During the previous weeks I formed two main elements of my model: the wing and the main part of fuselage. As you saw, I could not resist myself for adding some details to the wing (like the ribs and spars of the flaps).
Now I think that this is a proper time to stop modeling for a moment and compare the shape of the newly modeled parts to the real airplane. If I find and fix an error in the fuselage shape now, it will save me from much more troubles in the future! If I find an error in the wing shape – well, I will have more work, because I already fit it with some details which will also require reworking… You will see.
The idea of using photos as a precise references emerged from the job that I did two years ago. One of my colleagues asked me if I can recreate the precise shape of the stencils painted on an airplane. He wanted to determine details of the numbers painted on the P-40s stationed in 1941 around Oahu. He sent me the photo. I started by fitting the 3D model to this historical picture, finding by trial-and-error the location and focus of the camera (as in Figure 29‑1):
Then I made the model surface completely transparent. I placed the opaque drawing (texture) of the large white tactical numbers on its fuselage, and the black, smaller, radio call numbers on the fin. I rendered the result over the underlying photo, finding all the differences. Then I adjusted the drawing and made another check. After several approximations I recreated precisely shapes and sizes of these “decals”.
The key point in this process was to recreate the location and focus of the camera that was used to make the particular photo. Now I realized that it is possible to use the photos in the same way as precise references for my model. All I needed was a high-resolution picture.
I decided to begin with one of the archival photos of the SBD-5 from VMSB 231, made in spring 1944 (the original photo from Figure 29‑2 is 2127px wide):
To find the camera projection that fits the model into airplane contours on the photo, you have to coordinate the location of the camera and its direction (I used for this purpose an auxiliary “Target” object). Yet another parameter to be adjusted is the camera lens length:
The whole process is an iteration: I started from a rough first approximation (as in Figure 29‑3). Then I enhanced it, gradually determining the ultimate camera and target location, as well as the focal length.
In this process I based mostly on the elements which dimensions were determined by the “hard” evidence. It pays off that I placed most of the the fuselage mesh edges along the original bulkheads and longerons. (I will also benefit from this during further stages of my work). I was quite sure of the bulkhead stations because they were set according the original diagram. Thus I started by fitting to the photo the fuselage between the firewall (station 0) and the last bulkhead (station 271). Then I tried to find the proper camera focus that fits the middle bulkheads to the rivet seams and panel lines which are visible on the photo (Figure 29‑4):
I was also able to fit to this photo the root rib of the outer wing panels. However, I could not match the position of the wing tip! When I fit this tip to the photo, the fuselage deflection was wrong. Otherwise, as you can see (Figure 29‑4) the wing tip of the model was a few inches below the tip on the photo. I started to wonder why I have such a problem…
Figure 29‑5 shows the best projection I was able to find. The fuselage bulkheads fit well the seam lines from the photo. It seems that the bottom contour of the tail was somewhat lower than in my model:
The only weak point is the different elevation of the wing tip. I cannot say that it had a greater dihedral, because it was dimensioned on the original general arrangement diagram (7⁰ 30’ along the upper wing contour, in the front view).
Finally I came to conclusion that what I can see on this photo is the elastic deformation (bending) of the loaded wing! This aircraft here is depicted in the flight, right? This means that these wings are carrying the load of about 4 tons of its weight. Their structure was stiff, but not absolutely rigid: every beam deforms (more or less) under the load. The airplane wings are not the exception: while flying in an airliner (like Boeing or Airbus) you can observe how their wing tips bend in the air. Of course, the relatively short, wide wings of the SBD Dauntless were much more resistant to such deformations. Nevertheless I think that we can trace the slight bending of this wing leading edge on the other shots of this airplane. For example, the white sun reflection on the photo in Figure 29‑6) allows me to reveal this dynamic deformation:
We can see here the bending of the outer wing panel. However, there was another deformation: in its joint with the center wing. The root rib under loads slightly rotates around the wing chord, which elevates the wing tip even further (Figure 29‑6b). I suppose that the center wing was much stiffer (it had thicker airfoil and shorter span than the outer panels).
Using a side photo of a flying airplane, always try to estimate the elastic deformation of its wing, especially the wing tips! Usually such a deformation makes these photos less usable as the precise reference for a 3D model.
