Eggcrating JSC hull designs

Discussion in 'Tutorials' started by Darwin, Oct 1, 2005.

  1. Darwin

    Darwin Member

    This has been stitting on the back burner waiting for some inspiration on how to create some illustrations for the text. With the new site format being so graphic-unfriendly without reliance on photo archives external to (which I personally don't care for, since links dissappear so easily into the ether, and if not done properly, turns attempting to load a forum string into a true nightmare if you are stuck with a dial-up connection), I figured what the hey? Might as well post it strictly as a text file, and see how well I managed the description using text alone.

    As demonstrated by the Helena build, even the older JSC kits can be enlarged and turned into a fairly decent large scale model (1:250 or 1:200). Although not exactly overwhelming, I have received some interest in the techniques I use to upscale JSC kits. This fits into the larger framework of the tutorial string I started for the Image Forge graphics program (or it will after I finish taking a sledgehammer to it).

    The process is broken into several phases: (1) scanning the hardcopy original, (2) enlarging the scanned files, (3) cleaning up the enlarged files, (4) creating a set of individual part files, (5) redesigning the hull framework, (6) redrawing the individual parts, (7) adding additional detail, and finally (8) creating new parts pages. Be warned that this process can take as much time or more than actually building the model. It all depends on your personal preference (and the quality of the original kit).

    To start at the beginning (what a novel idea)…..I assume you own, or at least have access to a flat bed scanner and know at least the basics of operating it. The main point for discussion in this tutorial is selecting the scanning resolution. This selection is going to be a compromise between image quality on one hand and computer hardware and software limitations on the other. The higher the scan resolution, the better image quality will be. The higher the scan resolution, the less image detail will be lost in the process and the sharper the final image will be. However, the higher the resolution, the larger the image file, and the more computer resource is going to be used. Perhaps this is a bit simplistic, but I consider image quality to vary linearly with scan resolution. Double the resolution, and you double the size of an image detail that will be captured in the final image, and jaggies will be half as noticeable. However, demand on computer resource varies exponentially with scan resolution….double the resolution and image size (hence demand on computer memory) quadruples. Again a simplistic way of viewing it, as image resolution increases, the disadvantages increase faster than the advantages. So far as bounding the resolution range you work with, the lower boundary (smallest dpi number you want to work with) is totally subjective, depending only on the user’s tastes. Most modelers seem to be satisfied with a final image resolution of about 150 dpi. There are no statistics available (and I’m not suggesting someone rush out and do a research project to establish some), but it is probably a safe assumption that image quality acceptability will conform to a simple bell curve, with at least half of the paper modelers being satisfied with the image quality associated with a 150 dpi resolution and at least 95% of the paper modelers being satisfied by images with a 300 dpi resolution. The upper boundary is non-subjective, being determined by limitations of the computer hardware and/or software. When you push the envelope too far, you will know it immediately because your computer simply locks up on you. In my experience, the single biggest factor establishing the upper boundary for resolution is system RAM. My system has 256 Meg ram, and is severely taxed by not having separate graphics memory, etc. Not being particularly a computer guru, I don’t know exactly what amount of memory is actually working for the graphics package, but would be surprised if the effective memory available to the graphics package is much more than half the installed RAM. In short, my system (including my preferred drawing package, which is notoriously weak with respect to memory management) starts letting me know it is unhappy at an image resolution of 300 dpi, and absolutely pukes when I try pushing the envelope to 600 dpi. Thus, my personal image resolution envelope is from 150 to 300 dpi. If you intend increasing the size of the model, my personal recommendation is to set the scanner at 300 dpi. Generally speaking, scanner resolution is not that big a concern if you intend reducing the model, unless you have an older (and I mean ancient) printer. There are not many printers still in use that cannot handle resolutions of 600 dpi or higher. So, unless I intend really squashing down the model size (for example, reducing a 1:33 scale model down to 1:144 scale), I just leave my scanner at 300 dpi. If I wanted to downscale a 1:24 car model so it will fit on an N-scale train layout, I would drop the scanner resolution to 150 dpi, or maybe even less.

