Card Modeling FAQ

Appendix: Designing Paper Models

There are many ways to design paper models, and what's collected here is not a complete reference or tutorial on the subject. However, it is hoped that the information here will give prospective designers some insight into how others set about the process. Much of the discussion here focuses on the use of computers in the design process. This is probably because many of the FAQ readers and contributors are computer users. However, it should be stressed that computers are not essential to the design process. After all, paper models were being designed a century ago. Nor is a computer-designed model necessarily superior to a pencil and paper design.
from Kell Black <blackk@apsu01.apsu.edu>: No software can compete with the thrill of using my old German drafting set to layout a card model. The velvet lined leather case is more handsome than my computer screen, and the various compasses, dividers and triangles are much more fun to use than the electronic palettes in Touch 3D. The price was right, too: $23.00 at a flea market.

It is also not necessary to be a mathematics wizard to design paper models. Mathematics is often useful, but there are graphical or cut-and-fit methods that can accomplish the same tasks as the equations. However, it is this editor's opinion that there is a general concensus among experienced designers that you will not go wrong if you consult a basic text on technical drawing and familiarize yourself with the basics of drafting, three-view and perspective drawings, and how to determine dimensions from scale drawings. If you wish to be able to model curved surfaces, look for a book on sheet metal layout (the older editions are usually better, in my opinion) and look at some examples of how such surfaces can be laid out. Consult the bibliographic sections of the FAQ for some specific texts.

Regardless of how you design your models, the process will require some patience and a lot of trial and error building.

Examples of the design process

from Matso Limtiaco <matsolim@wolfenet.com>: I've designed two large-scale (1/144) cardstock models of US Navy airships, the Shenandoah and Los Angeles, and had lots of fun doing so.

Basically, I start with a 3-view drawing, and get measurements for hull heights, widths, and section lengths in the right scale. I design the hull frames and large paneled sides using an ancient version of MacDraw, and print these out as patterns.

Using removable glue stick, I stick the patterns onto medium-weight poster board (Mead 10-pack for three bucks!), and then cut and score as usual. I build up the airship hulls as large polygons with center frames (reinforced with foam board) that are joined with tabs. Once the complete hull is finished, it gets a light coat of gray primer, and then two light coats of silver enamel (I'm going to try silver dope on the next one).

After the hull is complete, I design and assemble the gondola, engine cars, and fins of simple box construction. These are then attached at the appropriate points. I use a pair of foam board "cradles" to support the hull off the work table so these more delicate parts are smashed. The fins are of the same kind of poster board, but for engine cars, struts, and smaller gondolas I use index card stock.

Once the entire airship model has been assembled, it's time for insignia. Because I couldn't find ready-made decals of the correct size, I created the US Naval symbols on the Mac and printed them on a color laser printer - then had this copied onto a sheet of decal film and applied them to the model as usual.

If anyone would like more info on how I designed these models, or would like to see a couple of pictures, please Email me. I'm certainly looking for anyone with design experience to suggest a better way of doing it, as I think my method is quite slow.

As for my scratchbuilt designs, they are more or less built like the old Alan Rose Hindenburg model - cross-section formers (reinforced by foam board) supporting the hull surface, with simply constructed fins, gondolas, etc. I'm learning more about how to work in this medium as I go.

So far, I've figured out how to measure adjacent hull sections so they fit together better. I had to rebuild five or six different sections on my first model - trial & error at its finest - because they wouldn't fit together. So far, on my latest ship (USS Akron) , the fit has been very good and I haven't had to redesign anything.

I would advise that my methods are quite different than building a kit, so I have more room for mistakes. I build out of cheap poster board, and spray paint after the hull has been completed. So I can afford a small glue smear here and there because it'll be covered up eventually. Insignia are applied as decals, which are produced using a draw program.

from Peter Richardson <par@cct.u-net.com>: I am putting together an item on design for the FAQ which I am sure will be of interest. Essentially, I flat-bed scan from appropriate drawings, import the scans into the background layer of Freehand 5, scale appropriately, then draw over the scans with the bezier tool, curve smoothing as appropriate. The "tracing" then can be re-scaled, re-dimensioned, 2D tweaked as required to translate into component developments. These are printed and test assembled. With experience and some simple geometrical and maths calcs (PiR2) I can now produce most component developments on screen immediately without having to test build assemblies. The first models I produced for the Hooton AirCraft range required loads of prototype assemblies to get them right (The Rapide took 8 full models with countless trial sub-assemblies before I got it right - took about two months - the latest Hawker Harts came straight off the Mac in a week! ) My biggest challenge is not the models themselves, but the instructions. For these I use a combination of Freehand and Infini 3D to draw isometric and perspective line drawings from the previously computer generated component drawings. This is a bit specialist, and wouldn't really be of use to Joe modeler.

I am presently developing some architectural models. If I haven't got measured drawings to use the Freehand trace from scans technique I use Apple Quicktake multiple images of the elevations of building to be modeled, composited together through Photoshop/Quark Express (which is quick) and EPS transfer into Freehand, then trace. I am producing a range of facades of buildings in plain white card which are to be mounted on backboards for framing and wall hanging as 3-D pictures. Avoids the problem of displaying 3-D whole buildings!

