USS Iowa

[Darwin]

With my recent experience with the Blue/Helena builds, it seems to work better for me when I have two builds going rather than just one, so I dusted off the GPM USS Iowa kit I managed to score from Ted during his recent Ebay insanity. For some reason, I just can't bring myself to build a kit "straight from the box," so am doing some heavy-duty computer work on the scan in preparation for the build. The original GPM kit is 1:300 scale, full hull. I'm coverting to 1:250 so it will fit in with the Helena and Blue, which are now proudly sitting on the top of my credenza in my cube at the office. In light that there are some 18 pages of scans, with lots of little bitty parts, I decided not to do a redraw. However, the print quality of the kit leaves a lot to be desired (it is an older one, GPM #61) with lots of blemishes and color registration problems, so still have a couple of weeks of computer work ahead of me. The kit is on A4 paper. I scanned with a 300 dpi resolution. The rescaling part is dead easy....I made myself a bunch of legal-size pages at 1:250 resolution, and am just cutting and pasting from the scan to the new pages. The parts fit on the legal-size paper is just perfect....even the largest of the kit parts fit on the page after rescaling without having to be chopped up into subsections. There is some juggling around needed, since the page width of the upscaled A4 is considerably wider than legal paper. As there is no actual drawing going on, I am doing the entire task in Photoshop 7. I am trying to pretty the parts up by healing the boo-boos (color voids and blemishes) using the retouching tool (the icon that looks like a little band-aid) in the centers of the colored regions, and copy-and-pasting near the edges of the color regions. For some reason, the retouch tool goes a bit berzerk when used too close to an edge, and smudges rather than heals. I'm also using the 'copy-and-paste" method to fill in the white spaces left by the color registration errors. It's easy enough, but time consuming as all hell. The final product makes it worth it to me, though. Gives me time to decide on whether to make it full-hull, or waterline so it fits in with the other two models. Thanks to Lief's threads, I was able to correct the color shifts introduced by the scanner.

No pics at this point....this post is just marking my territory (and, with a thread started, may help keep my feet to the fire to actually finish another project, though not as quickly as the others).

 

[Jim Krauzlis]

Way to go, Darwin!

Just a question on how you upscaled this...it's probably a very simple matter, but you mentioned cutting and pasting the scan to legal sized pages...did you scan the original at 1/300 and then enlarged the scan files in Photoshop to 1/250, and cut and paste from that to the legal pages, or did you do something different that I didn't catch? Does the 300 dpi provide good resolution for these files? Some of the scans I make of the JSC kits (because I make a lot of mistakes, okay? ) it seems the printed copies are never as sharp as the original pages...am I doing something wrong?

Sorry to get off track on your thread, but your comments about how you scanned IOWA made me think about my experience and I wanted to know what you thought...

Hopefully some pictures next time around?
Really looking forward to this one too!

Cheers!

Jim

 

[Leif Oh]

Well done, Darwin. Nice trick with the 300 - 250 dpi resolution! And I am so glad some of all that on scanning and recolouring could be used for good purposes by others too. (By the way - and I hope this is alright with you, Darwin - the stuff on scanning is here.)

Jim, the way it works - I'm thinking - is that an inch of pixels in the scanned original (at 100 percent, 300 dpi) no longer "fit" into an inch in the new 250 dpi empty sheets Darwin made up for the purpose. In fact, they take up just the right amount of more space to equal enlarging from 1/300 to 1/250. Very neat, isn't it!? Just copy and paste, as Darwin says.

In Darwin's case it was so pretty and easy (once you have somebody like Darwin suggest the idea!), since the scales and the dpis are so numerically similar. But the same principle will hold, of course for other scales as well - but then you would have to do a teeny-weeny little bit of calculating, to ensure that the proportions between the dpis stay right (desired scale divided by original scale, times the resolution of the scanned original, equals the resolution you should set in your empy sheets for pasting parts into; remember what Darwin did - 250 scale divided by 300 scale, times 300 resolution in scanned original, equals 250 dpi set in new empty legal size document to paste parts into).

More on clarity of print, Jim: I found exactly the same thing when test printing my recoloured Storch. Never really thought much about that before, since I always enlarged and then printed, on fairly coarse paper at that (which I use on account of size and thickness, and availability). I never printed an original at the same scale, and therefore never had reason to think about the loss of quality ink jet printing means. Any loss of quality I attributed to unavoidable losses in scanning, or bad paper quality, or something not within my control. NOT SO!