    Now to the resizing part. You can make this step in the process as complicated or as easy as you want. So far as the mathematics is involved, there is nothing more complicated than simple ratios involving two basic numbers….the scale of the kit you are starting from, and the desired final scale you want for your model. For JSC kits, the starting scale is usually 1:400, or 400 scale. I would guess that most enlargements would be to 1:200 or 1:250 scale. Other possibilities are 1:144 (N-scale), 1:100, or 1:86 (HO scale). Regardless of the desired final scale, the principles are the same. Start by determining the ratio between the scales. For simplicity, divide the initial scale by the desired final scale. For 1:200 scale, the ratio is 2.0. For 1:250 scale, the ratio is 1.6. For N-scale, the ratio is 2.78. For 1:400, the ratio is 4.0. For HO scale, the ratio is 4.65. This ratio determines how much bigger or smaller the final image needs to be compared to the initial image size. How here is where things get complicated. Image size is affected by two variables….the number of pixels in the image, and the image resolution. You can change image size by changing the pixel count without a change to the resolution, or by changing the resolution without changing the total number of pixels, or you can change both pixel count and resolution at the same time. Like I said, you can make the process as simple or as complicated as suits you. For myself, I like simplicity, so I only change one of the two variables. From my perspective, changing the total pixel count has some big disadvantages associated with it. First, the computer software has to go through some kind of pixel manipulation scheme. If enlarging, it has to make more pixels, and for each pixel it adds, it needs to figure out both where in the image matrix the new pixel is to be placed and figure out what color value and luminance to assign to it. Sometimes strange things happen to the image in the process. Also, since pixel count is increasing, the electronic size of the image file is increasing (by the square of the scale ratio, remember) and more demand is going to be placed on computer resources. If you change the image resolution, no changes occur to the electronic image the computer is working with. This has a disadvantage that jaggies become more noticeable when model size is increased; however, if the final image resolution is within your satisfaction zone, all will be well with the world when the task is finished. To use a few examples, if my scanner resolution is 300 dpi, a final image resolution of 150 dpi will give me a 1:200 scale model, and a final image resolution of 188 dpi will give me a 1:250 scale model. (For the mathematically challenged, these numbers are derived from the initial image file resolution divided by the scale ratio.) And, if you want things even simpler (and the numbers happen to lie within your image resolution comfort zone), just set the scanner resolution to the original scale, and use the desired scale as the final image resolution. As I mentioned in my Tufur string, there are a couple of ways to change the image resolution. One is to immediately use the resize function of your image processing program to reset the resolution of the scanned image files before going on to doing image cleanup, etc. However, by far the easiest way is not to worry about rescaling until the very last step. When you create the new parts page files, just set their resolution to the desired final value, and cut-and-paste the parts onto the new pages. The scale conversion happens “automatically.”
  2. Darwin

    Darwin Member

    Once I have a set of scanned page files, I clean up the images. Of course, the easiest way to clean up the scans is to scan carefully to begin with. Make sure the scanner plate is clean and free of lint (wiping it with a piece of microfiber cloth works wonders). Make sure the pages being scanned are flat and squared up on the scanner plate. I first remove the staples from the model booklet and scan individual pages, rather than trying to “scan the book.” I make sure the page is snug against the top of the scanner window, and then place a flat, thick backing piece over the page to try to press out any wrinkles (I find my cutting mat ideal for this purpose….it fits just about exactly over the scanner window, and has enough weight to hold the paper reasonably flat). Do your scanning in with minimal background light (after placing the scanner lid over the cutting mat/page sandwich, I then cover the scanner with a towel to further cut down on stray background light). And, if you live on the end of the distribution line in your electrical grid, don’t scan during periods of peak electrical use or when line voltage is fluctuating (like when the big irrigation pumps down the street are cycling on and off). And, if possible, don’t have a large, intermittent load (like a microwave oven) on the same electrical circuit as your scanner…..don’t ask me how I know that one. However, sometimes even with all those precautions, the image still comes out with a blue background, or even worse, the hardcopy original has an off-white background color. First, I color-correct at least the backgrounds of the scanned images in Photoshop. Leif has a great string for doing that….I refer you there. I usually go one step beyond the color correction by using the magic wand tool to select the image background area and then hit the delete button in order to eliminate most of the sparklies in the background. I’ve found a range setting somewhere around 16 to 24 for the magic wand tool will eliminate nearly all background artifacts without noticeably eroding the foreground image. I then correct blemishes in the image areas using Photoshop’s touch-up tool or by copy-and-pasting small “patches” from adjacent, unblemished areas of the image over the imperfections. If there are serious color registration errors on the original page, I also use the copy-and-paste patching technique to fill in color voids in the image. Once I have the original page files in as good a shape as possible, I create a folder and save the images to the hard drive.

