I can't seem to get BMP Tracer to work

After seeing this thread:

I tried installing and using BMP Tracer and I’ve had no luck with it. I’ve tried a dark version of what I’m trying to trace, where the details are white:

And a light version, which is all white with just black outlines of all the details:

with the same results, a pause and then nothing, even if I zoom extents.

What am I doing wrong?

If you can post the picture, maybe I can help you figure out why it’s not working for you. Also, if you open the Ruby console before running it, diagnostic and/or error messages should help to determine the problem.

Also, if you haven’t seen this video, it might help: https://www.youtube.com/watch?v=uQXPL89zhH4

A picture of what? I mean I posted gifs of trying it with 2 different BMPs

If you’re looking for the BMP, then the first one is the dark one and the second one is the one where all the details are converted to black outlines on a white background.

Check that, BMP is not a valid file type for upload here, here are screenshots of both bmps:


I’ll look at the video later when I’m sitting still…

I took your image and chopped some of the detail away. Then I used Gimp to convert it to black and white with a low threshold and added a white border. Then saved it as a color BMP file:

This is the SketchUp file after running the importer: test.skp (292.3 KB)

If you can eliminate the fine print and add a border, it seems to work just fine.

What is causing the problem? Can you send me the bmp that you used so I can try it? The fine print is not that fine in the real file, I believe I resized it to 4x it’s original size something over 8000x7000

I’ve probably tried it with half a dozen BMPs (things like just searching my hard drive and finding the verisign logo bmp, mostly small but a few larger) and I’ve had exactly the same results in as in the gifs that I posted. The only time I had any output is when I took a screenshot of the apple logo in the old thread (linked at the top) and saved it as a BMP and tried it and I got the body of the apple and no leaf at the top.

I guess I need to figure out how to open the ruby console… will look later.

The console can be found on the “Window | Ruby Console” menu.

There needs to be a border around the image of at least two pixels or it won’t work. It’s also important that the image be essentially monochrome even though it is a color BMP file. The RGB values you specify when running the program dictate what colors are sampled … many images may look black, but they have slight variations in blackness. I usually create my own artwork in black and white and then save it as a color image. Or filter it through Gimp to convert it to black and white before saving it as a color BMP file.

Here’s the image I used (I deleted some more of the text labels):

This is a PNG file, but you should be able to save it as a BMP and import it to get this (I used exhaustive de-pixelation:

image

test2.skp (127.1 KB)

OK, so my outside lines were not >2 pixels wide so I made them all wider (smallest that I could find was 3 wide), and I get this when I try (it’s been sitting like this for 30min, I’ve tried it on 2 separate machines and it does the same thing):

I tried scaling it half the size (I checked and all my lines were at least 2 wide) and I just got the outside outline and nothing inside it and then it crashed right away before I can get a screenshot or anything.

I don’t know how much memory you have, but the dimensions you’re using is about 48 megapixels which works out to more than 140MB of RGB data. I suspect that the import might finish someday, but I’m not sure how well Ruby handles arrays of this size. I’ve imported images up to around 2000 x 1600 before, but when I get above 3 megapixels or so, it starts to take forever. Since the algorithm works its way around the image in loops, data that is being paged from disk can potentially require almost continuous disk fetches just to grab the next pixel location.

Try reducing the dimensions to 1975 x 1576 and see if this helps.

Since that is such a simple layout, why not just draw it from scratch and get on with your project? You could import the image and use it for a guide while you draw with the drawing tools.

2 Likes

I do most of my work on my laptop (not bad, but an ultrabook, Yoga 2 Pro, i7, 8GB…), but the second machine I mentioned I tried it on is a quad processor machine with 32GB and about 16TB worth of drive (1TB SSD for OS and swap…). If it doesn’t finish it quickly nothing will.

I tried shrinking it down to half size and like I said, I get the outside boarder, none of the inside detail and then SU crashes. I can try shrinking it down to >2000 later, but then nothing will be over 2 pixels wide without re-editing everything.

That’s why I disappeared for a few days after asking the question.

I probably tried to get an extension to do it for me for a few hours, then I said screw it and just did it manually (imported it as a jpg, drew a rectangle the correct dimensions, scaled the JPG to match and drew from scratch on top of it), it took like 5-10 min and it came out quite good:
image

Honestly it took me about 5x as long to take the .jpg scan, scale it, clean it up, turn it into B/W and then an outline and save it as a BMP then it did to just draw it manually in SU.

I came back here to see if there was anything new about getting this thing working when I had some time a few days later. I figured that at some point it might be something useful to have as a tool, I loved your Superman logo in the other thread, and as a father of 5y/o twins something like that may make me more of a hero than actually building something useful :wink:

I had thought about suggesting doing this, but your question was how to import the image. The importer was mostly intended for importing logos and artwork. In trying to answer your question, I apologize for dragging a red herring across the path :frowning:

Your final result looks good, BTW :slight_smile:

Honestly, most of the questions I ask here are “How do I get this working, I want to learn how.” not “I just need this stuff done help me.” So there’s no red herring here, I can see this as a useful thing for logos and patterns and stuff, stuff that I’ve used in 3d printing and am trying to figure out how to use.

THAT SAID…

um, I just RDP’ed to the beefy machine that has been chugging at this since last night and guess what I found:
image

:slight_smile:
(After scaling it to the correct proportions, exploding it and giving it dimension)

Now I wish I knew why the smaller copies are not working :-/

Why are the top faces reversed?

What do you mean by: “Now I wish I knew why the smaller copies are not working”?

When I scale the BMP to have smaller dimensions (1/2 or 1/3 of the one that I posted a screenshot of it with the ruby console open and hung) it imports only the outside outline and then crashes.

[quote=“DaveR, post:13, topic:52024, full:true”]
Why are the top faces reversed?[/quote]

How do you always notice that? Somehow I miss it entirely. Does it cause a problem with something?

To be honest looking at it better it imported the image as geometry but it diid’t work well enough to use. EVERY line is doubled and with the fact that some of them are squiggly and other issues it makes it just about impossible to fix so I have closed faces, at least now without tons more work than just drawing the whole thing manually. :frowning:

It occurs to me that maybe I’m asking the wrong question, and maybe this is a bad example but something I’ve wondered before. Given a scan of an object (raster, in this case a jpg or png, and since it was a 2d scan of a 3d object you’re going to have some shadows and stuff), is there an easy way to import it as geometry in SU (lets say that it was more complicated than this or you wanted to preserve font styles…)?

Honeyview_21931130_10213966989349441_248893461_o

You could edit the style so the back face color is more noticeable.I use a green I’d never use in my models as a material. It’s easy to see, even for just a few pixels on screen so holes are easily spotted. Some people set the color to magenta or some other annoying color so they fix the face orientation when the faces are reversed.

It can. Many rendering programs don’t render back faces even if they have a material applied to them. They are just rendered as black like there’s nothing there. And for 3D printing, correct face orientation is critical.

There are programs that convert raster to vector. These programs create vector lines between contrasting pixels much like BMP Tracer. Often you’ll see sort of stair step lines around curves. They can be cleaned up but you need to weigh the cost. Do you spend more time cleaning it up than you save by importing the converted image? Like any computer related stuff, GIGO applies.

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.