How to draw aircraft

[Jemiba]

"If you can set up a camera in the correct position and with a similar virtual lens,
then you can render out what should be an identical view of the object and check."
That's more or less the method, i'm using, too. Not easy and not fast, but mostly
with reasonable results ... mostly !
Regarding my problem, I've forgotten one point, that I already knew and that spoils
the result : As most published photos don't show the complete photo, but just a
cut-out, you never know, where the center point is.
Thanks for mentioning your experiences with graphic tablets, some more arguments
for thoughts about such devices !

 

[prolific1]

I'm an old fashion 3D. My software is pencil, paper, and an eraser...electric eraser that is. I do need to learn how to do this in Illustrator or Photoshop. It's cleaner. Are there any tutorials for PS that show how to draw with paths? I'm only good at photo manip and coloring in PS.

 

[overscan (PaulMM)]

My two cents...

Photoshop isn't the best thing for vector work. You should look into Illustrator.

 

[Jemiba]

For vector drawings, my favourite still is CorelDraw. Xara is good, too, but
the versions I know, are still too limited with its short cuts, so you have to
use the mouse too often.

 

[Merv_P]

I use Corel Draw 9. You can get legitimate copies of earlier versions of Corel very cheaply on Ebay; mine cost £10. Inkscape is a good freeware alternative, but I've read somewhere that development of the Windows version is being discontinued.

prolific1, don't give up the pencil and paper entirely - that's a superb illustration.

 

[prolific1]

Thanks. I'll look into those programs. In the mean time I'll keep on doing it the semi old fashion way. Semi because I import it into PS to clean up the dirt and splash on some color. My Project Pluto SLAM (color) is coming out in APR soon.

 

[Jemiba]

"..don't give up the pencil and paper entirely"

Can you give us a link to download "pencil and paper" ?

 

[prolific1]

Ha ha ha!!!

 

[Jemiba]

Experienced in the reconstruction of drawings ?

One good photo of an aircraft should be enough for a reconstruction, if 4 (four ! ) 3-views are already
available. But ... for me, the cockpit of an aircraft is its face. Get this detail wrong and you’ve got the
whole drawing wrong ! In this case, I have a photo of good quality now ( thanks to deltafan ) and I’ve
the mentioned 3-views. But to my opinion, the angle of the windscreen is too shallow in all four
drawings and not just by a small amount of, say 3 to 5 degrees, but at least of 10 to 15 degrees.
Or is this an optical illusion, due to the fact, that the windscreen is slightly curved ? I’ve tried to
measure the angle on the photo, but after thinking about it again and again, I’m less certain now, than
before. To my opinion (and obviously the opinion of my software, too), the angle won’t change, as the
curvature is backwards. But the voting is 4 to 1 against me ! Am I right, am I wrong ?

 

[Matej]

Can you give us a link to download "pencil and paper" ?

This reminds me one alert message that I reprogrammed when I was young. Then it stated: "LPT1 is not responding. Use the backup paper_and_pencil.sys?"

This old thread was reborn because I am once again searching for the solution of the problem, how to make a 3D model from the 3view. Today I did my first touch with the Blender. Nothing easy but we will see....

 

[Demon Lord Razgriz]

Bah, all you guys using these fancy smancy 3D model makers. Give yourself a challenge and use good old MSpaint, it may not be fancy, but I'll be damned if it doesn't get the job done.

 

[Matej]

I used MSpaint when I was 12. Now it is time for something a bit sophisticated.


So... After a first few hours with the Blender, the result is... :-\ Seems that it will be a long way.

 

[Flitzer]

May I ask if some of the softwares mentioned in this thread are suitable for Macs or are they just PC compatible?

Many thanks
Peter

 

[Michel Van]

May I ask if some of the softwares mentioned in this thread are suitable for Macs or are they just PC compatible?

Many thanks
Peter

"Corel 9","MS Paint","Paint.NET" are 100% Windows

only "CorelDraw 11" was made for Mac
but Corel abandon Mac versions.

"ArtRage" is also for Mac
"Inkscape" need X11 to work on Mac
"Blender", "Gimp", "Google Sketchup", "Photoshop", "illustrator", "Corel Painter" are multiplatform

"Lineform" only for Mac

of those software are these Freebee (no Money !)
"Inkscape", "Blender", "Gimp"

 

[Flitzer]

Thanks Michel

looks like I'll be doing a bit of hunting.