    Now we are ready for the serious work to begin. The typical JSC hull construction may work acceptably for a 1:400 scale model, but it just doesn’t work well for the larger scales. If nothing else, I redesign the hull framing by converting it to a more conventional egg-crate style. And, if I happen to have a set of scale hull drawings available, I may even go all out and convert to a full-hull design rather than a waterline. It all depends on how much computer time I wish to put in, rather than building time. Unfortunately, there seems not to be just one “JSC construction method, “ other than they all seem to be along the lines of “first, build a box….” Some of the models start with a base plate, and build as if they are an egg-crate without a keel. Most start from a deck platform, and have no covering at all over the bottom of the hull. Some use a construct of “beams” down the length of the hull rather than a platform. Some have a triangular beam down the length of the hull that skewers all the bulkheads. And not a damned one of them has a decent drawing to show us non-Polish-literates what the hell the designer had in mind when he set pencil to paper. From my experience so far, though, it appears that the most popular style uses a deck platform to which the hull bulkheads and side pieces are attached, with an open bottom. For destroyers and smaller, the hull framework builds as a single piece. Cruisers and larger, the hull frame is usually built in fore and aft sections that are later joined together (see the in-progress build string on the USS Indianapolis for an example of the latter).
  3. Darwin

    Darwin Member

    The example I will use for this discussion is the German Z-10 destroyer from JSC kit # 31. It is an example of the single-piece box construction, with a slight complication in that it has a stepped hull. I started by creating small files of each of the kit’s hull framework pieces. To keep things tidy on the hard drive, I created a folder called “raw” and store the unaltered piece files there. I assign the kit parts number as the name of the individual files. If you are planning to totally redraw the kit, you may as well make files for every one of the parts in the kit. If you are happy with the scanned hull skin and superstructure parts, all you need individual pieces for is the hull deck platform (parts 1 and 2), the main deck (parts 15 and 17), the hull step bulkhead (part 16), and the hull formers. Now create a new window in your graphics program that is wide enough to contain the entire hull deck platform and high enough to allow for placing a hull keel pattern and a hull base platform below the deck platform. For my example, I chose canvas dimensions of 4000 by 3000 pixels. As it turns out, this image size is at the limit of what Image Forge can handle with my hardware, and it let me know it….big time. Frequent lockups, but none serious enough for me to abandon the attempt altogether. Now, use the “paste from,” “standard” selection from the “edit” menu and import the deck platform piece that includes the stern end of the hull. Use the “rotate” tool to position the piece so the hull platform will extend across the canvas, with the stern on the left side of the canvas (this is my personal preference….if you prefer seeing the ship running from right-to-left instead of left-to-right, or top-to-bottom, go for it). Leave about 50 pixels of margin between the edge of the canvas and the end of the hull platform, and far enough down from the top of the canvas to allow about the same amount of top margin at the widest part of the deck platform. Using the “paste from,” “transparent” selection, import the next deck platform piece, rotate as needed, and position it so it joins with the first piece. If the background of the parts file contain so many artifacts (sparklies) that you cannot tell when the part outlines are precisely joined, don’t panic….this is a pain, but there is an easy workaround. Just position the second deck piece elsewhere on the canvas and then use the rectangle-selection tool from the tool bar to surgically select the second deck platform piece (in other words, position the edge of the selection marquee exactly at the end of the part outline you are joining with the first platform section). When you have the selection rectangle positioned just where you want it, nuke the selection (click on the upper toolbar button that looks like a mushroom cloud) and drag it up the canvas to join with the first deck segment. If the deck platform has more than two pieces, repeat this process until you have the entire deck platform as a single image. On the Z-10, the designer joined the hull box side pieces with the deck platform. For this portion of the design, the hull sides just clutter up the image, so my normal practice is to cut them out of the image….however, leave the side segments at the stern and bow ends, because we do need to take a measurement or two from them later in the design.