I have tried to summarize my findings in the evaluation of recolouring the Storch. I really had hoped that would be useful reading for a lot of people, even if they never recolour. The gist of it is that the weak link is not scanning, not resolution, and not even the amount of mistreatment you manage to inflict on the original in your computer - but the final printing process.

You lose A LOT of quality in inkjet printing. There are a few things you can do, however, and they are about finding the adjustment in your print dialogue (often hidden under scary options like "expert" or "advaced") which specifies how to print at 360, 720, 1440 dpi (or similar options). CHOOSE 1440 (or similar) WHATEVER RESOLUTION YOUR ORIGINAL IS IN. (Any higher than that is excruciatingly slow; and 1440 is already slow.)

What this means (I think) is that the printer ink heads go over the same spot of paper much tighter, irrespective of the resolution of the original. I did some fairly rigorous testing (since I was so shocked at the poor result using my usual setting) and found beyond doubt that this is where you can actually make a difference.

The resolution of the original (i.e. dpi in scanning or whatever you do in your computer) does not matter very much compared to that, if you make sure to stay higher than, let's say, 100-150 dpi. I tried resolutions down to 90 dpi and they turned out better, printed at 1440 printer setting (not document resolution), than high-resolution images printed with standard settings.

I would dearly like to see somebody else replicating some of these tests to validate the results, particularly if you use high-quality paper. I imagine the main conclusion will be the same, but that you will start to see a difference sooner at low resolutions.

Best, Leif

 

[bholderman]

Greetings,

I thought I would toss in my 2 cents as a large format blueprinter.

Leif is correct on his assumptions on printing. Each individual printer has its own default settings as well as its capabilities.

I've done the same process of enlarging a sheet (or its selected image) and sending it to 3 different printers. The result was 3 differents prints of both quailty and scale (?).

A lot of the digital paper models are in Adobe .pdf format. This normally has a default setting of trying to fit an image to the paper size, whic can commonly destroy any percentage settings. Also, one should thouroughly examine all the available settings in the printer dialogue, especially under the "advanced" button. From there, its a bit of trial and error for scaling.

Then there is paper. I would'nt recommend plain paper for anything but scale testing discussed above. It simply absorbs and blurs too much of a fairly detailed file. At a basic level, I would recommend a color copier paper (we use Hammermill "Color Copy Paper" (go figure), at 28 lbs. with a Brilliance of 96 as our generic laser jet paper). Is dense and avoids bleeding unless a color section is especially dense (i.e. a red hull). When printing this much solid color even this paper gets "wet" and wrinkles. Thats when you finally open your printer manual to find out how high a poundage (is that a word?) a paper the printer can manage. I personally managed to get an HP2500c for just a couple of hundred dollars via ebay, and its worked wonderfully. I'm currently experimenting with heavy canvas papers on Currell's R101.

Briefly, most tend to overestimate or ignore a printer's ability, or even worse underestmate it. The trick is to experiment without running out of ink, that's the expensive part.

In regards to the Iowa, my Dad was born on the day of its commissioning. I can't wait to see the build.

Cheers,
Brad

 

[Darwin]

Jim, unfortunately, the second law of thermodynamics seems to apply to cardmodelling as well. No scan is ever going to be as sharp as the original, especially with equipment affordable to the average home user. The higher you set the resolution of the scanner, the closer it will be to the original. I am limited by computer RAM....much higher than 300 dpi, and it starts groaning. However, considering the way ink bleeds on cardstock, I really think we are just kidding ourselves to think we are getting super quality of printouts by scanning and printing at a kajillion dpi and working with jigabyte-sized image files. I really don't see any appreciable increase in printed image quality above about 300 dpi, and in some cases the increased resolution actually degraded the final output, in that the details became so fine that the printer couldn't handle it. (I'm waiting for the blazes from the technofiles in the group, but then again, they may have access to a lot better hardware than most of us can hope for.)

As far as the scaling....and again, I upset the purists....there are lots of ways of skinning that particular feline. You can hold the resolution constant, and manipulate the pixel count of the image, which is what I suspect most of the purists will advocate. Or, you can hold the total pixel count constant, and change the image resolution. :twisted: Or, if you really like hurting yourself, you can do both at once. If you keep the number of pixels constant and change the image resolution, the computer does not have to do any magic with the image, thus you don't have to worry about aliasing, noise introduction, color shifts, etc., etc. (On top of that, after working months in my drawing program to get every pixel in the image exactly where I want it, I damned well don't want the computer kafluxing around with it outside of my control.) The downside is that, particularly when enlarging, the image tends to "pixelate" (loose its sharpness) and the dreaded jaggies become intrusive. When going in the opposite direction (decreasing size), the printed image will actually improve in quality (appear to become sharper and crisper, lines look smoother, usw {threw that one in just in case some of our German friends may be reading this}). My own preference is to make small scale changes (read that as less than a 50% enlargement) by simply changing the resolution.