Dr. Zarkov of ModelArt has provided the following account of the stage in the design of a paper model. ModelArt's designs rely heavily on CAD techniques--this is by no means the only way to do designs, but Dr. Zarkov's discussion does bring out some of the advantages of the computer in the design process. For instance, note how straightforward it was to produce a variant with different markings.
from Emil Zarkov <modelart@tusk.icn.bg>: Sometime ago I proposed designing a group project - the IAR 80. But from my point of view there was not enough interest and I finished the project myself. I'll explain with details what I did:
  1. I found all available materials concerning the project.
  2. I scanned side, top/bottom and front view drawings of the airplane, saved them as TIF files and imported them in AutoCAD R14. Then after some scaling, rotation and moving I prepared the base of the design - the raster image drawings that represented the aircraft in 1/72 scale, allowing me to draw over them.
  3. I drew model drawings on this base, referring also to the information from the pictures of the real thing and correcting some mistakes in the base drawings.
  4. I drew 3d polylines that determine the shapes of all segments of the model's parts - fuselage, engine cowling, wing, tail, cockpit, bombs etc.
  5. I drew the theoretical shapes of the cutouts, using my ModelCArD addition to AutoCAD, and created all the intersection lines between them where it was needed. For example, I defined the wing with two 3d polylines that coincide with the center and the end profile of the wing. Then I created the intersection line between the wing segment and the two fuselage segments, obtaining very accurate fitting. I created the wing loungerone in the same way, defining a flat surface with two 2-vertex polylines - one in the center and one in the end of the wing. I produced loungerone lines simply by finding the intersection lines between this surface and the wing segment. Using helping surfaces I obtained the defining 3d polylines of cowlings between wing and fuselage (much easier to be done or shown than to explain.)
  6. After finishing the geometrical design of all parts, I started the preparations for coloring the kit. AutoCAD is very convenient for drawing, but not for coloring. For coloring of the model I used CorelDraw. First of all I arranged the cutouts in the sheets with preliminary determined format, then drew the lines, signs and camouflage patterns on the parts (still in AutoCAD). For obtaining smooth borders between green and brown patterns I created 3 offset lines across each border (one at 0.15 mm and two at 0.3 mm distance from previous one). After that in CorelDraw I filled the green and brown closed patterns with corresponding colors, setting their border lines to none, and making the created offset lines thickness 0.3 mm, setting their color with interpolated values in corresponding order. Thus I avoided using input of the bitmaps in CorelDraw, keeping all the design as vector objects. This gives much smaller in size PDF files of the final downloadable version of the model and avoids inappropriate color calculations of raster images during printing.

    The last needed preparation in AutoCAD is making a rectangle over the parts for exporting with fixed known dimensions. It is needed due to the bug in the import filters in CorelDraw (present even in release 8).

    When all that was over I exported the cutouts as DXF files (selected objects only in ACAD 12 format).

  7. I prepared the CorelDraw layout for the kit. It is standard for 1/72 ModelArt downloadable kits and contains all the necessary information for high quality printing, including color tables for exact color matching. Inserting a temporary page with guidelines placed to the borders of the container, I imported the model's pages using snap to guidelines mode, adjusting the size of the container by moving and scaling, achieving the exact scale.

    From then on the things were easy. I put every page in its place, set the line thickens of every detail and filled the parts and set the lines with appropriate color.

  8. I produced 3d isometric views of the instruction drawings in AutoCAD, exported and placed them in CorelDraw project in the same manner.
  9. I made a test printing and building of the kit. Before this I made one more variant of the IAR 80 - IAR 81 dive bomber with different camouflage scheme and markings. It was easy - I simply mirrored the standard camouflage scheme and reversed the colors. I do not have exact confirmation for this color scheme of IAR 81. But with a high level of probability the real thing can be painted in this way. It was the normal practice of painting the camouflage patterns in this way during WWII. (Remember scheme A and B for English RAF fighters. If anyone of you [especially plastic modelers] has compared these schemes, soon you could find for yourself that they are with the same standard patterns, but mirrored and with reversed colors. You can examine pictures of my Spitfires Mk I and Mk II on Paper Paradise as illustration).
  10. After test building and making few corrections on the layouts I took some pictures of the models and inserted them in the final version of the CorelDraw files.
  11. I prepared PDF files and they are available now from Paper Paradise. It is only 1/72 scale variant of the kits. I prepared also 1/48 scale variants with interior and a transparent canopy but I have had no chance to publish them yet.
from Erik Johnson <engineer@pcisys.net>:
Getting Started.
Having built scale paper models over the years, I finally decided that it would be fun to design one, but which one? My library was bulging with 3-views, photos and plans of all kinds, mostly World War II. I found that scale construction plans in RC magazines are a fabulous source, because they have the outlines, cross sections and lots of details.

One ship that interests me is the German DFS-230 troop glider. I had modeled it as 10 ft span RC scale and already had a lot of data and pictures of it. Not only that, it would be ideal because of the simple lines. It is sort of a huge Piper Cub. Besides that, it was one ship I could keep an interest in long enough to get a design done. I decided that a wingspan of about 12 to 24 inches was a good size, so 1:32 scale was chosen. With subject matter and size known, the task began.

Drawings on the Wall

It is hard to extract accurate dimensions off of a small 3-view in a book, so the image had to be enlarged. I had done this before by shooting drawing in a magazine with a 35mm camera & slide film, and using a slide projector, which works great, but this time I wanted to use an overhead projector. I have also tried scanning a 3-view and importing into my CAD program, but found it easier shooting it up on the wall, measuring, and then drawing it in CAD.

I copied the 3-view drawing onto an 8x10 clear sheet for the projector, and shot it onto the wall. Since the drawing was only for sketching, old construction drawing paper was used. The back side is blank and engineers and architects throw them away by the ton. That created another problem. The copy process, or perhaps the projector, created a 'fish-eye' effect seen in curved spar lines that I knew were straight. To fix that, I first drew the centerlines, wing span limits, length and other known lines on the paper in red for guidelines. The next step was to move the projector in and out, and re-focus to fit the image within the wingspan lines. Because the drawing would be used to extract dimensions, size wasn't too important, as you will see later. The fuselage and each wing projected to about 36" long. Because it was impractical to draw the complete top view, the major parts, fuselage and wings, were done on separate pieces of paper.

To counter the fish-eye problem, the paper, not the projector, was shifted so that part of the red centerline would line up under the projected centerline. Moving the projector made it almost impossible to line up again on lines already drawn. Shifting the paper was far easier. I started with the top view, nose pointing up and worked down to the tail about 8 inches at a time. Paper was held to the wall by masking tape to make moving it easier. Lines and other details on each side of the centerline were lightly sketched in. A yardstick helped for the long straight parts. Since projected lines have width, I had to decide where to draw lines, and usually chose the center of a fat line. When the fuselage top was done, the sketched lines were connected and cleaned up. All of this was repeated for the wings and empenage.

