My first scale project ...
Research 2: Transforming the side view photograph
I made a Photoshop file at 1:4 scale and 72 dpi (screen resolution)
which makes a heavy file of 75 Mb, but my pc can handle it and it
is convenient for taking measures. I added the drawings, an image
of the Cirrus Hermes engine and green bars for the measures I could
be sure of: cabin height, chord, the distances between the formers,
plus the supposed length of 7,80 m. These measures were then the
base for the set of guide lines (the thin blue lines).
Everything put indepently in layers of which the visibility can be turned on and off.
At this point it was already clear the Koolhoven drawing is wrong. The Desoutter Mk.I drawing had a much better fit.
I enlarged and mirrored the side view photograph and tilted it to get the aircraft horizontal. I determined the vanishing point (white/green symbol in the window) and added some 'white' above to get the vanishing point vertically centered.
The vanishing point is where the photographer stood, the square line
from aircraft to camera. To get the entire aircraft in picture however,
he had turned his camera towards the tail. The nose is closer to the camera;
the tail is far from the camera. The nose is shown to big in the photograph;
the tail is shown to small. The vanishing lines converge to the right.
For a start I made the left side smaller and the right somewhat bigger
to get the vanishing lines parallel. There were no lines visible
that could tell me how much, so I put my trust in the drawing and made
the front of the nose and the end of the tail corresponding.
Next I cut up the photograph and moved the pieces aside. I treated the photograph in slices, putting every compartment in place, stretching to the correct width and height, plus making horizontal lines horizontal. I could identify the compartments because the panel lines were just visible enough.
Transformation Perspective works symmetrical; only the bottom corners of the selection need to be moved. The vertically centered vanishing point prevents asymmetrical distortion between the upper and lower halve of the photograph.
Inbetween I transformed the entire tail, making all panel lines corresponding to the guide lines, in order to determine the size of the rudder. It measured 46 cm., but since rudder and fuselage side are at different angles, it had to be somewhat smaller, so most likely it was a round 45 cm.
I had used the Desoutter's rudder size, 55 cm., in the length bar, so I had to correct this bar and the outside guide lines. (To be sure I also checked the 55 cm., but it resulted in a very obvious over-stretched rudder.)
After this intermezzo I made the left side corresponding to the cabin and enlarged the right side, making all horizontal lines horizontal.
I put the tail aside and continued slicing the compartments, putting them in place and stretching them to the correct width using the regular transform tool.
Another great thing of this method: you get to know the correct position of details! No need for eyeballing the position of the registration.
A few steps back I copied the area with fin and rudder and put it aside in a layer on top. In the image below I made it visible again. I cut off the tail (see dashed selection line) and stretched it, getting the rudder to its correct size. This also got the fin to a reasonable correct size.
And a little fine-tuning, putting the horizontal lines of the rudder horizontal. Then I merged the layer with the entire image below.
The nose turned out just fine with a single transform.
In the image below you see the end result with measures, Desoutter drawing
and engine. There was no way I could get the propellor within the specified
7,80 m. The front of the cowling however seemed spot on. Apparantly this
is how the 7,80 was measured.
I added the photograph of the flying G-AAGC. This photograph
is taken from very far and has little perspective distortion. The close
resemblance of the proportions of both is a strong argument for my end
result being correct.
Finally a last comparison to the Koolhoven drawing which I thought to use as a base for my model. Nevertheless this drawing is not useless because some of it is correct.
The cowling and windshield needed some fine-tuning because of the parts close to the centerline. I used two other side view photographs for comparison; the different angles of view gave me a better understanding of the perspective in this area.
Next I needed to determine the position of the wheels and the oleo leg in the sideview. For this I first needed to know the track width and the fuselage width, as well as the distances of the wing ribs to determine where the oleo leg is attached.
According to the apparantly most reliable Desoutter construction drawing
the fuselage must have been about 97 cm. wide. A peculiar measure
which seemed more believable if I started from
the specified cabin width of
90 cm. (guessing that was Koolhoven's starting point too) and add up
longerons plus sheeting. At a photograph of the last F.K.41 under construction
I
was able to measure that longerons and all other woodwork was
3,2 x 3,2 cm. Plus sheeting it adds up to: