Top header photo
 Trainiax logo
HomeNewContactLinks

Drawing Accuracy

I believe Albert Einstein once said "The more I learn, the more I realize I don't know." That largely sums up the history of the drawings. While I wanted to believe that my early efforts were accurate, I did not know (or even look up) most locomotive dimensions - nor did I know how inaccurate my renditions were as a result. I was only 14 years old when I started the 1:55 series in 2000, but with the relative simplicity of the drawings, the series quickly evolved to encompass almost every major locomotive model within about five years. Back in the day, I was working in a world of few reference books and a much smaller Internet, and the result was my broadest (and until recently my most well-known) drawing series... of not-so-accurate drawings.

Over the years, I've learned a lot and upped my standards, now striving for absolute accuracy in the latest 1:18 drawings. I have several books devoted solely to locomotive dimensions, along with thousands of my own photos and tens of thousands of reference files obtained from the Internet. But the more dimensions I measure and the more resources I look up, the more sources of error I find - not to mention that I'm highly prone to missing occasional small details - and so I still end up correcting my more recent drawings after their initial completion. I doubt I'll ever redraw in 1:18 all the locomotive types I drew in 1:55 because I know there are some that I probably could never draw accurately.

Moral of the story: My old drawings are definitely not accurate, but my newer (1:18) drawings are as accurate as the dimensions I get from diagrams, photos and real-world measurements. And if ever you come across any inaccuracies in my 1:18 drawings, please let me know - so I can correct them!

What's In an Inch?

I've often come across online discussions about models (particularly in HO Scale) where one person will raise a concern about a model's dimensional inaccuracy, while another person will respond something like "a couple of scale inches is too small for me to see or care about in HO."

Can a one- or two-inch discrepancy be noticeable? Well, it depends.

Consider a drawing or model that is 2" too long over the coupler pulling faces, but in which that extra length is spread evenly and proportionally over the entire locomotive. At a glance, it would be absolutely impossible to spot the difference. But now consider a set of handrails or grabirons that are 2" too thick. When the original pipe is only 0.75" to 1.5" in diameter, an increase of 2" would be hugely obvious.

The effect is that an inaccuracy is noticeable not so much based on the absolute amount, but by the proportional error. In many cases, that's how I've come across inaccuracies in the drawings, where a difference as small as 0.5" is not noticeable at first glance, but becomes obvious when components don't fit when transferred to another locomotive model, or when a logo or paint scheme doesn't overlay the hood properly.

Of course, the prototype is not always consistent either, sometimes varying by up to 1" - generally in non-critical dimensions like handrail stanchion spacing.

Scale Comparisons

Below you'll see how the accuracy of the drawings has evolved over the years, with comparisons of an EMD SD40-2 in each scale (in black) with my latest version in 1:18 (in orange).

1:55 Series

As my first major drawing series that lasted for more than a year, the 1:55 series has the widest variation in accuracy, which can be divided in three phases. For more information on the history of the 1:55 series, see the Drawing History page.

Phase 1: 2000 - 2004

Drawings from 2000 to 2004 are not accurate, and in fact their inaccuracies spurred me to redraw many of them only a few years later. In particular, length dimensions were hit-or-miss, pilot-to-coupler distances tended to be too short, and walkways tended to be low, making the hoods too tall and the trucks "squashed". Some of the worst offenders were late-model EMD SD units, which initially had underframes not much higher than a GP9, and the largest error I later corrected was in ALCO Century series drawings from 2002, in which the truck centers were 18" too long. Most drawings were generic; there were very few railroad-specific versions.

These drawings have white backgrounds, except for some rolling stock drawings from 2004, to which I retroactively applied cyan backgrounds and which were somewhat more accurate than my locomotive drawings of the time.

Show Image

Phase 2: April 2005 - January 2007

In the "long term drawing update" (as I called it) starting in April 2005, I redrew many 1:55 drawings to be more accurate. These revised drawings replaced (rather than supplemented) older versions, which were removed as revised versions were completed. At this point I also began to add huge numbers of railroad-specific versions.

These drawings follow major dimensions more closely and are generally accurate to within a few scale inches in other measurements. Although not perfect, they corrected many flaws found in the older drawings. These drawings have cyan backgrounds.