The CAD Program:

There are two ways to create the drawings---by hand or by CAD in a computer. Having some basic CAD and drafting skills, I chose CAD. Besides, it gave me something fun to do and improve CAD skills at the same time. The computer is a fabulous tool for creating a precise drawing, but it is not a magic wand. Like a piano, it is just an instrument and the results depend on what you can put into it. If you can't draw, CAD won't save you.

I use a shareware CAD program called Draft Choice available from Trius, Inc. It really doesn't much matter which software you use as long as you stay with it long enough to get proficient at it.

The CAD drawing was set up to use dimensions directly from the wall drawings. I drew the plane as a 3-view and then created individual parts from it. If it was 23 inches on the wall, it was 23 inches in the CAD drawing.

With a wingspan of 24 inches, there were some details that would be difficult to model, so they were omitted in the CAD 3-view. When all of that was done, I set the wall drawings aside and continued on with the CAD 3-view file.

Designing the Paper Parts

The completed 3-view was then reduced to the final model size of 24 inch wingspan. This is why it didn't matter what size the wall drawing was, and is another reason why drawing in CAD is easier than working on the drawing board. Since fit of parts and paper thickness were major factors, it worked out better as the final size of the model. If a part was 1/32" short of fitting, I changed the part by 1/32" in CAD. This is different than the normal CAD practice of drafting in the dimensions of the actual airplane (72 feet span), not the model.

Fortunately the DFS-230 fuselage is mostly straight lines, so the geometry and flat projection process was a easier than with a complicated design, like a Spitfire. It is a real challenge to layout a flat projection of a compound curved surface. It can be done, but it's not a good choice for the first model. Like one old engineer once told me, "Take baby steps first, you will learn to run later."

The wings and the other major parts were printed out on plain paper. I developed the general layout of the design by marking up these pages with the probable location of the lines to separate the paper sections that composed the fuselage. This is where it started to become a paper model. At this point, it became obvious that the layout of the DFS-230 canopy area was wrong in the 3-views, and the whole front end would have to be designed, so back to the wall drawing. Keep in mind that many 3-views are created from minimal information and errors are quite common. Because of that, they also conflict each other, so try to use photos of the real ship to sort it all out. Looking at the photographs again, I discovered that all of the cockpit sides were flat so the fuselage below had to have parallel sides. When I got the Smithsonian microfilm, that observation proved to be right on. This explains why I had so much trouble with the earlier RC model, oh well.

I tried drawing the typical paper fuselage sections, one after the other, but it was too hard to keep track of how they all fit together, so I copied a piece of the 3-view off to the side and worked with that to design each paper piece on the one CAD drawing.

Another advantage of CAD is printing the drawing out at any size you want, so I did it with some of the pieces that were having fitting problems. Cutting and pasting bigger trial pieces helped in clarifying the design. It was a constant process of choosing between a big piece, or several simple ones. There was alway an option. Printing on regular fan-fold paper in the dot matrix printer was quite efficient and cheap. Pieces were thrown away after revising the CAD drawing.

The design itself pretty much followed techniques other paper kits use, and I soon found out why some of them were drawn the way they were.

As the design layout began to take shape, printing the pieces at the final size became important. 8.5 x 11 size paper was the initial choice, so each page of the project would have a box border of specific size so it would fit on the paper and contain as many parts as could be put in it. Since I wanted to use my own ink-jet printer, use of larger sheets, such as 11 x 17 inches was impractical. Besides, there aren't very many paper models out there that couldn't have been done as 8.5 x 11, but those designs went straight to the commercial printer and the paper options available there are quite different.

The 3-view became surrounded with these 7.5 x 10 boxes filled with fuselage surface parts, glue strips, bulkheads and such. Each box was later exported as a new file for printing. It was important to do all of the designing and fitting on a single drawing, rather than importing individual CAD files back in to see if pieces will fit together. Printing these boxes also allowed for parts to be the same scale. If fit was a big problem, the parts were again printed at double size to work out the problem. Each time it happened, and it happened a lot, I got better and quicker at fixing things.

An especially useful tool is 'Rollation'. This is the process of creating a flat surface that will fit between two cross sections. The fuselage skin between two bulkheads is a perfect example. See the Paper Model FAQ for more on how to do this.

Several prototypes were completely assembled to check fit and ease of construction. Things were constantly changing to make it better. Note that all of this was with a 24 pin dot matrix printer and 65 pound paper (cover stock). Camouflage coloring and an ink-jet printer were yet to come.

Another issue that came up was keeping track of drawing versions as changes were made and then maybe changed back. After losing work because I deleted the wrong drawing, I began numbering drawing sheets with letters, such as '7-d.dch', which means sheet 7, try number 'd'. The dch is the Draft Choice default extension. To keep track of what 'd' meant, I had a notepad with remarks scribbled on it. When I was done making changes, all of the old drawings were erased to avoid mixing them up with the final version.

Finally it was done. The parts fit and it looked reasonably buildable by other people. Buildable, that is, if they had instructions. This is where most of the paper models differ. Some instructions are well written with lots of illustrations or drawings, while others are not much more than a checklist. Those from Europe tend to be short, because the hobby is so well founded, and the basics are common knowledge. Because of that, I decided to go with full instructions with line drawings for illustration. Still no color, but it had to be right before that.

Composing the text wasn't too difficult, but the drawings presented a challenge. The first attempt was with freehand sketches scanned and inserted into the text as BMP files. A freehand sketch looks really amateur and shoddy. To combine text and drawings, a good word processor is needed, and I used Microsoft Word 2 for Windows. If you don't have software that can do this, now is the time to move up. There are a lot of them out there. Anyway, the manual came together and I decided on double column format to make it easier to read. Now I had to get the drawings to look better.

