My first post in this series described how I produced a working drawing for a model of GWR No.184. Now, it's time to consider the tender and, for this, information is less readily available. Wet and windy weather has kept me indoors, so that progress has been rather quicker than I had anticipated.
The various photographs of No.184 show it partnered with a variety of different tenders, so I decided to try to model the tender shown together with its 1893 re-build. I have not found a broadside view of this engine/tender combination, so have had to work from the oblique view shown in my previous post.
In a forum thread on Estimating Dimensions from Oblique Views , I described a technique for correcting the effects of perspective by using the tools in the photo-editing program 'Photoshop Elements' (PSE). This program can adjust the magnification across an image, to correct for the angle between the subject and the camera. Once this has been done, it becomes possible to compare the distances between different points on a photograph, providing these points all lie in the same plane.
In order to use this method on the photograph of No.184, I first identified some parallel lines that are as well-spaced as possible in the image. I chose a line along the top of the boiler and the line defined by the rails below the engine, and marked these in red on the photo, as shown below:
The next step is to use the 'Perspective' tools in PSE, to distort the image until these two lines are made as parallel as possible,. ( I use an early version of PSE but the principles remain the same in later versions, although the menus may differ in detail.)
First select the whole image (Select | All) and then select 'Transform | Perspective' in the Image Menu. Use the mouse to drag the handles that appear around the image, to offset the effects of perspective by enlarging the more distant parts of the image, as show below. It helps to turn 'on' the grid display, when judging when the lines have become parallel.
Once the image has been 'squared up' in this way, it is possible to compare distances measured along the frames of the engine and tender. The coupled wheelbase of No.184 is known to have been 8 feet, so I used the 'Line' tool to draw a line between the wheel centres, in the plane of the engine frame, and measured the length of this line in pixels. On my image, this length was 188 pixels. I then drew similar lines between the tender wheel centres, which both measured 130 pixels. The 'real' distances between these points are in direct proportion to these lengths, giving an estimated tender wheelbase of 130/188 X 8 feet, or 5' 9" between each pair of axles.
Unfortunately, the photograph does not show the entire length of the tender, so I could not use this method to determine the overall dimensions of the tender frame. There is, however, a rather similar type of tender, shown behind GWR No. 244, in Part 3 of the RCTS "Locomotives of the GWR" (Fig.C60), which appears to have the same 5' 9" + 5' 9" wheelbase. ( * This is not certain and is based on my assumption that the wheelbase of this re-built engine was 7' 0" + 8' 0" ). This tender scales to 19' 3" platform length.
I have used the side-on photograph of this tender to produce a line drawing. First, I increased the Contrast of the photo, using PSE, and then selected 'Stylize | Find Edges from the Filters Menu, to produce the rough outline shown below. I used this outline as a template to draw a more refined outline drawing and also included some of the details shown in photos of No.184.
I adjusted the sizes of both the engine and tender outline drawings, such that 40 pixels correspond to 1 foot on the real engine, and set the scale to 100px/cm, so that they will print at 4mm/foot scale. I then pasted the two drawings (engine and tender) together, to provide an overall impression of the complete 'system' that needs to be modelled:
Now that I have some scaled sketches, I can start to design the main parts for a model engine. My first step will be to draw, and then cut out, some paper templates, which I shall use to explore a feasible method for construction, and to check clearances for (00-gauge) wheels, motor, etc.
In previous models, I have cut out these templates by hand but, this time, I intend to use my Silhouette cutter.
* My assumption is supported by the ratio between the wheelbases being extremely close to 7 : 8. I used a spreadsheet to find that the most precise match to my pixel ratios lay at 6' 10" + 7'10", which would yield a tender wheelbase of 5' 8". The tolerances in my measurements are, however, greater than ± 1" so I have kept my initial estimate.