Show Image

Drawings from December 2006 to January 2007 (beginning with the SD50-SD60 series) had much more attention paid to the dimensions of smaller details. A few series that had already been revised in 2005, such as the GP40-2(W) and SD40-2 series, received a second revision that involved many minor dimension corrections and redrawn details such as traction motors and handrail stanchions. Relatively few drawings received this treatment before I progressed detail one step further in February 2007.

Show Image

Phase 3: February 2007 - February 2008

On February 9 2007, I started the third and last phase of the 1:55 series, which represents the shortest of the three 1:55 eras. These drawings observe dimensions more closely and also received substantial upgrades to smaller details. Their accuracy is generally to within one to three scale inches. These drawings have sky blue backgrounds, which were adopted for all later scales as well; they were also the first to use white-outlined (rather than solid black) air hoses.

Show Image

Why haven't I corrected all the old (and not-so-old) 1:55 drawings to be perfectly accurate? Well, given the variety of locomotives and rolling stock (some quite rare or obscure) I don't think I could ever actually make all of them 100% accurate - and even if I could, I figure my time is better spent making new, larger and more accurate drawings from scratch.

1:36 Series

February 2008 - February 2012

The primary goal of the 1:36 series (introduced on February 25, 2008) was to include more detail than what the 1:55 series allowed, while carrying over the goal of increased accuracy started with the last 1:55 drawings. These drawings follow almost all major dimensions exactly and are generally accurate to within one scale inch in minor dimensions. I made no more 1:55 drawings after the start of the 1:36 series. Since they are drawn at 2 pixels per scale inch, the finest increment is 1/2 inch.

Show Image

There are two noteworthy discrepancies that affect a large number of drawings. The first affects the majority EMD hood units. I used photos to measure the hood height on the first unit I drew in 2008 (an EMD RM-1, or GP38 rebuild) and my measurements were off. As a result, all EMD hood units derived from that first drawing have a hood that is 1 to 1.5 scale inches too short above the walkway (depending on dimension rounding). However, I only discovered this after completing many drawings, so I decided to live with it.

The second discrepancy affects all 1:36 ALCO and MLW hood units. The hood and cab are both 2 scale inches too tall above the walkway. This was the result of errors or inconsistencies between ALCO specifications for the frame height vs. railroad specifications for the cab roof height for the Century series. I only discovered this error when measuring a preserved M-636 in 2013, after I had already stopped work on the 1:36 series.

1:18 Series

November 2012 - Present

In November 2012 I started the 1:18 series, which is double the size of the 1:36 series (4 pixels per scale inch), making the finest increment 1/4 inch. While I aim for perfect accuracy in every dimension and detail, I can't make such a strong claim; nonetheless, the 1:18 drawings follow all major dimensions exactly and are generally accurate to within a fraction of a scale inch in minor dimensions. Any inaccuracies are the result of uncertainties or variations in the measurements I've made - not the result of "eyeballing" the dimensions.

As before, I now draw in the new scale exclusively and have made no more 1:36 or 1:55 drawings since the start of the 1:18 series. More information about the making of the 1:18 series can be found in the Development of 1:18 series and Creating the Drawings pages.

Show Image

To a greater extent than with previous drawings, I measure everything. I've drawn many common components - trucks, couplers, horns, air dryers and even air brake control valves - based directly on patent or manufacturer engineering diagrams. When measuring from photos, I take repeated measurements until I get results within 1/2", or preferably less. Occasionally I've reached a stumbling block where one component (such as an ATS shoe or a unique antenna set-up) slows me down - or even prevents me from finishing a series of drawings. If I can't find sufficient reference material to make an accurate drawing (at a minimum, the major dimensions and a few good high-resolution photos) it doesn't get drawn in 1:18.

Despite my best efforts, inaccuracies can still creep into the drawings. In early 2019 I discovered that the majority of my EMD locomotive drawings were affected by a dimensional error with the front hood seam, which was 1" too close to the cab. As a result, parts of the hood were off by 0.25" to 1", and the dynamic brake fans and exhaust stack on SD units were off by almost 2". As I deemed these errors too high for the 1:18 series, I decided to correct them - affecting a total of 915 drawings in the biggest drawing update since the 1:55 series. A second massive update arrived in late 2020 and early 2021 to roughly 2,000 drawings, when I corrected minor dimensional errors in the handrails and (on some units) sand lines and fuel tank details, which I had not accurately measured when first drawing them.