Remembering the ability of the CAD program to import a BMP file as background, I borrowed a digital camera and photographed the pieces alone and partially assembled. A photo was imported into a new CAD file, and I traced lines over the edges and important lines. This also accounted for perspective and angled curved wingtips. The CAD drawing was saved and the BMP background erased. To keep my terminology straight, I call these construction illustrations 'Figures'. The drawing was saved as CAD and then exported as a BMP file. Well, red, green and yellow CAD layer lines is not usable in the black & white manual, so I went back to export the BMP as black on white. That was nice, but the lines all but disappeared when reduced to 2 inches wide figures in the manual. Back to CAD, I converted all lines to 'fat' before going the BMP export. CAD programs let you define line thickness from the default 'thin' lines. That did the trick, but the figure was cluttered and had lots of unwanted white area all around it. That got fixed by editing the BMP file with Microsoft Paint, which comes with every copy of Microsoft Windows. After editing and inserting part numbers, arrows and such it was imported into the instructions file opened in the word processor.

The editing process was to do the usual paragraph breaks, word underling, bold, size changing, and insert all of the figures. This also met the goal of keeping everything computerized and easy to duplicate for other modelers. Cutting and pasting the instructions for the copy macine would have worked, but I also wanted to improve my computer skills.

Now it was time to color the drawings. Using several bit map painting programs showed it could be done, but it was hard to do and saving the colored drawings as BMP files introduced jagged lines and other unwanted features when printed. The answer was a graphics illustrator, which for me was Corel Draw ver 3. It is fairly old considering that version 9 is out now. Anyway, there were problems getting the CAD drawings into Corel Draw. It could import the common DXF format, which I knew, but Corel Draw had trouble reading what the CAD program produced. I don't know what the problem is, but experimenting around, I found that exporting in HPGL pen plotter format instead of DXF worked just fine. Once in Corel Draw, the surrounding box once again came in handy, as Corel Draw can enlarge or reduce the image to fit the chosen printed paper size. I did that to get the 7.5 x 10 inch box exactly to that size. I added the colors next, one sheet at a time.

Coloring is straight forward and in many ways easier than in a BMP paint program. To get the colors in the right place, I went back to one of the white prototypes I built and carefully marked it up with a red pen to locate where camouflage lines changed, what color they were, and the measurement to nearby lines or borders. Each part on each sheet was then colored in and finally each drawing was complete. I had the paint chips, so printing a few color sample sheets showed which colors matched the chips. What is seen on the screen varies dramatically when you adjust the brightness and contrast, so creating sample color printouts to compare with your paint sample is very important. One other thing that came up was my setting the screen resolution higher than what my version of Corel can produce. It does 256 colors, not 2 million. 256 colors is enough, but not matching the screen/printer resolution can cause unexpected printing results.

A color laser printer is fantastic, but a color ink-jet printer will also work. The Paper Model FAQ has a great discussion area where builders talk about the printers they use. Definitely check that out before you run out and buy some strange printer. Not having a laser printer, I pressed on with my ink-jet.

I knew I could build it, but could someone else? That meant using 'beta testers' to build it. That was frustrating, because the first volunteers put it on the shelf or stashed it away somewhere. I think half of the entire modelling hobby is collecting unbuilt models and never building anything. I should talk, I have a garage full of unbuilt kits dating back to when I was 8 years old. Finally several people actually did build it and passed on comments. Nobody likes criticism, but in this case, I actually welcomed it.

So here it is, a model that someone else can build and something nobody else has done before in paper. Like a lot of builders before me, I have to decide how to make it available to others. We shall see. I still have that list of 29 other models that have not been done in paper, so I better get back to work.

from Joseph M. Cangero <jcangero@ix.netcom.com>: I need some help. I don't have any sophisticated CAD software, so I'm doing this all within PaintShop Pro 5.03 -- and it's my first crack at designing a model. I've just completed the design of the forward mount and 13 1/2" I-beam rails for the lifting bed of the Miellerwagen for Ralph Currell's V-2 Rocket. I intend to make it fully articulated and functional. The ideas are just swimming in my head. I've made some test assemblies and they are unusually light and strong. I've gone through a lot of measuring and transferring those measurements into PaintShop Pro. Here's a little summary of what I've done and then my questions:

My first designs were made in an 8-1/2" X 11" image file at 90 dpi with one-pixel wide lines. When the rough designs were done, I increased the image size by 400%, copied the images and pasted them into one layer of an 8-1/2" X 11" image file at 360 dpi. I then created another layer and traced over the images using a one-pixel wide line. When the file prints, the lines are thin, but very crisp with virtually no pixelation at all (just what I was looking for). I made a copy of the black and white draft image and saved it as a gif at 256 color depth. The resultant file was 131KB in size. Since the Miellerwagen will eventually be composed of various shades of gray, I may even be able to decrease the color depth further by the time I'm done.

Is a gif file the way to go or should I consider a compressed jpg file to reduce the size further? Since I'll only be using the most basic of coloring, will the image quality suffer much? I'm particularly concerned about losing the crispness of the lines. I'm comfortable with what I'm doing now, so I'm working directly in a 360 dpi image file from now on. Considering all I have is PaintShop Pro, is there an easier way to create the design drawings?

Since I'm working strictly from photographs and Ralph's model of the V-2, consistency of scale is a real concern. There's a lot of measuring and eyeballing until things look "right". If anyone has any insight into how to keep a consistent scale under these conditions, I'm all ears. Any and all suggestions are welcome.

from David Jackson <djackso1@stny.lrun.com> I use Photostyler (similar to Paint Shop Pro) for my work from digital photos. A CAD program would be OK if the models are just for yourself and not for distribution because receivers would have to have same program in order to print them out. The Cad program I use for other things (DesignCAD 2000) has the capability to convert from vector to raster format, but the results are lousy. Think they just take a screen snapshot.

