axle had a fairing over it and the braces that support the gear horizontally. Cool thing is, the axlw doesn't splay as it is a straight axle.

[Olham 164]

Only just found my way here through your post in "General Discussion", Bullethead.

What you're building there, looks so amazing, that I want to try 3D stuff again.

 

How long would you say a semi-intelligent bloke would need to get as far as this?

Do you use Gmax or also other programs?

[Bullethead 12]

How long would you say a semi-intelligent bloke would need to get as far as this?

Do you use Gmax or also other programs?

 

It shouldn't take you very long. The hard part is learning how to use Gmax. While the tutorials give you the basics, building airplanes requires specialized techniques that you can only learn by asking the experts like Stumpjumper and Hairyspin, plus they also know shortcuts that speed things up. You can see all sorts of great info they've given me in this thread plus others.

 

I started this airplane in early December and have, due to time constraints and waiting for answers to questions about how to proceed, probably only worked on it an average of 1 day per week in all that time. So say 12 days total. In that time, I have started over a couple of times and rebuilt a number of the major pieces several times each as I've learned better ways of doing them. Despite all this, I have the exterior LOD100 airframe complete except for the landing gear and bracing wires, and have completed and textured the engine. So the actual building of the airplane doesn't take much time--what takes the time is learning how to use Gmax. But once you know which commands to use for what jobs, the airplane takes shape very quickly.

 

I should point out that while this MS AI is my 1st "real" airplane, as in something I hope to fly in the game someday, I have previously built a few other things for practice. This wasn't really good practice because I didn't know what I was doing, so if I ever return to building those planes I'll start them over from scratch. However, at least that experience taught me enough about Gmax to enable me to ask meaningful questions of the experts.

[Olham 164]

So where would you recommend to start reading/studying?

[Bullethead 12]

So where would you recommend to start reading/studying?

 

Start with the official Gmax tutorials available at TurboSquid, where you get Gmax itself. This gives you a good idea of the basic interface.

 

Then get the Bitmap to Bullets tutorial linked in one of the stickies above and all things that that tutorial links to. Also get the Cesena 162 tutorial in another sticky above, and all it links to. The B2B tutorial teaches very little about using Gmax itself, but tells you everything you need to make and how to hook it all together to get a flyable plane in the end. Do it first because then you'll appreciate the scope of the task, of which building the external LOD100 model is just the tip of the iceberg. If you're not intimidated by that, read and perhaps do all the others.

 

Having done that, pick an airplane (not necessarily from WW1), something that you're a little bit interested in but don't care that much about, and try to build it using what you know so far. The result will be crap (like my Bristol M1c) but you'll LEARN a lot doing it. After numerous restarts, you'll get to the point where you think you're doing it pretty good to start with, but when you get towards the end you'll think it's crap and want to start over yet again. At that point, you've probably gotten as much out of it as you can. Then perhaps do a "real" plane, something you're very interested in and think OFF needs, or start another test subject. And all the while, read threads here and ask the experts for help.