How can I machine that logo?
This page consists of an article written by Bill Elliott,
President, Northwood Designs, Inc., Antwerp, NY, in the early 90's.
For some reason, it's still applicable today.
CNC user's who have tried to machine logo's and
lettering have found the process to be time consuming and nearly
unworkable, yet there is a solution to this problem.
A number of years ago our company became aware of the growing number of
artistically oriented drawing packages available. This software was able
to help the user design logo's, create lettering, and generally produce
artistic images with an ease that was unmatched in the CAD world. Smooth
flowing curves and text manipulation was refined to a high degree in
many of the available software packages.
About the same time, the idea of scanning original artwork and using
these scanned images for other purposes was coming into its own. Excited
by the possibilities of using this software and/or a scanner to create
files which we could then machine, we quickly purchased a scanner and
some illustration software and set to work.
Many frustrating days later we came to the realization that this wasn't
going to be as easy as it seemed. There were a multitude of problems we
encountered when attempting to machine these images. These problems
included: open contours with endpoints that didn't meet, extremely large
files composed of thousands of short line segments, and inappropriate
geometric entities such as bezier curves which both our CAM software and
our control were unable to manipulate.
These were only the most glaring difficulties, there were many other
aspects of the process which kept this from ever becoming something we
could look forward to doing on a regular basis. The underlying
complication with all of the problems was that the software that was
good at producing these artistic files, assumed you were going to print
them. As you can guess, the accuracy necessary to produce acceptable
printed output falls far short of what is required to machine an image.
A closer look at the above problems showed us that the open contours
made it difficult to "chain" the geometry we wished to cut with our CAM
software. When we attempted to define the contours we wished to cut, the
software was stopping at small gaps or endpoint mismatches and we would
spend hours zooming in on these areas and inserting lines. Even after
all of these gaps were closed the sheer number of small lines required
to approximate a curve, often produced block transfer problems with the
machine's control. This would cause the machine to cut slowly and often
stutter as it was cutting. If we told the software to produce less short
lines on each curve, the finish of the part suffered from flat spots on
the curves. The third major problem was files with inappropriate
entities, primarily Bezier curves. These files were simply impossible to
import directly into our CAM software, and any software we used to
convert these bezier curve files lead back to the first two problems. In
spite of all of these problems -and others not mentioned above- we
weren't ready to give up just yet.
We Need Help!
The next stage in our saga involved researching the software market for
other products that might alleviate the obstacles we were encountering.
It would be inaccurate to say that we made no progress, in fact we were
able to machine most of the images we wished to machine, but this was in
spite of any additional software we purchased not because of it. In
fact, we were only able to machine this artwork after spending
inordinate amounts of time and effort manually fixing the problems. One
and one half years and many software packages later, we were essentially
no closer to a quick and easy solution. There was a good side to this
however, we were becoming real experts on the problems!
If You Want Something Done Right, Do It Yourself.
Eventually we came to the realization that there were no real solutions
available to the problems we were encountering. Even software that
claimed to help, never really met with our expectations or appreciably
shortened the time we spent to machine these images. The only possible
choice was to develop the solution ourselves.
In November of 1990 we took our first small step. The original idea was
to use the HPGL (plotter) files we could export from some of the
software and to translate them directly to our CAM systems toolpath
format. After running the post processor on these files we were able to
force the machine to jump over any gaps in the geometry. This worked
reasonably well for engraving type operations but we still had problems
with slow cutting and finish quality, and this method still produced
files of an enormous size. It was not unusual to start with six lines of
lettering and wind up with 10,000 lines of machine code.
What we really wanted was a geometry file which could be imported into
our CAM software that consisted of lines and arcs. We reasoned that if
the file had lines and arcs the total entity count would be much lower
and the block transfer problems would go away. In addition, if we could
import this file into our CAM software, without the problems, we would
be able to use this geometry to produce contours with a tool offset and
even pocket areas of the file!
We soon found out why we were not able to buy the software we wanted
from someone else. The software we wanted had to be able to analyze a
series of points which might make up a letter "A", and then
intelligently fit lines and arcs while determining which point should be
a corner, what should be an arc, where one arc should stop and the next
should begin, whether the meeting point of two entities was smooth or
pointed, ad infinitum. This turned out to be more difficult than we had
anticipated, but we gradually started to obtain the results we were
hoping to see.
As the entity count was reduced and endpoints matched, we started to
address some of the other problems we had encountered over the previous
years. We added a facility to check for gaps in contours and close them.
Soon we started to import the Bezier curve (EPS and AI) files which were
typically produced by illustration software. Once imported, we were now
able to fit the lines and arcs to these curves. At this point we finally
had a software solution which addressed the major problems we had
In The End...
Over the next several months we continued to add features to the
software which helped us to make the entire process quicker and easier.