I use .jpeg format to minimize file sizes. I have trouble trying to tell the difference between a gif, tif, bmp or high quality jpg. The .jpgs can then be converted to .pdf (the apparent format of choice for this list) for distribution. I use Photoshop for that. AH, there's the rub. Paint Shop Pro 4.0 does not handle .pdf format. Maybe 5.0 does? One other comment, since this model is mostly black and gray, why not convert it to gray scale and save even more space. Or have an unfilled version where users could do their own hand coloring.

from Robin Day <rday@pangea.ca> If the results of the JPEG are too rough, decrease the amount of compression. Depending upon where the JPEGS are saved from there is usually a dialog which allows the user to set the amount of compression. For black and white files, they should be saved as Greyscale format (single 8-bit channel for black) and they can then be saved as JPEG. For example, from Photoshop, when the Save dialog is closed with the JPEG format selected, a dialog will open with a slider setting the amount of compression vs image quality. Experiment with these settings to determine the highest amount of compression possible while maintaining image quality.
from Gunnar Sillén<gunnar@modellbyggare.com>: Is it really necessary to use the JPEG for a b/w or greyscale drawing or picture? Or is it actually possible without losing too much? When I have tried I have just got my drawings turned into less organized heaps of scattered black dots. Maybe I have done something wrong (as usual) but my opinion is so far that JPEG is useless for b/w drawings.
from Karl Guttag <karlg@topher.net> JPEG generally does a poor job of "sharp" high contrast edges. Furthermore they tend to add "block/square" artifacts. "Artificial/man-made" objects and line drawings are full of sharp edges. Among other things, JPEG tend to cause "ringing/rippling" (known as aliasing due to errors in the high frequency of the image) around the edge the first time an image is compressed and then each successive pass makes it worse. Even with a "high quality" JPEG the image will degrade fairly rapidly. Current JPEGs are based on block DCTs and behave fairly badly (the next JPEG standard version, last I heard, will be based on Wavelets which are better, but not perfect either). But note, any lossy algorithm (not matter what the claims) will have "generational" problems.

In contrast, loss-less algorithms almost always work best on high contrast and flat shaded (single color) objects. Wanting to keep sharp

From what I remember, a "typical" photograph can only be compress about 2.5 to 1 to about 3 to 1 with a good loss-less algorithm (my experience is that none of the "standard" loss-less algorithms are very good at getting a good compression ratio of a "natural" photograph). At about 7 to 1 compression of a typical photograph, few humans could tell the difference after a single compression, but the image would degrade after many passes.

Generally, I find for most purposes, one can do fine with an input of a JPEG with a low compression ratio (higher bits) of about 7 to 1 to start with. Then all re-touching work has to be done in a loss-less method (I use Photoshop's PSD which seems to greatly expand the file size). Then out putting to JPEG for distribution base on the needed output quality.

from David Hathaway <david@dhathaway.freeserve.co.uk> I would suggest a different approach, based purely on my preferred way of working. I would always draw as complete a plan as possible as a first step. The reasons for this are multiple: With existing plans that can be scanned, you can import the scan to the tool, then take measurements or digitise over the top of the scanned image. Working from the plan, then design the model. May seem a lot of work up front, but the results in terms of consistency and speed of generating the model from the plans pay back the investment.

I would always use a vector-based drawing tool:

I use a second-hand copy of Corel Draw V6.0 (cost me UKP£20), a free copy of Micrographx Draw v5 and the free drawing tool Mayura Draw (converts .ai to .pdf).

In the case of the V2, the model of the rocket can provide the reference points for size/scale - so you probably need accurate plans of the rocket too. I would draw the rocket; then draw the trailer--taking measurements from a number of photos until complete; then design the model.

from Robin Day <rday@pangea.ca> I agree wholeheartedly. Vector based applications are the way to go. I use Adobe Illustrator and/or Macromedia FreeHand (these two are very similar in function and use) for most of the construction files for the models and Adobe Photoshop for the colours/shading.

Unfortunately, these are very expensive packages, although a reasonably priced bundle including Illustrator and Photoshop is available from Adobe. Although first designed for the Mac, there are PC and Unix versions of these packages (except for FreeHand - PC and Mac only). These are the packages that graphics professionals have been using since the bad old days of paste-up and colour mock-ups.

If you have to make a decision on Illustrator vs FreeHand, in terms of card modeling, Illustrator has a couple of small advantages. Primarily in relation to the rotation tool: FreeHand will only rotate in 0.1° increments, but Illustrator allows for 0.001° increments. Quite important for those angled pieces. Illustrator also places points in the center of primitives (rectangles and ellipses) and this makes it easier to relate to the centers of these types of objects. Other than these two small details, both packages are equal in ability and function.

As far as file formats are concerned, a non-compressed format for the construction files is in order, but as far as the final file, it depends on how it is being created. If Adobe Distiller or QuarkXPress & PDFWriter is used to convert the file to PDF, the images are converted to whatever compressed format is desired automatically and are embedded in the PDF file. Therefore, all you need to keep are the construction files. If money is tight and you can only buy one of these packages, I would recommend Adobe Illustrator (although I don't think I could continue living without Photoshop) since it is vector based and exports to a variety of formats.

By the way, you can make measurements from photographs. Not very accurate, but a good way of judging proportions. The difficulty lies in the foreshortening or persepctive effect of the photo. Assuming that a normal focal length was used to take the picture, the distances as they move away from the picture plane will compress by a progressive proportion. If one of the distances along the desired axis is known, use cross-multiplication (proportions - similar triangles, and all that) to relate the measured length, actual length and the measured distance of the required part. This works reasonable well to about a three-quarter view at which point the distortion due to foreshortening becomes a little too great.