 

[Michel Van]

On price of good Software

"Photoshop CS4 Extended" €1486
"illustrator CS4" €855
together (with Flash CS4, Dreamweaver, Fireworks and Acrobat 9 Pro)
in "Creative Suite 4 Design Premium" for €2616

"Corel Painter X" for €415.31
"CorelDRAW Graphics Suite X4" for €745

heavy Prices !
but we can get this CHEAPER

you are a student ?
Congratulations, you can buy those software for 1/10 of Price B)
like "Creative Suite 4 Design Premium" for €260.

oh your not a student
so do you really need the lates version ?
in that case welcome to eBay
Photoshop CS2 for €220
CorelDRAW for €150

yes i know there alternative for "Free Download"
but no handbook, no support and if you unlucky get Virus, Trojans
and in worst case some Lawyer at your door

so with eBay you got cheap full version and get support or upgrade

 

[Matej]

The cost of the Corel DRAW Graphics Suite X4 in Germany is €745. In its neighbour Czech Republic the same product from the same web corel.com costs €530...... :

I know that because I bought Corel Paint Shop Pro Photo X2 Ultimate a few weeks ago and I saved some 15 Eur! And no matter where you buy it because it is still in English.

 

[Yoyodyne]

Commercial programs cost much less when they aren't the latest and greatest version. Sometimes you can get a hand-me-down from someone who has upgraded. Having said that, the free tools listed above are really very good.


I recall finding a program called 'Piranesi' that was aimed at making artsy-looking drawings from 3D models, the result looks drawn by hand.


For creating view drawings from a 3D model, there are a couple of ways. Load your net-downloaded 3D model, and use the view controls of the application. Most 3D apps have perspective and orthographic cameras. Look for the camera's settings or properties, and use orthographic, this way you need not mess with focal lengths.


Another method would be to use the window view controls - many apps have a single button or command that creates a four-window display with top, front, and side views. These views can be seen in wireframe or shaded modes in most apps. You can choose a view for a single window, and pick top, left, etc. You could grab this as a screen shot, and draw from there in another application.


The next step would be to learn how to cut or slice a 3D model in your application, this will be in the help or manual under 'Modeling'. I don't read online manuals much anymore, but I do open them and use search to find what I want. This slice or cut tool will let you slice the model at any point, and then you can take more screen captures/make renderings to get the cross sections. Don't save the sliced file over the original!


In the process you will probably decide you hate the model you found, and could do a better job yourself, and then you'll be making your way into the deep end of the pool...


Deriving designs from photos has already been addressed well, I would add that the field is called photogrammetry, and searching for that will get you articles and links to some of the software packages.


Using Photoshop, you can scan drawings, then select parts with various tools like the magic wand tool, which can find edges. Once you have refined a selection you can convert it to a vector path, and repeat until your scanned drawing is completely represented as paths. It's pretty easy. No problem to do it in Illustrator either, but they overlap a bit.

 

[BAROBA]

Hi jemiba,

I don't think anyone has answered your questions:
Image 3 is the best one, because you can see that both wings are almost on the same line facing the camera.
So this one is the best for measurements and proportions.
For the shape of the nose you need to guess it.
If both A and B are on the same line, it would be easy.
If not, try to make a 3D model from the pics you have.

Modo is also available for mac,if someone wants to know

Cheers,

Rob

Ok, the reconstruction of 3-views only from photos won’t give really accurate results.
But for the following problem, I think reasonable accuracy should be possible :
I’m just looking for a method, to get the correct proportions between A and B, if B is
given. No, simply stretching the lines, to bring B to the same length on all photos to get A,
don’t work, because A and B aren’t always on the same side of the optical axis .... BTW,
Would it work then ??
Anybody, who has a simple solution ? Or any clue, where to find further reading about this theme ?

 

[BAROBA]

Hey Matej,

How is the model doing? Did you finish it?