Again, there are a couple of different ways of doing that. One can go through the process of changing the original page file. Select the image size function, enter in the desired resolution (250 dpi in this case), which, in Photoshop, changes the pixel count, because the program tries to hold the canvas size a constant), do some creative swearing because you didn't happen to write down the initial pixel count, start over again, and then after changing the resolution, reset the pixel count (image width and height measured in pixels instead of good, old-fashioned inches) to the orignial values. Because the new paper size of the image is now bigger than anything you can force into the maw of your printer, you now make a new image file that is sized to fit your printer feed slot (letter, legal, A4, whatever), with the same resolution as your new image file. I did this, for well over a couple of years, until it finally sunk in that, when you cut and paste part of an image to move it between files, the image size of the piece you are moving stays a constant. If it is 100 pixels in the original, it goes to the clipboard at 100 pixels, and finally to the final location at 100 pixels. DOH!!!! I could accomplish the resize simply by making the resolution of the destination parts page different (by the proper ratio, of course) from that of the original image file, and avoid all the aggravation of working with the image size function. (Lief, I know this is going to offend your sense of technical purity, but I'm gonna say it anyhow) To make it real no brain, and as long as we are working with typical ship model scales), here is the Grigg "rescaling for complete imbiciles" method. Set the resolution of the scanner equal to the scale of the original kit (in this instance, 300). When the scanned file is in your image processor of choice, make the resolution of your "new" parts pages equal to the desired final scale (again in this instance, 250). Then cut and paste the parts from the original file to the new file, hoping that each of the resized parts will somehow fit on the new pages (if one doesn't, just transfer it over in pieces and glue it back together during the build....what's one more seam among thousands?). Hope this helps.

 

[Darwin]

Leif, some day I'm going to remember how you spell your name. Hope you don't mind my taking your name in vain (particularly since I was blind to your reply whilst typing up mine own), but if one can't occasionally do a little teasing, what is the value in keeping up the hassle of drawing another breath? I greatly admire your technical ability....please don't think I am in any way being disparaging. Bye the bye....could you please let us know the phonetic spelling of your name? Took me nearly a lifetime to realize that Sean was pronounced "Shawn." so am curious.

 

[wunwinglow]

Just to throw another spanner in the works, and slap me hard if you already mentioned this before, but what about interference between the printed (offset litho) spot of colour on the original printed paper, the scanning resolution, any scaling by changing the number of pixels involved, and finally the resolution of the printer/paper combo? Doesn't this mush things up anyway?

I love all this techno stuff!

Tim P

 

[Darwin]

Tim, you really know how to hurt a guy.....going back to the second law of thermodynamics (everyone should know these....zeroth law, stuff always flows downhill. First law....TANSTAAFL. You aint ever gonna get something from nothing....the best you can do is break even. Second law...you aint even gonna do that, so why try?), anything done to an image short of redrawing it is going to degrade the quality. Color values are going to change through every step of the process. The only question is "HOW MUCH?" (And secondary question is always "how much change can I tolerate?") Third question is how far off track can a string drift before it assumes an entirely new identity. Hey, this is a discussion forum. As far as I'm concerned, as long as we're having fun and no one has died from boredom, all thoughts are fair game....anyone doesn't like it, they don't have to read. {klik...klik....klik....the sound of offended members signing off. Did I ever say that PC I am not?} Seriously, you are now delving past my limited technical expertise. My world is a simple one....what I understand is AM (amplitude modulation). Everything else operates on FM (freakin' magic).

 

[Leif Oh]

Tim, the interference between the print pattern in the original and the mesh imposed by the scanner and printer, I became wise to in the post about moirée patterns. I knew about this from an earlier life as a journalist but never thought about it in paper modeling, until I detected the unmistakeable "moirée" pattern in one of my own scans. Since then I've seen it in many a scan posted on different sites.

The solution, I found in that post, was to avoid any setting in the scanning and printing that has anything to do with "photo". Stick to "line art" or "document" or anything that is NOT "photo". That way you do not impose a new mesh upon the existing one.

Darwin, to me most everything encountered in this sphere is still "FM" (Freaking Magic - love that one!). I just like to write about it when some aspects of it - often with the help of others - become less so. The trick about changing resolutions you described here certainly was one such instance!