In terms of rotating parts, the best solution I have found is to connect two cylinders with a piece of copper wire running through their centers. Use a smaller disc at each end and bend or tie the wire so that it doesn't slip through.

from Karl Guttag <karlg@topher.net> Actually, with the help of computers, the perspective distortion of photographs can be used to create fairly accurate measurements (called by some photogrammetry). With a few known measurements and telling the computer where some straight lines are, the computer can figure out all the rest of the measurements. In fact, I think if the computer has enough perspective distortion in the scene and enough line segments the give enough X, Y, and Z clues, and some information on the film format, a computer can figure out everything from the where the camera was, the lens, and the size of anything in the scene.

But it is a LOT of work best left to a computer. By hand, it would be hard to get accurate results.

Also, Adobe Photoshop's "Arbitrary Transform" can be used to do a true perspective correction and does not just "stretch the object to fit." I think Photodelux also does it "correctly" but I have not checked it.

Developing dimensions of the prototype

from Kell Black <blackk@apsu01.apsu.edu>: Over the years I've developed a pretty reliable method for recording the info I need to make a model. I pack a camera, a notebook and a retractable metal tape measure for photographing possible modelling subjects.

Using the tape measure I mark off a perimeter around the object to be photographed, say 20 feet around a bulldozer. This is the most important step! If all photographs are taken from the same distance away from the object, then all the photos will be in proportion to each other. I then reset the tape to 3 feet and place the tape on or as close to the base of the object as possible. This serves as a later scale reference. I then move around the object, taking a pictures from all sides. If there's LOTS of detail I'll mark off a new perimeter, set the tape to 1 foot, and repeat the process. In the notebook I jot down the measurements, sketch any details that might be hidden in shade, etc. For my current project, a 1:30 excavator, I took about 30 snapshots of a machine that was parked next to our motel in Pennsylvania.

Design techniques

For some more mathematical and graphical techniques, see the Appendix, Maths Primer for the Card Modeler.

Pattern development

Pattern development is the process of `unrolling' a three-dimensional surface to a flat pattern. This process can be done either with conventional drafting techniques or with specialized CAD software, but even if you plan on using CAD techniques, your understanding will be enhanced if you learn some of the basics. Pattern development is also used in sheet metal work, and there are good books on the subject. Consult the bibliography sections of the FAQ for more suggested references.
from robin day <rday@pangea.ca>: The best method I have found for creating compound curves lies in a combination of sheet metal techniques and polygonal modeling techniques from 3D software. Using an orthographic drawing, break the curves into sufficient number of straight segments along the axis with the least curve. For example, in creating an engine nacelle, the leas curve is along the profile, not the cross section.

Once the triangles have been drawn, use projection techniques to find the true lengths of the triangles so that they can be drawn flat. Choose a line for the seam of the object and begin building the triangles from the information determined from the drawing. As the triangles are created, they will create a flat representation of the part which can then be used as a guideline for drawing the curves required to create the part. Keep in mind that since straight segments were used to draft the shape, the resulting part will be somewhat smaller due to truncating the curves. This will only have an impact on fitting cross sections, however.

Although this technique can be very tedious, it is the mos accurate I have found without resorting to complex geometry formulae which can create problems in joins, or, heaven help us, differential geometry. Any other techniques out there? or perhaps someone has a way of streamlining this one so that it isn't so tedious.

from David Hathaway <david@dhathaway.freeserve.co.uk> I am currently using 2d packages that basically allow me to use my school technical drawing skills to draw out plans and derive the flat shapes that make up the solid models from them. This is fine for regular shapes e.g. blocks, cylinders, cones; intersecting regular shapes; singly curved surfaces; etc. However it does not get me as far as accurately laying out a shape that bends and twists in 3 dimensions - eg the side of a ship. Think of the bows of a ship curving in and the waterline curving in faster than the deck level.

This can be done by prototyping/rolling the shape onto a piece of paper, but there must be a way of doing this on paper using 2D tools? That must be a generic enough problem to have been solved by the draughtsmen of old.

from Robin Day <rday@pangea.ca> I use traditional projection techniques to reslove compound surfaces into triangles. The true lengths of the triangles can then be determined using still more projection methods. The triangles can then be drawn flat to describe the shape. I am sure there is a much better way of doing this, but this is how I do it.

Tedium is the main drawback to this process as it takes a long time.

from Tazman <jhuffldn@gateway.net> Robin is correct. I've been a drafter for over 12 years and this would be the simplest method, however boring it may be, it will yield an accurate result. Any standard mechanical drafting instruction manuals will teach you about projection if you don't know the technique. If you can get your hands on some 3d modeling software (shareware is available) that might be an easier option.

Rollation

Rollation is a mechanical method of pattern development.
from Kell Black <blackk@apsu01.apsu.edu>: I'll use an airplane for an example since that is my biggest modeling interest. To make the pattern for a section of fuselage from balsa kit plans, first cut out two consective cross sections (fuselage formers) out of heavy cardstock. Make sure that you maintain the center of these pieces as given in the elevation drawing. Using the plan view, cut out the connecting fuselage contour. Mount this at right angles to the two formers. You now have a skeletal section of the fuselage; two formers connected by the silouette of the plan. Looks kinda sorta like a cardboard barbell.

Now, get yourself some chalk and a piece of black construction paper. Dust the edges of the two formers with the chalk. Carefully place the contraption on the construction paper, and roll the whole thing one complete revolution. Viola! You have just traced a flat pattern of the fuselage section bounded by the two formers. Make sense? This is the kind of thing they teach in junior high shop classes. I guess instructors figured that a bunch of adolescents couldn't get their hormones to stop raging long enough to learn the math, so this was invented. However, it works! Make sure that you use really stiff board, and that the formers are centered and attached EXACTLY at right angles.

Here is more on rollation.
from Erik Johnson <engineer@pcisys.net>: I started the He-177 in 1:50 and did a fuselage section built to 2x model size, about 3" dia x 8" x 1.5" dia. It is a square bulkhead, with rounded corners, down to a circle bulkhead. For material I used cardboard from a dry cereal box and added stiffeners to keep it from flexing while rolling. Edge of the bulkheads were lightly coated with Elmer's to seal and when dry I added stamp pad ink (VERY carefully with a Q tip). It rolled out nicely. I don't think it would have worked very well at the smaller 1:50 model scale.