I would make it this way
I draw the outlines, just like you did.
Then I would slice up the shape in quads ( polygons with 4 points)
And then I would add more slices that would from the details like the cockpit and engines, while keeping everything quads.
After that I would thicken the shape, so you get a 3d-shape.
Then I would start to move points and polygons to get the right shape.
After that I would start to cut out control surfaces.
Then I would convert all the polygons into Sub-division-surfaces to get a nice smooth surface. But this also will soften the sharp edges, so I would add more slices to get the sharp edges again.
Then I would add final details and texture afterwards.
That gives me a basic stealth plane

CHeers,

Rob

I used MSpaint when I was 12. Now it is time for something a bit sophisticated.


So... After a first few hours with the Blender, the result is... :-\ Seems that it will be a long way.

 

[Jemiba]

Image 3 is the best one, because you can see that both wings are almost on the same line facing the camera.
So this one is the best for measurements and proportions.

Thanks for your answer, that was my choice, too. In fact, in this case reason wasn't to create a new drawing
from photos, but to correct an older one, based on a number of other 3-views, which, to my opinion, were all
more or less faulty. So, the differences I was looking for were relatively small, something most methods are reaching
their limits, I think.

 

[BAROBA]

How to determine a 3view from a picture.

Hi all,

Someone here asked a long time ago if someone knew a good way to draw plane if only there were pictures of the plane.
Apparently someone working for the CIA found a way (a long time ago)
The process is described here:
http://ghostrocket.blogspot.nl/2012/04/early-cold-war-intelligence-estimation.html

The example in the article is very impressive
I hope that someone here finds it useful.
Cheers,

Rob

 

[huckebein]

Re: How to determine a 3view from a picture.

Thank you!

 

[CJGibson]

Re: How to determine a 3view from a picture.

Thanks for that, very interesting. I think if you've had any kind of engineering drawing training and an understanding of projections you can do this, but I've never seen it set out like this. A mis-spent youth looking at GA drawings probaly helps as well.

Chris

PS I've been accused of being a FTD engineer in the past, but with a different meaning.

 

[Stargazer]

Re: How to determine a 3view from a picture.

Link's not working on my end. Any other members experiencing this?

 

[flateric]

Re: How to determine a 3view from a picture.

no, CIA links are OK
try to install ol'good https://www.torproject.org/projects/torbrowser.html.en

 

[quellish]

Re: How to determine a 3view from a picture.

Hi all,

Someone here asked a long time ago if someone knew a good way to draw plane if only there were pictures of the plane.
Apparently someone working for the CIA found a way (a long time ago)
The process is described here:
http://ghostrocket.blogspot.nl/2012/04/early-cold-war-intelligence-estimation.html

The example in the article is very impressive
I hope that someone here finds it useful.
Cheers,

Rob

It's not too hard to do, and a lot easier with some of the tools available today. The discipline is called photogrammetry:

http://en.wikipedia.org/wiki/Photogrammetry

OpenCV and a number of other machine vision libraries out there implement various degrees of what you'd need to determine the 3D shape of an object from photos or video. You don't actually need stereo images to derive a 3D shape, but having multiple images (or video) does help quite a bit. I have a terribly detailed model of a location in China (and the aircraft there) that was derived from various photos that made their way out of China.

 

[UpForce]

Re: How to determine a 3view from a picture.

... I have a terribly detailed model of a location in China ...

"Terribly" detailed or "terribly" detailed?

Photogrammetry is advancing by leaps and bounds these days, it's not too long ago that I too got a little bit of insight into the field ... now considering whether I should look deeper. It's exciting, but so are many things. Not that there aren't available academic/beta/os applications out there, so no intrinsic need to know how it's done though. The thing that is obviously interesting in doing perspective projection by hand (or in this case the inverse, as in the article above) is that it highlights that very different things are difficult in human cognition vs. applying raw computing power.

"Passive" photogrammetry programs' accuracy (from 2D) does sometimes suffer from a lack of fixed reference points, scarce pictorial data, reflections, changes in lighting, etc. so for very detailed requirements types of 3D scanners are probably still preferrable (... the kind that can project their own references on the scene/object). But no matter what way it's done, the field and potential demand for applications (think semantic, human interface, machine learning) seems to be huge both for hard- and software.

 

[Jemiba]

Re: How to determine a 3view from a picture.