Frankly speaking, this conclusion was a little surprise. I have not noticed such a deformation before — maybe because I was mainly focused on the WW II fighters? Fighter wings are the stiffest ones…
NOTE! After couple of weeks I discovered that this is false theory (or, at least, overrated): the Dauntless wing were much stiffer, and the small curve of their leading edge visible in Figure 29-6 can be caused by the lens barrel distortion. In the level flight the Dauntless wing tips have no significant deformation (see in this post, Figure 32-7). The deformation of the photo from Figure 29-5 can have other reasons – for example physical deformation of its negative.
In the airplane standing on the ground the wing deformation is minimal, thus such a picture suits better the reference purposes. Ultimately I decided to use some of the photos published by the Pacific Aviation Museum on flickr.com. Figure 29‑7 shows the result (I had to flip this photo from left to right because I modeled the left wing):
As you can see, the wing perfectly fits its contour in this photo. I have found that the left aileron airplane was rotated upward by about 4⁰. I can see some differences in the hinge location of the upper wing flap (on the photo it seems to be placed at somewhat different angle, and shifted to the rear). The contour of the aileron bay also seems to be a little bit lower. On the fuselage you can see that the bottom contour of the tail is placed lower than in my model — confirming the observation form the previous photo.
To effectively match a 3D view projection to the photo, you have to be sure that at least some points of your model are in their real locations. In the case of my Dauntless model these “fixed points” were:
- Positions of the fuselage bulkheads (I read them from the stations diagram);
- The basic trapezoid contour of the wing (deduced from the general arrangement diagram dimensions);
- Positions, shapes and sizes of the root rib (the rib at the joint between the center wing and the outer wing panel. I deduced its chord and spatial location in relation to the firewall from the general arrangement diagram dimensions, while its shape is declared as NACA-2415 airfoil);
Then you can precisely determine the location of the camera and its lens by matching these elements to the photo.
When you find deviations as these that I have found in the aileron and flap hinges, it is always a good idea to check them on another photo. Thus I fitted my model into a different picture from the same PAM photo stream on flickr (Figure 29‑8):
In this case I opened the wing flaps, because their straight contours helped me in precise positioning of my camera. It was possible to fit the bottom flaps to this photo (I just discovered that in this photo their deflection angle is 40⁰, while according to the specs it should be 42⁰). I was able to verify locations of their ribs and spars. (It occurs that these ribs, set according the stations diagram, are in the proper places). However, the upper flap did not fit properly into its contour in the photo. It was only possible when I shifted its hinge to the rear, placing it as in Figure 29‑7. In this way I confirmed that these wing elements require corrections.
After these initial findings I decided to verify both: the wing and the fuselage, to fix all the differences I would find. Of course, it required more photos. Matching the model projection to a single picture takes me several hours of work (usually — one evening). I assigned to each of these pictures a separate camera (as well as the camera target object). Their names are three-letter shortcuts of the source photo followed by the ordinal number: thus PAM-1 means “Pacific Aviation Museum – 1.jpg”, UND-1 is “Unidentified – 1.jpg” and so on. I think that these reference pictures will be also useful in the future stages of this work. Switching between these cameras requires several steps: you have to type the path to the corresponding photo, as well as to alter the scene renderer aspect ratio. To facilitate this operation I created a dedicated add-on, which allows me to switch between these pictures with one click (Figure 29‑9):
The ability to immediately switch between various reference photos definitely makes the difference! It encourages to study the same fragment from all possible sides.
The photos can be extremely useful reference, but they do not replace the traditional scale plans. First you have to create a 3D model that is close enough to the real shape, using the plans. Then you can project such a model onto a high-resolution picture for the further improvements.
In this source *.blend file you can evaluate yourself the model matched to the first picture from the Pacific Aviation Museum.
In the next two posts I will write about the results of this verification. In the first one I will describe the errors that I found in the shape of my fuselage. In the second post I will describe the differences that I found in the wing. Sometimes fixing these minor errors require several hours of work… But this is why we are the modelers (“a slightly different human being”).
6 thoughts on “Matching the Model to the Photos”