    In the Z-10, the designer chose to make the hull former at the step as an integral piece of the hull platform. We need to cut that section out in order to be able to use the hull platform to give us the dimension references for drawing the keel and base platform. We also need to have the step former available as a dimension reference for the keel. Use the rectangle selection tool again to select the portion of the hull platform where the step former is located. When the former region is selected, to up to the edit menu and select copy, then select “paste,” “new image.” This creates a new window with the former. Save this in the “raw” folder with an appropriate name….”step former,” or whatever strikes your fancy as a means of locating it later. Close the step-former window, and now use the rectangle selection tool to surgically cut the step former from the deck platform. Include the fold lines where the former bends down from the hull platform with the hull platform pieces (make the cut just inside the former portion of the original hull platform image). Now use the rectangle selection tool to capture one of the hull segments, with the selection marquee snugly against the fold line (which now is effectively the outline of the end of the hull platform segment). Nuke it and join the two hull segments into a single image again. Note that the hull platform on the bow side of the join is a little wider than on the stern side of the join…..this is as it should be, since the hull sides at this point slant inwards at this location, rather than being vertical. When making the join, vertically center the two segments around a common centerline….the amount of extension of the bow segment past the stern segment outline should be about equal at top and bottom. At this point, it would be best to save the image into the “raw” folder. I usually name this file as “hull stackup”…..use whatever name you will later be able to recognize. A comment at this point regarding the JSC models that use the “two boxes joined together at the middle” method of JSC design….the former that makes up the end of the box is integral with the deck platform segments. You can just directly join together the two deck platform segments and then treat the area at the join as just described for a step former, or you can cut the formers off the platform ends before you make the first join….it doesn’t matter when you do it, so long as you remember to make an individual parts file for the formers. If there is no step at the deck where the join occurs (not the case with the Indianapolis model), all you need to save is one of the two formers.

    Starting on the design of the new hull frame……select a foreground color that does not appear (and will not be used) in the image (I usually pick purple). We will use this color for putting in our construction lines. (The reason for the color choice is so at the end of the designing process, we can use the color change tool to set the construction lines to the background color, thus in one brief instant erasing them without disturbing any of the image we wish to keep.) Starting at the very end of the deck platform, we are going to draw in a vertical line at each significant dimensional point on the hull. Draw a vertical line (all the way to the bottom of the canvas) at the extreme end of the stern end of the hull platform, at the extreme end of the hull side (this is the point where the hull side joins with the end of the hull….what I would call the transom if it were a row boat), at each hull former location, and at the extreme bow end of the deck platform. VERY IMPORTANT…..because the bow, and usually the stern, of the hull are usually raked (slanted), the end points of the base plate are not going to be the same as the end points of the deck platform that we are using to establish the construction lines. Now go the little piece of the stern end of the hull side piece I told you to save. If the kit you are working with doesn’t have the side pieces integral with the deck platform, they you will have to open the file for the starboard (right) hull side piece and use it for the next step. Using the rectangle selection tool, place a selection marquee around the vertical line that is the end boundary of the hull side. It is best if the vertical sides of the marquee are snug against the line segment we are capturing. When you have the marquee positioned exactly around the line segment, nuke it (hit the mushroom cloud button in the upper tool bar), and then drag the line so that the extreme upper end pixel of the line segment lies on the construction line coming down from the stern end of the hull platform. (It would be best to position this segment just far enough below the deck platform that it is in an empty portion of the canvas.) Now select the line tool, place the cursor over the extreme bottom end pixel of the line segment, and draw another vertical construction line down to the bottom of the canvas. This line tells us (approximately) where the base plate (and the bottom end of the keel) ends; the construction line coming down from the very end of the hull base plate locates the top end of the keel. Repeat this process for the bow end of the side piece vertical outline. This has all the vertical construction lines in place. Go about halfway down the canvas and draw a horizontal construction line completely across the canvas. (This line will be the bottom of the keel.) Now drop down about half the distance from the keel construction line and the bottom of the canvas….about 1/4 the height of the canvas….and draw another horizontal construction line completely across the canvas. This will be the centerline of the hull base plate, and is the last of the construction lines we need (at least for the moment).
  4. Darwin