Specific Tips

Curved joints

from B.J. George<bgeorge@gna.net>: I would like to know where is the best place to design tabs on automobile tires? I've seen them of the threads (and you glue the round tire to it) or on the round tire (and the threads are wrapped around them). I seem to have trouble making a good looking tire either way. Which is better?
from Maurice O'Brien <mnob@picknow.com.au>: For "curved" joints I prefer to forget about ordinary tabs. Instead put a crease around the curve that follows the curve exactly, i.e. not a succession of short straights. Every 1,2,3 mm or so (depends on the size of the bits) put a single cut at right angles to the crease and don't bother with "V" cuts as they are a disadvantage. You now have a series of near rectangles which fold over very easily and of course will overlap each other (on an outside curve anyway.) Doesn't matter because I only put glue on the edge of the "tabs" next to the crease and the other ends of the tabs sort themselve out and they are out of sight. Gives a much more curvy curve.

Recommendations for CAD software

from Lou Coatney <mslrc@uxa.ecn.bgu.edu>: ModelCAD, by ViaGrafix, the guys out of Pryor, OK. $100.00 ... and accurate to 7 digits. It's what I do my ships with and, it gives you everything conceivable a 2D program should ... and that's for *commercial* use, if you wish.

from Bob Bell <bobbell@portal.connect.ab.ca>: I do a lot of designing myself, generaly of steam and diesel locomotives which are my main interest. I use Corel 5 which gives me snap to accuracy to two decimal places and the only math formula I have had to use is pi d (3.14 x diameter). I use plans from model railroad magazines which gives three views and as they are already the correct scale I can take measurements directly from them. First I scan and save the plans after cleaning off all unnecessary lines and info, so that I can keep calling up these diagrams and erasing everything except the part I am working on. For example if I am designing the cab I open the diagrams into Corel Paint and erase everything except the cab sides, front, and back. These I then cut and past into CorelDraw where I can assemble then into a flat piece that will assemble into the 3D cab. Instead of measuring with a ruler, which can cause errors, I use the diagrams imported into Coreldraw so that I can visualy adjust the ruler guides in the program to position around the sides and top and bottom of the part then draw a rectangle which snaps to the rulers thus giving me the exact dimensions I need. I have also made my own clip art file of loco parts such as lettering and logos, different styles of wheels, different styles of sand domes, etc. which I can use on different loco designs. This saves a heck of a lot of time. Those of you who design buildings might want to use this idea for different styles of windows, doors, brick and siding, etc. I store these images on the windows card file and I can easily drag and drop them into the project I am working on.

Anyway this probably does not make much sense to most of you. What I am trying to say is that you don't need to be a math wiz or a computer scientist to design your own models. All you need is a good design program (I prefer Corel Draw), a good photo editing program, a scanner (preferably flat bed) and good diagrams which can be found in all model magazines.

from Emil Zarkov <modelart@tusk.icn.bg>: I would like to tell you some additional words about ModelCArD Software package that I'm using in my cardmodel designs. I wrote it myself a few years ago and 'til now I'm its only user. Recently I wrote all the necessary user interface and hooked it to AutoCAD 12-14, so now two variants are available - stand-alone application and AutoCAD extension. I personally prefer and use the second.

Comparing ModelCArD to Touch-3d (I didn't test Touch-3d because I haven't access to a Mac) there are a few differences:
Touch-3d allows 3-D modeling and provides all steps of the creation of relatively simple models.
ModelCArd helps only to create the cutouts of already designed 3-D models with the help of another 3D application - AutoCAD for example. The stand-alone version uses input data as ASCII XYZ list with coordinates or DXF file of 3d polylines that determine the part designed. As an output ASCII and DXF file with results is generated. For example the module that creates cutouts needs two 3D polylines that describe the ruled surface of the part as input and generates file with two 2D planar polylines of corresponding cutout us an output. The module for calculation the intersection lines between two parts uses four 3D polylines that describe the two parts and optionally four 2D polylines that describe their two corresponding cutouts. Then a list with both 3d for the model and 2d for cutouts polylines is generated as an output. Creation of the input data and utilization of the output data are up to user. It may be done with every 3d package that allow works with 3D polylines.

The ModelCArD is professional tool. It requires some skills and usage of the third party CAD software. Its main advantage is that it can process precisely very complicated cutouts that are parts of very large projects (My last kit F-15 E Strike Eagle contains 1711 parts!). For such purposes Touch-3D is not convenient.

As you see ModelCArD is not for beginners and is far from the convenience of the glittering commercial software packages. This, together with its restricted field of usage doesn't allow me to offer it till now as a commercial product. Of course I will do it if there is some interest.

from Claes Lundström at Lundström Design <ludesign@algonet.se>: Touch-3D is a Macintosh based integrated free form 3D modelling and unfolding program with the ability to unfold very complex models. The unfolding function is an integrated part of the program and essentially works as a view. This means that any updates on the 3D model will automatically update the unfolded pattern, which is very useful for optimizing the use of material. You simply change the model and click on the unfold view to see what happened. Large models consisting of many parts can be handled by using the built in layer function which treats the content of each layer as an individual part when unfolding it. Touch-3D also has built in editing tools for controlling the unfolding order (pattern) which is very useful for optimizing the use of material. The unfolding pattern may also change the appearance of the model, so this is really a key feature.

Touch-3D is not limited to just unfolding models generated in Touch-3D. It can read and write files in formats compatible with many other design tools.