An interesting article really and my first thought was "Damn ! I obviously wasted hours and hours in the
past, but it's that simple ! "
But I'm still not completely convinced ...
"..Once the "enclosing box" has been established,.." You'll have to determine all three axes and you probably have to do this
accurately, or you could as well have spent your time for other important things, like lawn mowing or to do the washing up
(and your wife will agree !). The lateral axis should be easy, just a line between the wing tips, but the longitudinal axis isn't
and the vertical axis neither.
Another point, which I still think is important, is the knowledge, where the central point of the photo is. Due to the perspective
distortion, the sides of the "enclosing box" aren't necessarily parallel and the position of the box compared to the central
point determines the deviation from parallelism.
Ok, if the photo was taken from a greater distance, this problem may be less important. And those guys analysing photos taken with
calibrated cameras certainly never met this problem, because the even had the original negatives. But people like me, had have to
use photos from books or magazines can run into problems, because those printed photos often are just cutouts from the originals
and the real central point may not even be on the printed photo !
Maybe I'm completely off the mark, but to my opinion, this method can only be used under very favourable circumstances (standing on
the Red Square with a calibrated camera), which at least I met very rarely.

 

[quellish]

Re: How to determine a 3view from a picture.

Maybe I'm completely off the mark, but to my opinion, this method can only be used under very favourable circumstances (standing on
the Red Square with a calibrated camera), which at least I met very rarely.

Modern digital cameras and camera phones encode a lot of that information in the EXIF data - making every one of those devices equivalent or better than those calibrated cameras for these purposes. Additionally, you can often get time of day and geographic information that can be very useful.

 

[Jemiba]

Re: How to determine a 3view from a picture.

Well, that may help if Aurora, or maybe its russian equivalent will pass (slowly, I hope, or I'll miss the
opportunity for taking a photo ..) over Berlin ! But for those one-offs, or rarely shown types which
sometimes show up on old photos (the Aerosalon Le Bourget during the late '40s/50s was a good example),
most of the EXIF data are lost in the meantime ... :-\

Kidding aside, I'm well aware that for aircraft like the Chengdu J-20, or Sukhoi Su-50 PAK FA it actually is
a worthwhile source for data and the fact, that we saw detailed 3-views quite fast quite probably was the
result. But for those photos dug out from old magazines, maybe printed in double stamp size in then common
newspaper quality, problems will remain, although here, too, modern software can enhance details.
And despite all those problems , there still are lots of types, which I really would like to see on such photos !

 

[CJGibson]

Re: How to determine a 3view from a picture.

I forgot to add that this is quite time-consuming, but really satisfying, when two years later a 3-view turned up and I was ever so slightly wrong.

Chris

 

[Jemiba]

Re: How to determine a 3view from a picture.

That's a feeling I know well, too !
And not bad, too, was the feeling, when I once couldn't collate a 3-view I had made using
a photo, with two or three published 3-views ... until I got one, which was from the manufacturer.
It took me some time, before I realised (and accepted), that it contained a simple fault, which so
was spread over several publications !
So, never believe in 3-views, you didn't manipulate by yourself !

 

[Stargazer]

Re: How to determine a 3view from a picture.

Although I'm extremely interested in this topic and think I understand the general principles, I find that some steps are missing for me to reconstruct the whole process.

Here is a series of pictures showing how I obtained the "box" from an old painting of the 1940s.

1°) This is the unidentified 1940-ish project I've chosen to work from.

2°) I drew the thrust line and the outer limits of the box at the wing tips.

3°) Then I drew the perpendicular lines defining the upper and lower limits of the "box". I realize I made a small mistake in perspective on the perpendicular lines (those which cross the main fuselage's axis) but from my calculations this shouldn't change a lot to the proportions of the box itself.

4°) Then I drew the vertical lines that correspond to the elevation of the box.

5°) Then I materialized the sides of the box in different colors with transparency.

6°) And then the top of the box.

7°) To evaluate the dimensions of the box, I first isolated these shapes.

8°) I then rotated them so that the longer side would be horizontal.

9°) Then I put both shapes in perspective to treat them together.

10°) Then I used my software to unslant the shapes. This gives me the proper measurement in pixels. If I knew the real-life dimensions of the aircraft, I would be able to determine a scale (x pixels for y meters) but I don't, so I'll stick with the pixels for now.

11°) And that's pretty much where I'm stuck...

QUESTIONS:

1°) According to the article, I should determine "a/b" and "b/c". Is this supposed to be a division or a multiplication?

