Modeling the Empennage (1)

The horizontal tailplane has similar structure to the wing — but it is simpler. Thus I started it in the same way as the wing, by forming its root airfoil (Figure 32‑1):

Figure 32-1 Modeling the root airfoil of the horizontal tailplane

In the most of the aircraft the tailplane has a symmetric airfoil. So it was in the Dauntless. I did not find its signature (family) in any of the reference materials, thus I carefully copied its contour from the photos (its rear part — the elevator — seems to have modified shape, anyway). It has incidence angle of 2⁰, so I rotated the rib object and used a Mirror modifier to generate its bottom part.

During this work I decided that I will use this rib as an auxiliary reference object for shaping the horizontal stabilizer. To precisely match the contour copied from the photos, I rotated part of this curve in the top view. Now it runs along the outer edge of the tailplane fairing (Figure 32‑2):

Figure 32-2 Modeling the reference airfoil

However, because I am going to copy this rib into the initial edge of the horizontal stabilizer, I already prepared three vertices for the leading edge of the elevator (Figure 32‑2).

During this work I was struck by the idea that it is stupid thing to model the whole empennage, and then to verify it against the photos. The much better approach is first to “draw” in the 3D space their contours and match them to the photos, then to model their surfaces. In this way I can identify errors in my reference drawings before I start the modeling! The parts formed in this “verified” way and continuously matched to the references will have better quality!

Thus I interrupted forming the horizontal tailplane, and quickly shaped another auxiliary object — the contour of the rudder and fin (Figure 32‑3):

Figure 32-3 Fitting the fin and rudder contours

What’s more, I decided to recreate in the model the basic reference “trapezes” of the fin and rudder. They are determined by the explicit dimensions in the general arrangement drawing, which I already used some months ago to draw the 2D reference drawings (Figure 32‑4):

32-4 The explicit dimensions of the basic trapeze of the fin and rudder

While in the model space the 1 unit corresponds to 1 inch, I did not need to multiply every dimension by the scale coefficient. It was a big surprise when the trapeze drawn according these re-applied dimensions occurred shifted left by 0.7” (Figure 32‑5):

Figure 32-5 The basic trapeze of the fin, drawn anew

 I immediately did the same test for the horizontal tailplane. It also was shifted by 0.7” (Figure 32‑6):

Figure 32-6 The basic trapeze of the horizontal tailplane, drawn anew

Well, such a coincidence suggest that I made a kind of systematic error in calculating locations of the elevator and rudder axis for my scale plans. Most probably there was something in their extremely long position, measured from the wing leading edge (see the general arrangement diagram in Figure 32‑4). For example, it could be a rounding error of the scale coefficient!

If I was wrong in this case, I could made other errors. I decided that it is proper time to re-use the original photos from the web page of Chino Planes of Fame Air Museum. Their resolution is only half of the resolution of the photos from Pacific Aviation Museum Pearl Harbor. However, they were made using a long-lens camera. (You can read that the standard length from the EXIF section of the Chino photos — it was 400 mm. The photos from Pacific Aviation Museum Pearl Harbor were made with the standard lens length: 36 mm).

Using this focus length, it is easier to fit the model and the photo (Figure 32‑7):

Figure 32-7 The long-lens (400mm) photo of the flying airplane

As you can see, this is a flying airplane — and there is no visible dynamic deformation of the wingtip! This means that the whole “theory” that I described in my post four weeks ago (in Figure 29-6) was wrong! The wing is much stiffer than I thought. The deformation of the historical photo can have other reason. It could be significant barrel distortion of its lens, or the deformation of the negative. I do not know.

I verified contours of the horizontal tailplane by matching the model to another photo (Figure 32‑8):

Figure 32-8 Fitting the top view contours to another long-lens photo

Note, that this is another photo that I used to draw my scale plans. However, this time I left it unaltered, to avoid eventual errors that I could made by setting it horizontally and scaling.

In general, the model fits this photo pretty well. However, there are small differences at tailplane and wing tips. I started to suspect that such a photo can still have a small barrel distortion.

Finally, I used the third Chino photo to further verify the side view (Figure 32‑9):

Figure 32-9 Fitting the side contours of the model to another long-lens photo

The dimensioned contours of the empennage helped me to match better the other photos. For example, I slightly updated the projection parameters of the SBD-5 pictures from Pacific Aviation Museum Pearl Harbor (Figure 32‑10):

Figure 32-10 The updated projection onto another photo

The contours of the tail and fuselage fits the photo pretty well. There is just a visible difference in the wing tip spans — I think that this is the effect of the barrel distortion.

In this source *.blend file you can evaluate yourself the model from this post.

Well, I started to build the tailplane in this post, but this process ended in another verification. However, it spared me from similar check that I would have to perform on the finished empennage. Now I can quickly build this section — in the next post I will finish the horizontal tailplane.

4 thoughts on “Modeling the Empennage (1)

  1. Hey! Nice work there! I own two of your books and I’m very satisfied with them! I’m modeling a plane for X-Plane and your books have been a great help in the process. I very much like the way you approach the modeling process of an airplane and all the resources you use.
    One day, if you have time you could make a tutorial on how you match the photos to the model. It could take me days and not get even close to match them like you do.
    Keep it up! Cheers


    1. Thank you! I am really happy that you have found these books useful!

      About the reference photos: I still have to verify all the results of this method, before I will be able to recommend it in a tutorial. At this moment I am aware of the small but annoying differences between my model an these photos. Each of them is different and individual to a particular photo. They may be the effects of the lens barrel distortion (which is possible to determine for a ground photos, using Hugin softare). It is more difficult to cope with the differences in the long-lens, in-flight photos. (I have some differences in matching the model to the photo from Figure 32-8. I am not sure if this is a result of dynamic wing deformations, which are smaller than I had initially estimated, but still visible, especially in a turn like the one depicted on this photo).

      In general, I use following tactics for such a match:
      1. Define a separate Blender camera and its target object (camera tracks this object using “Track To” constraint, including the rotations). I have found that it is more convenient to direct the camera using such an auxiliary object;
      2. Read from the EXIF section of the raw photo the original camera lens length, recalculate it into the standard 35mm “film” length and set in the camera data (lens length) in Blender;
      3. Set the photo as the background image (“Camera” projection) and set the render height and width to the size of this photo (i.e. to have the same aspect ratio);
      4. Move the camera and the target object (and eventually rotate the target object) to project (roughly) the model onto the photo displayed as the background image;
      5. Fine-tune this initial setting, matching the model elements that have known, explicit dimensions. (Example of such elements: locations of the bulkheads along the fuselage centerline, which you can read from the stations diagram);

      Of course, point 5 has to be repeated. The more distant, “dimensioned” points you have in your model (for example: fusleage bulkheads, the fin, tailplane, wing tips), the more accurate results.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s