Touch-3D is a professional tool being capable of handling very complex models. One of the original uses of Touch-3D was to design boats (real ones, not cardboard boats) where the available marine designing programs where unable to handle the complex shapes/unfolding patterns we where using. I have personally designed over fifty full sized boats with it. Our building technique resembles paper folding, but at a *much* bigger scale. Check out the Sailboat example on our web page where the single hull sheet is 10.757 x 5.684 m of 5mm aluminum. This is really king size paper folding. I also use Touch-3D to produce cardboard mock-ups where I plot directly from Touch-3D on cardboard using a flat bed pen plotter being able to handle cardboard. This allows me to produce an accurate 1:10 scale model/mock-up in a couple of hours (including drawing the model, unfolding, plotting, and assembling it).

For more information, visit the Touch-3D information site. Here you will find:

from Chip Fyn at Fiddlers Green Ltd <chipfyn@quest.net>: We here at Fiddlers Green create new models by, what I think, using the best from both worlds. Personally, I think there is (almost) no bigger thrill than paper 'sculpting' an airplane from a 3-view, photos and the actual plane. Working with pen & ink on a drawing board and then building test models is soooo much fun.

But, there's lotsa fun in coloring, resizeing and adding text with Adobe Photoshop and Claris works. It's best fun watching prepress color separations from Photoshop come out of the printer knowing that I just saved $80 a set.

from Thomas Pleiner <Thomas.Pleiner@t-online.de>: What I've found is that the step moving from the conventional way of drawing with a board, rulers, ink-pencils etc. to CAD is not as easy as most people think. Looking for a CAD tool which allows working in a very intuitive way I found the ASHLAR products helpful to get past that obstacle.

As the professional ASHLAR 3D product is at a very high retail price I may draw your attention also to ASHLAR drawing board - it's a 2D package at US$175 - a fair price for one of the most user friendly products.
http://www.vellum.de/download.htm: refer to this site for downloading a free demo of the latest ASHLAR 3D product (be aware that's a german site).
http://www.ashlar.com/high/db/index.html: refer to this (US) site to learn more about ASHLAR DRAWING BOARD.
But, also the expensive ASHLAR 3D package does not create flat structures from 3D drawings.

from Gary Au <garyckau@Glink.net.hk>: Did someone work with any Vectoration program for CAD? Please give me some suggestion and what will be the price and how to obtain?

from King Butler <kbutler@mandala.ca>: I presume you mean raster to vector conversion software. There are many choices. The top of the line, as far as I know, is VPMax or VPLite. It very sophisticated and has lots of features but the bad news is that it costs about $3,000.00US, although there is a "VPLite" version for about $1,500.00US. The good news is that they have a free demo version available. The demo does not expire and you can use it forever, but it is limited to a small area (about 1/4 of an A4 sheet, as I remember). I suppose you could do a larger drawing in sections and paste them together, but I've never tried it. I have messed with the demo, but I found the learning curve pretty steep, and the software was frustratingly unstable. It tended to crash if you did anything wrong. See their web site.

At the low end of the scale, TurboCAD is widely available for around $100.00US and does support raster to vector conversion. I haven't used it, so I can't comment on its capability.

There are also traceover programs available that may be quite inexpensive, but are very labour intensive. But then, the above ones are as well (labour intensive, that is).

from Gary Alexander <gary@gasp.u-net.com>: I am surprised that Adobe's Streamline isn't mentioned. I converted a whole books worth of scanned line drawings with that program and I would recommended it highly. It's much better than the trace tools built into things like Illustrator and Corel.

The CAD package form-Z can unfold complicated 3-D shapes into 2-D patterns. They say, "the unfold operation offers an effective method for constructing physical cardboard models." It's pricy, at least by the standards of the hobbyist, about $2000, and I don't know anyone using it for paper models, but I'd like to hear about it if you are.

from Ned Batchelder <Ned_Batchelder@iris.com>: I've had great luck designing models with Visio. It is more of a diagramming tool than a CAD tool, but it is up to the task. It deals with drawing scales, and can manipulate shapes in intelligent ways. The best part is SmartShapes, which are re-usable pieces that can be parameterized. For a model of my house, I created a window SmartShape that I used for most of the windows (and doors), even though they are all different sizes. I could adjust each one by specifying its height, width, molding thicknesses, etc. The latest release (Visio 5) allows you to design custom fill patterns, so you can create textures for bricks, roof tiles, lattice, etc, and fill shapes with them easily.
from Kell Black <blackk@apsu01.apsu.edu>: I draft most of my models using a really simple draw program, ZedCore Desk Draw. It's a dumbed down version of Illustrator, but I've been using it for years and have learned lots of shortcuts.

I also use Touch 3d for some things, but it's not nearly as fun as doing the math yourself. The price was right: only US$200. for a single copy with an educational discount. I bought the software about two years ago, and then the regular price was about US$250. Still a bargain compared to other packages. However, it's only for Macs.

from Darek Lipinski <TonClass@netcom.ca>: Personally I am using Rhino 3D and I love it. For the casual card model designer it could be a bit too expensive, around $800.00, but for all those interested in three dimensional designing software, you can hardly find any better. Actually you guys can try it for free, since there is a fully functional demo version available from the website.
from Joseph M. Cangero <jcangero@ix.netcom.com>: For those would be card model designers out there who, like me, can only afford a relatively inexpensive graphics program there is good news. I just received my upgrade copy of PaintShop Pro 6 ($49.99 for registered users of a previous version, $99.00 for the full version) and one of the biggest improvements has been the addition of vector objects and editing tools. Not only that, Jasc has also added support for several new file formats including AutoCad dxf files -- unfortunately they still don't support reading of pdf files.

Although the vector tools are probably not as robust as those found in programs like Adobe Illustrator or Corel Draw, they make working with vector objects pretty easy. Since I'm in the middle of my first card model design project, a meillerwagen for the V-2 rocket, the ability to work with vector objects couldn't have come at a better time for me. I'm in a lull right now awaiting some additional reference material anyway, so it will give me a chance to learn to work with the new vector tools in PaintShop Pro 6. For those interested in more info, you can visit the Jasc site.

Links

Back to Card Modeling FAQ
Copyright © 1997-2002 by Steve Brown. Copyright notice
Mail cardfaq@bellsouth.net | Steve Brown |