Leif

PS. You would probably pronounce my name "leaf" which is very OK. In Swedish it's more like Star Wars princess Leia, with the "a" at the end substituted by an "f".

And now - on to the Iowa! Sorry for hijacking your thread...

 

[wunwinglow]

Leif, thanks for the slap! I feel much better now. I will go read the Moire stuff right now.

Tim P

 

[Leif Oh]

No slapping intended, Tim, sorry!!! Just wanted to keep that thread alive, since there were so many valuable contributions. Altogether it made for a lot of information. - Leif

 

[wunwinglow]

No, Leif, I told you to slap me! I should have been paying attention earlier. I agree, the amount of info that comes out of these discussions is fantastic. I might even have a go at bitmap editing myself sometime!

Ah well, back to my Arizona...

Tim P

 

[Jim Krauzlis]

Thanks, guys!
I now have a bit of reading and experimenting to do.

Very nice article, Leif, and thanks for all the advice...now I just have to figure out all this computer and printing lingo...computer challenged, am I. :lol:

How's the Iowa coming along, Darwin? I am looking forward to seeing how she is coming along, any pictures yet?

Cheers!

Jim

 

[Gil]

Darwin,

The guns for the Iowa were all tested out in the desert not far from where you live during WWII. Think it was run by Springfield Armory.

On the subject of moire patterns and sampling theory the following web page has a fairly decent problem description and better yet describes fixes for the most popular desktop paint packages. The fixes all result in filtering "down" the patterns. Interesting point is that you should sample at 2-2.3 times the original dpi content. Filetering high resolutions with a pixel width of just over 1 will give good results as related in the article.

http://graphicssoft.about.com/od/glossary/l/blmoire.htm

Gil

 

[Darwin]

From what I understand about the history of the site, the armory was over in Pocatello, and the guns were set up about where the Central Facilities Area is located at the Idaho National Laboratory. The South Butte was the most frequent target for the testing. We are still uncovering various pieces of ordinance out on the desert, and part of the environmental monitoring scene is checking the growies, creepies and crawlys for explosive residues. All kinds of fun making the arrangements to send a couple hundred haunch of mouse to the laboratory to the analytical lab at San Antonio for testing. Oh, well, it's a paycheck. Could be worse....could be the one fending off the rattlesnakes while checking the traps for the samples.

 

[Darwin]

Jim, I still have a couple of weeks of computer time ahead of me to get to the point of cutting paper, so don't expect to have much in the way of pics until near the end of the month (of the Iowa, anyway). I am far enough along with the design work of the Mogami to start butchering paper on it, though....I will start a string on it as well, once the paper starts flying. I picked up a building board for it my last trip over to Home Depot, but still need to get a much longer one for the Iowa.

Hmmmm.....no wonder the trackball was getting a bit flakey....does anyone know if there is a name for the little pads of gunk that build up on the bearing points of trackballs?

 

[Gefahren]

I don't know the technical name for them, but gunk works well as a discription. I know I clean mine at least once a month to make sure that those things don't build up since they effect my game playing. If there is a technical name for them, I'd be intrested in why someone spent the time thinking of one.

 

[OldSalt]

I believe the correct technical term is "mouse ball toe jam."

 

[Darwin]

Works for me...Let's hope it never gets mixed with bellybutton lint.....

 

[Darwin]

I thought I'd give an update. Still slaving away on the computer. My decision at the beginning was not to redraw this model, but try to go as quickly as possible to the buiding part after resizing the parts. Well, to make a long story brief, the color offsets of the original kit were just too much for me to ignore, knowing I could fairly easily correct them and the various scanning flaws, etc. It may have been quicker just redrawing the damn thing. I now have about 16.5 of 18 parts pages touched up to the point I am satisfied with them. I have been using photoshop, first correcting flaws in the interior of large areas of color using the retouch tool (looks like a bandaid), then using the old cut-and-paste method to correct flaws too near the outlines for the retouch tool. As a tip if you try this, minimize the pain by, when a part is repeated, just correct one and then make a copy to paste over the other similar parts. The pic below shows a before-and-after comparison of the cleanup (and this is one of the better areas of the original with respect to flaws).

I don't know how fast I'm gong to get to the cutting part....I'm now about two weeks into trying to heal from a bit of a fall I took when carrying a heavy sack out to the garbage can. Somehow during the fall I managed to get one of my hands jammed up underneath my ribcage and separated a few ribs. Man, it takes a while for things to quit hurting.