Military AI Works

After reading Mark Grigg’s trials and tribulations with 3 views and using photos as backdrops in FSDS , I decided to write this short summary of the problems I encountered.
Accurate 3 views and good photos are necessary tools for ,in our case ,AI aircraft .
In preparation for the F/ CF 101 I had gathered no less than 7 different sets of 3 or 5 view drawings .
I’l l start of by saying there are too few ACCURATE modern aircraft 3 views out there . ( I only wish that Willey Peeters had drawn a F-101 .His F-104 3 views are incredibly accurate .)
I initially used a Canadian set and quickly discovered errors . One error that persists in ALL F-101 3 views is the shape of the main gear doors . A detail that was incorporated in the 1st ( by date) set and that has been copied/plageurised by all subsequent in their 3 views . Some drawings had the canopy in different locations along the x-axis .As much as 30 inches in one case ! .All front views are mostly a figment of the draughtsman’s imagination and a are useless ! So whatya do ? Use some photos right ? That is fraught with its own pitfalls as documented here
http://gizmodo.com/5857279/this-is-how- ... retty-face
http://helpx.adobe.com/creative-suite/u ... a-raw.html
using a 35MM lens
http://paulbourke.net/miscellaneous/lenscorrection/
good example of pincushion,barrel and perspective distortion here
http://www.cambridgeincolour.com/tutori ... ctions.htm
and here
http://www.dxo.com/ca/photo/dxo_optics_ ... distortion
So we are stuck again ? Not quite, one can correct distortions in Photoshop CS2 ,or any other good program such as Paint shop Pro to correct some linear distortions
Here’s an excellent tutorial here showing how to correct some distortion
http://www.kenrockwell.com/tech/correct ... ortion.htm
In conclusion I’ll say that using 3 views and photo documentation you have to be aware of where the errors are . In a photo , one has to know where the point of interest of the camera lens was ,and realise anything outside of that is prone to error .

Brian,
In most instances that I have found, 3 view drawings on the net have come from people tracing the outline of photos. That is why there’re differences between these drawings in scale. The drawings are not intended for flight simulator model development.

Depending on the focal length and angle of the lense, etc of the photo being traced, the scale will differ.
Only manufacturer blue prints can be relied upon to be correct. You will find 3 view drawings on the web that contain cut-out sections of the fuselage and wing to provide assistance to develop an aircraft but they too are loosely based and require a bit of extra input from visual aids like high-res photos to complete the shape.

As you know I've been working on the F-35 model and in doing so I've looked at 100's of photos and even taken a close up look at other models developed in G-max and 3D Max, and similar models made in professional programs and found so many differences between models its not funny.
Nobody can expect you to make a perfect designed model in FSDS or G-max so don't try or you'll drive your self insane. Even those skilled people making high-poly 3D max models are getting it wrong along with a heap of Payware guru's out there. I have the PMDG 737 range, probably considered the best model set and even they haven't got it totally spot on.

As this is your first attempt at developing an aircraft model, may I suggest that you keep to a simple design, like an ultralite and then work your way up to more complex models.
The issues with 3 view drawings are not going to go away, so we have to do the best we can with what we have to work with at the time.
Cheer's

Some3 views are inaccurate because they are based on photos which “ lie” in the 1st place and you have to be aware where that occurs. In other words you have to know where to back off but still represent a realistic AI A/C.
Engineering blue prints ARE OK but are generally of small parts or are huge drawings .The F-101 station point drawing I have is 48 inches long !!
Your comment on focal length and angle of the lense in relation to the azimuth is exactly what I was referring to in my original post .I’ve used two graphing/plotting programs to check photos and to produce accurate drawings in the past.You don’t need a lot of XYZ plots In the case of the T-38 I used 54 ( checked Kevin Reed’s T-38 and it’s an extremely accurate model ) ,F-14 = 87 points , F-4= 62 and the F-101 =76 points.
Too late to start with an “easy” model. I purposely chose the 101 due to its complex structure with its compound,parabolic and ellipital curves morfing from the circular arc around the radome in the pointy part .

Brian,
The purpose of AI traffic is to represent a realistic representation that is not over built. My point is that you can not expect a perfect shape when you need to cut corners to keep the model within the boundary of an acceptable AI model. Depending on which photo you overlay onto my models, you will find big differences. Some will fit very closely while others look nothing like the same airframe. Depends on the lense that took the pic and without knowing all that info, its a best guess for me.

I do agree that that Kev Reed does make accurite AI and places a lot of time into each and every part of his model to ensure that he can get the best shape from the least amount of polys being used, and he does that perfectly.
cheer's,