Sources of Dimensions

Diagrams and real-world measurements have benefited many models in the 1:18 series. With only a few exceptions, the diagrams I've used for reference were freely available online and I simply found them from many hours of browsing.

The 1:18 GP30 drawings - among my earliest 1:18 drawings - were the first locomotive drawings on the site that I derived almost entirely from EMD mechanical drawings, including cutaway diagrams of the trucks, underframe, cab and most of the hood (supplemented with photos). In 2013, I also started taking detailed real-world measurements of preserved locomotives; the 1:18 MLW M-636 drawings were the first drawings to benefit from such measurements. Several series of 1:18 drawings (mostly MLW locomotives) have benefitted from these measurements, and several other models remain to be done.

When I don't have specifications or measurements to use, I derive measurements photos - or from related models, such as extrapolating a few hood dimensions from SD45 diagrams for drawing the GP40.

mlw m636

An illustration of some of the measurements I took of a preserved M-636 for use in the 1:18 MLW M-liner drawings. This locomotive benefitted from more detailed real-world measurements than most of the other locomotives I've drawn. For more measurements, see the Locomotive Dimensions page.

Painted Drawings

In the past, I matched paint colours by eye. However, starting with the 1:18 series, I match the colours directly with photos. I take a 16 x 16 pixel sample from 5 to 10 properly exposed photos showing clean, unfaded paint, and determine the combined average RGB values of all the samples. I then increase the saturation very slightly and adjust the luminosity (slightly lower for dark colours and slightly higher for lighter colours) to account for dust or glare. If I find a very good online paint sample, I give extra weight to the RGB values of that sample. The colours I use in the 1:18 drawings are listed in the Colours page.

The image below shows colour matching for the CSXT YN2 paint scheme, based on samples from eight photos. The RGB values in brackets indicate the sample average, while the second set represents the final colour.

colour matching

I also keep a list of fonts and a selection of logos (listed in the Lettering and Logos page) that in many cases can be simply copied and pasted into the painted drawings. In some cases, I locate the lettering and logo by scaling a photo and drawing on top of it, if a suitable photo is available. (I take many of my own photos specfically with drawing or painting in mind). I've also been building up a collection of maintenance and warning stickers, many of which are small enough that they only show the shape without legible text.

VIA F40PH-2 drawing

A comparison of my drawing of VIA F40PH-2D #6438 with one of the photos I used for reference (click for a full-size version). This was an ideal drawing scenario; most of my painted drawings don't benefit from such clear reference images.

Drawing Scale and Measuring

The number listed for the drawings (1:36, 1:18 and so on) ultimately has little meaning on the screen beyond establishing the relative size of the drawings - because different monitors have different pixel resolutions. The drawings are scaled at a fixed pixel-per-inch ratio, and the scale is based on a print resolution of 72 PPI. This is to maintain consistency with other drawings on the web (such as the Railroad Paintshop) which are often scaled based on a 72 PPI resolution. In terms of the actual size of the drawings, the 1:36 and 1:18 series are drawn at 2 and 4 pixels per scale inch respectively, so at a given printing resolution, they are exactly 1:(PPI/2) and 1:(PPI/4) respectively.

Obtaining dimensions from the images

For drawings in which I haven't listed dimensions, given the fixed pixel-per-inch ratio of the drawings, obtaining dimensions from the 1:36 and 1:18 drawings is very simple:

The 1:55 drawings use a more obscure number for the scale - 16.2 pixels per foot - and frankly, many of them (at least the older ones) aren't worth measuring.

To obtain a dimension from within one of the drawings, use a selection tool that displays the selection dimension in pixels, and overlap only one of the two lines to account for the line width. The example below shows a measurement of the first hood door on a 1:18 EMD GP38-2 drawing at 400% zoom. The selection indicates a width of 73 pixels, or a scale width of 18.25". (The door is nominally 18" wide, with a 0.25" gap.)

dimension measuring