Three years after our first attempts to machine artwork we were able to
machine virtually anything we received in a small fraction of the time
it had previously taken us. This was not to say that 10 minutes after
receiving a logo we were starting the spindle on our CNC. The time from
starting on a new piece of artwork to actual cutting still varied
greatly from as little as 10-15 minutes to as much as five or six hours
of preparation. It all depended on where we started from. If we received
a perfect piece of camera ready artwork at a relatively large size, we
often were able to scan the artwork and machine it in very short order.
If on the other hand we received an enlarged, faxed logo with complex
script lettering on an arc, the processes took many times longer.
In the real world, the artwork you are likely to receive from a customer
is usually somewhat less than perfect. In fact it's been our experience
that poor originals are the rule rather than the exception, and,
unfortunately, even at this point in time there's no software able to
make a silk purse out of a sow's ear.
The solution was to assemble the best software tools to handle the
artwork we were typically receiving. During the years we were working on
this problem, we tried many different software packages, some helped us
in one way or another, but none solved the problems in their entirety.
Our current recommendations for a complete hardware/software solution to
machine artwork and lettering includes: 1- A full page flatbed scanner,
2- Raster to vector conversion software, 3- Illustration software, 4-
Our own software "InterFlux", and 5- Your current CAM software. A
discussion of each of the above follows.
Item 1- A Flatbed scanner.
A flatbed full page scanner will always produce the best scanned images.
A scanner produces what is know as a "Bitmapped" or "Raster" image. In
its simplest form this is nothing more than black dots on a white
background. Bitmaps cannot be machined, but a bitmap is the beginning of
the process when you receive hardcopy (i.e.- an image on paper) from a
customer. In the best case scenario this hardcopy would be large, crisp
edged, camera ready artwork. In reality, it's often of very poor
The quality of the original artwork determines the direction to proceed
after the initial scan. If the original is one of the rare "perfect"
originals you may be able to perform what is known as a "raster to
vector conversion" on the bitmap, run the resulting file through
InterFlux and wind up with an easily machineable file in a matter of
minutes. More commonly, all, or part of the image will be less than
adequate for machining purposes and you will need to further manipulate
the file. It's important to realize that there is no magic involved in
machining artwork, it still requires the operator to analyze each job
and determine the most appropriate method to reach the desired result.
While on the subject of scanners, it should be pointed out that most
scanners have a scanning resolution of 300 dpi (dots per inch). We use a
300 dpi scanner ourselves. Currently it is possible to purchase scanners
which have resolutions up to 1200 dpi or more. The cost does increase
accordingly. The resolution of the scanner is mentioned at this point
because a higher resolution scanner will produce better results. The
penalty of a high resolution scanner is twofold, cost and file size. An
image scanned at 600 dpi is four times as large as an image scanned at
300 dpi. In practical terms, this means that you will probably want a
more powerful computer with more RAM to manipulate a file with a higher
scan resolution, but you are likely to spend less time fixing a file
scanned with a higher scan resolution.
Item 2- Raster to vector software.
The purpose of raster to vector software is to produce a vector file
(lines, arcs, or curves) from a raster or bitmap file. This is often
referred to as a tracing.
There are two primary types of raster to vector conversions; outline
tracing, and centerline tracing. During the outline tracing process the
software "looks" at the dots on the bitmap image and attempts to fit
vectors along the edges of the black areas. This produces an outline
which gives this method its name. The second type of raster to vector
conversion, a centerline trace, is usually used when the original image
is composed of lines of a consistent thickness. This method attempts to
find the center of the original scanned lines and fit a vector through
the center of the original lines.
There are many different raster to vector software packages available on
the market. We have had the best overall success with "Streamline" from
Adobe systems, Inc. Streamline is quick, will perform both types of
conversions (even at the same time) and has given us, by far, the best
centerline conversions. Streamline will show the original scan, as well
as the finished vector file. It also has the ability to smooth the
vector file before you export it. Streamline will export both DXF files
and AI files. The DXF files have the same problems we started with and
are not a good export option. The AI files are the most appropriate
Item 3- Illustration software.
Good illustration software when applied to the machining process should
allow you to edit bezier curve files, generate drawings from scratch and
produce and manipulate lettering. The best software for these purposes
is less clear than in the raster to vector software. Two good choices
are Adobe's "Illustrator" and Corel's "Draw!"
An obvious question often asked is, "What is the difference between
illustration software and CAD software". To answer this question we must
look at the purposes for which each of the two classes of software were
designed. CAD software was designed to mimic a draftsman, as such, it's
very good at drawing and manipulating geometric entities, if you need to
draw something easily defined with precisely placed lines and arcs, CAD
software is by far the best tool available. Illustration software was
created to mimic the graphic artist. While it is true that some graphic
art is very geometric, most artwork and lettering is composed of smooth,
flowing curves which are very difficult to manipulate with the tools
typically found in CAD software. Although these two types of drawings
are not mutually exclusive, there is no doubt that each type of software
is more proficient within its respective category.