2°) When this is done, I should obtain a coordinate. What do I do with it? Do I draw a line from there to point 0,0 to obtain the "yaw angle"?

More questions will follow, but for now I really need to get these clarified. Thanks in advance to the specialists on this site!

 

[Jemiba]

Re: How to determine a 3view from a picture.

You've chosen an artists impression, that can be a blessing or a curse . A blessing, when the artist used
the perspective correctly, a curse, when not.
In this case, perspective seems to be ok, perhaps we should assume, it's a photo. You drew the lateral
lines parallel, but using the given fixed points (e.g. the gaps between tailplpane and rudder and between
ailerons and wings), you'll see, that they actually aren't. Drawing the longitudinal axis on such a drawing
really isn't that easy, to my opinion. And as the outer lines at the wing tips aren't parallel to each other
and to the longitudinal axis neither, there's actually no real clue about the angle. Without additional data
it would be hard to get the proportions right, and that's the first aim, I think. The greater the distance
from the camera to the aircraft, the less those point would cause problems. But this drawing could well
be the view from the gallery in an exhibition hall (Aerosalon Le Bourget, perhaps ?), or from the tower
onto a taxying aircraft, so distance would have been quite short. Or maybe distance wasn't that short, but
focal length was long ? Or the other way round ? This actually can affect the result a lot, as I already found
out more than once to my disadvantage !
Regarding your questions: a/b and b/c are ratios, that means divisions.

 

[Kryptid]

Re: How to determine a 3view from a picture.

I do wish that the article had explained things more fully as well. I've been trying to see if I can develop a 3-view of Lockheed Martin's "Miss February". However, I end up with a leading edge sweep in excess of 60 degrees, which is unusually high for a modern fighter (unless this represents some increased emphasis on supercruise in the design?). I figure I may not be doing something correctly.

 

[theponja]

Re: How to determine a 3view from a picture.

I bought a while ago this book:

from: http://www.modelersnotebook.com/

The book describes in good detail the process to make a 3d view from pictures. The advance chapters go subjects as proyections for every shape. The book is easy to read. Sadly no news about new volumes in the series.

 

[sferrin]

Re: How to determine a 3view from a picture.

Hi all,

Someone here asked a long time ago if someone knew a good way to draw plane if only there were pictures of the plane.
Apparently someone working for the CIA found a way (a long time ago)
The process is described here:
http://ghostrocket.blogspot.nl/2012/04/early-cold-war-intelligence-estimation.html

The example in the article is very impressive
I hope that someone here finds it useful.
Cheers,

Rob

It's not too hard to do, and a lot easier with some of the tools available today. The discipline is called photogrammetry:

http://en.wikipedia.org/wiki/Photogrammetry

OpenCV and a number of other machine vision libraries out there implement various degrees of what you'd need to determine the 3D shape of an object from photos or video. You don't actually need stereo images to derive a 3D shape, but having multiple images (or video) does help quite a bit. I have a terribly detailed model of a location in China (and the aircraft there) that was derived from various photos that made their way out of China.

Back in the day this is how we derived the terrain/structure data for the Front Page Sports Golf/PGA Championship Golf/Tiger Woods golf. (We had really high res shots from planes though. And that technique doesn't work for squat if there are trees in the way. )

 

[steelpillow]

Re: How to determine a 3view from a picture.

The process is described here:
http://ghostrocket.blogspot.nl/2012/04/early-cold-war-intelligence-estimation.html

I can't get to that web site at all, it says it has been removed.

As for reconstructing a 3D form from photographs, in general one has to make assumptions about objects in the scene - that certain lines are parallel for example, or that a certain object is symmetrical. Otherwise, things like the focal length of the lens have to be guessed at and a bad guess leads to a distorted reconstruction. The underlying mathematics is something called projective geometry - a fascinating discipline in its own right, being the cleanest and most elegant of all geometries.

I'd love to be able to get at some of that software that people have mentioned here and put it through its paces. I have a couple of photos of a plane - courtesy of the Science Museum - that I'd like to turn into 3-views. However, my experience of magic software that promises the Earth is that in the end it falls short and doing the job by hand is the better bet. All I have to do is to remember my geometry lessons, ahem.

 

[overscan (PaulMM)]

Re: How to determine a 3view from a picture.

Well, there are several people on the forum who could help if you share the photos with them.