    Darwin Member

    Now, a little digression….why do I say we know the approximate location of the end of the bottom plate? Because we used the amount of rake at the hull side to determine the amount of rake at the hull centerline. If the amount of rake (slant) at the two locations are the same, the technique I described above will give you (within a few pixels) the true location of the end of the base plate. If the amount of rake at the hull side is not the same as the amount of rake at the centerline, the graphical approach gives you only an approximate location; the greater the difference in rake at the two locations, the greater the error of the base plate end point. You can get a feeling for whether this is going to be a problem by looking at the hull skin piece that covers the stern end of the hull. If the height of that piece looks the same at the center as at the end, the amount of rake at the hull centerline should be very close to the rake at the hull side, and the graphical technique I described is fairly accurate. If the height at the center is visibly different from that at the ends, the rakes at the two locations are different and the graphical approach is only approximate; you will encounter a fit problem during construction. I encountered this difficulty with the second model in the JSC #31 booklet (the T-24 torpedo boat). For the T-24, I used a mathematical approach rather than graphical to determine the base plate end location. Assuming I don’t loose all readers from terminal boredom in the interim, I will come back to this point at the end of this string.

    If you haven’t already done so, save the parts stackup image. Now use the “paste from,” “transparent” menu item to import the main deck pieces into the stackup image. Rotate them as needed to line up properly with the deck platform portion of the image, drag them to the proper location on the deck platform, and make a quick check for length dimension problems. For proper fit after assembly, the deck should completely cover the portion of the deck platform it fits on. In the case of stepped hulls, like the Z-10, the deck piece should extend past the aft edge of the step hull former by the thickness of the cardstock you are printing the model on. In most cases, that will correspond to about 3 or 4 pixels. Also check the tolerances at the bow and stern….the deck should not extend more than 3 or four pixels beyond the end of the deck platform. If it extends more than that, adjust the location of the vertical construction lines so that the bottom plate and keel will be the correct length to accurately fit the deck. In the case of the Z-10, I had to move the construction lines at the bow and stern by about 10 pixels each. If I didn’t make that correction, the finished hull would have been nearly an eighth of an inch shorter than the deck…a dimensional error large enough to be more than just a nuisance. To be exacting about it, the construction line marking the keel top and bottom end points need to be moved by the same amount…otherwise, you will change the amount of rake of the bow and/or stern. Although it may be overkill, I spread the dimension change over the entire hull, not just in the end bays (a “bay” being the portion of the hull between two adjacent hull formers). For example, if the bow has to be moved out 10 pixels, and there are 5 bays between the hull step and the bow, I lengthen each bay by two pixels. (I consider the location of the hull step to be the base reference point ….dimension zero, so to speak.)

    Getting back to the promised discussion of how I handled the case when the rake of the stern is different at the centerline from that at the hull side. Our old friend Pythagorus comes to save the day....for a right triangle, the square of the hypotenuse equals the sum of the squares of the sides. The keel piece near the stern of the ship can be broken down into a rectangle butted up against a right triangle that forms the end of the keel. We know the height of that triangle is the distance between the hull bottom plate and the deck (the height of the keel piece). The hypotenuse is the vertical dimension of the hull end plate (transom). Plug those numbers into the Pythagorean equation, turn the crank, and out comes the dimension of the base of the triangle, which is the dimensional difference between the top and bottom ends of the keel piece.
  5. Darwin

    Darwin Member

    Well, I hope this helps more than confuses. If you are totally confused on any of the particulars, get interactive and maybe I can figure a better way of passing on the information.

    Now for a rant.....this new forum format sucks for attempting anything other than the shortest and simplest of tutorials. Not only are the limitations of including graphics more severe, but there are now limitations to the amount of text that can be placed into a single posting. Burma Shave to the nth power!!!!!!! There has got to be a better way, but what is it? Maybe doing the tutorials as pdf files that can be downloaded/uploaded from the site? I really don't like that approach though, since it eliminates interactivity the forum format provides. Guys, after you get the site to the point you can actually navagate through it, please give some thought to how we can actually present tutorials effectively.

Share This Page