We use Corel Draw! as our clearing house for problems. For instance,
it's not uncommon to scan and trace a logo only to find that the artwork
looks good but the lettering is less than adequate for quality
machining. In this case we select the closest lettering style available
within Corel Draw! and then use Draw's! text handling ability to stretch
and move the text to match the original text without duplicating the
problems. There are times when we are given artwork that is unsuitable
for raster to vector translation. Corel Draw! has the ability to import
the original bitmap scan as a grayed out image. This image is then used
as a template to recreate the artwork using the drawing and text tools
available within the software, (this is, essentially, "on-screen
digitizing"). We also use Corel Draw! to produce our own artwork and
lettering from scratch. If all we need is a few lines of lettering to
engrave on a part or sign, we can start to cut the lettering in minutes.
Using the first three hardware/software items just discussed, we are
always able to find a method to produce the artwork and lettering,
regardless of what we are given as a starting point.
So far so good. However, this much we were able to do years ago. There
is still a problem when we want to machine the artwork, none of the
previous software packages produce machinable files, and now we come to
the final part of the solution...
Items 4 & 5- InterFlux and your CAM system.
We had experience with the artistic files and their manipulation, we
also had experience with CAD/CAM and machining. There was still the
problem of going from one to the other. This is exactly the purpose of
InterFlux. With the ability to import the files produced by the artistic
software and change these files into a machinable form, while retaining
a high precision and reducing the entity count, InterFlux proved to be
the necessary link in the final solution. Internally, InterFlux works
with a double precision database which gives us the ability to export
files which are completely accurate 12 places after the decimal point.
This virtually eliminates problems with endpoints not matching.
InterFlux was designed to be interactive so the user could visually
check the results produced within the program prior to exporting to a
CAM system. Many other facilities are included which apply specifically
to easing the transition to a CNC machine. Artwork can be easily and
accurately sized, placed in relation to the machines origin, rotated,
duplicated, mirrored, etc. If the final intent is to engrave with a
centerline cut, InterFlux allows the user to assign up to eight
different depths and feedrates to the image and will even show an
animated tool and cutter path based on a tool diameter entered by the
user. With an appropriate post processor InterFlux can then produce
centerline NC code.
Speed is of the essence when dealing with artistic files, the sheer size
of the files slows down most CAD/CAM systems. We have often worked with
files that started with 50,000 or more lines. InterFlux was optimized to
redraw these lines at a greatly accelerated rate. Once the files are
converted to lines and arcs in InterFlux it is common to reduce the
entity count by over 90%.
If more advance cutting operations are necessary such as contouring with
cutter compensation or pocketing operations, InterFlux will export to
common CAD/CAM formats in the form of high-precision lines and arcs.
These formats include *DXF, *CDL, and *NCI. There are plans to continue
to add additional file formats in the future.
After using InterFlux and importing these files into a CAM system the
files are either free from problems or at least, very nearly so. Complex
pocketing is easier due to the low entity count, and the finish quality
is improved because curves are cut with arcs instead of linear segments.
The low entity count also produces a faster cutting file without the
associated block transfer problems.
Perhaps our biggest surprise was the number of different industries
which have found InterFlux to be useful. Currently InterFlux is being
used as an integral part of the machining process in industries such as:
roll die manufacturing, engraving, laser cutting, water-jet cutting,
guitar manufacturing, church pew and alter manufacturing, jewelry,
signage, and more. Typically these users are able to produce machinable
files in 10 to 20 percent of the time it previously took them to
generate usable files.
This article was intended as a brief overview of the problems
surrounding machining artwork, and to demonstrate some of the possible
solutions. In today's competitive market, more companies are trying to
bring together all of the necessary processes to produce their work
within their own facility. Machining complex lettering and artwork has
been one of the more difficult operations to implement within the
machine shop. It is only recently that the technology to machine this
artwork has begun to catch up to the machine shop operator's need to
For more information about machining artwork and InterFlux, Mr. Elliott
can be contacted at Northwood Designs, Inc.
ABOUT THE AUTHOR
Bill Elliott, is the President of Northwood Desings, Inc. and started
as a consultant to the CNC industry in 1991. HIs original exposure to
the CNC world was as a light manufacturer using a CNC router.
Currently, Northwood Designs, Inc., develops CAM related software, and
provides specialized consulting services to the CNC industry.
*NOTE: DXF is Autodesk's- Drawing exchange format ,
CDL is Cadkey's- Cadkey Advanced Design Language, NCI is MasterCam's
native toolpath format, Corel Draw! is a trademark of Corel Corporation,
Illustrator is a tradmark of Adobe Corporation