Comparing 3D Filaments with the LS173 Colorimeter.

Both the Linshang LS-170 and the LS-173 have USB ports for charging.  But, only the LS-173 can also link to a computer via USB for color capture.  This was the primary reason for making that additional purchase.  While it turns out that the Windows app designed to work with the LS-173 doesn't report the complete range of color information as does the smartphone app.  It does perform one vital function that I have found to be very useful.

The Windows application is called Color Haze Meter.   The 'Haze' part of the name indicates that the software is designed to work with a variety of products including the LS155 Color Haze Meter for measuring and comparing the clarity of transparent materials and liquids.  I believe it is used in brewing industries.

Here is the screen layout of the Color Haze Meter software:  (Click to see full size)

Color Haze Meter Software for the Linshang LS173

 

TOP PORTION OF THE SCREEN 

The software expects to see two different scans before it renders a verdict on the outcome of the comparison.  In our case we wanted to compare the identically identified colors from Bambu Labs and Marswork.  In every case the published HEX values between the comparisons were identical.  The question we wanted to know is just how closely they actually match.  We first scanned the Bambu Labs and then scanned the comparable Marswork filament.

Since the primary goal of this application is to compare two different samples the color information presented is limited as LAB color values.  Each row contains the LAB color information for the "Standard" that was scanned and the comparison "Sample".  

As each pair is sampled, the last column in the Comparison Measurement spreadsheet provides a helpful comment.  In the case of the comparison of the Bambu Labs Mistletoe Green vs the Markswork Mistletoe Green it reports that the differences in the Lightness (L*) values was "Normal" and the Marswork sample had a "Slight Red" bias and a "Slight Blue" bias.

But, the most important notification is in the bottom cell where it says, "PASS", meaning that the two samples are within acceptable color difference limits.

In fact, in every case, using the default industry standard delta Thresholds of 5.0 every single comparison PASSED, meaning from a visual standpoint they were functionally equivalent.

BOTTOM PORTION OF THE SCREEN

The Color Haze Software allows us to scan batches of samples and with each two scan cycle it stores the result in a line on the lower half of the screen.  In this case, we made 5 different comparisons and changed the name of each line to indicate the color tested.

But, here is the really nice part. With just one button click we can export the results to an Excel spreadsheet where we can format the presentation any way we want!  (Click to see full size)

Formatted Excel Export from the Color Haze Meter Software

 

I find this to be incredibly helpful information.  For it tells me that should a certain color not be available from one vendor, there is another vendor's product that is virtually an exact match!

ABOUT THOSE SLIGHT DIFFERENCES 

Quite frankly, if I hold one of my test swatches for Bambu Labs Basic Jade White up against the side of the LS173, it's clear that the 3D filament is not precisely HEX #FFFFFF or RGB(255,255,255).  The plastic of the injected case for the LS173 is slightly brighter.  The question is, "Why?"

Actually, the published value is the "TARGET" used by the manufacturers to determine the colorants to be used to make the final product.   Asking AI about white filament revealed that raw 3D printer filament is naturally translucent.  So, Titanium Oxide is used to make it as close to the "White" target as possible.  But, from batch to batch there can be minor fluctuations even among the same brand.  That's normal.

This concept of a published HEX color being a "TARGET" is true of all colors.  What is important is the "Perceived Color" of the final product.  As we examine these results comparing LS173 scans of like colors from Bambu Labs and Marswork, we see some differing values.  (Click to see full size)

LS173 Comparisons: Bambu Labs / Marswork

But, if we look very closely at the far right it is clear that the "Perceived Colors" are virtually identical for the vast majority of users.  Can there be perceived differences over time?  Conceivably, but in theory, those same minor differences could show up in the very same product line, since colors are newly mixed with every new batch.

 

Colorimeter Targets, Scanning Setup and LS173 Scanning Tests

Before beginning the scanning tests, the specifications for the targets had to be determined and the targets printed.  Two color printing is required to print the target information; but, we also wanted to iron the top layer of the scanning target.  If you try to iron a 3D printed part having text of a contrasting color it's a mess.  So, we came up with a two-part system, using our existing Marswork ID tags, used with the Marswork Quick-Swap Spools and filament management system.

Two-Part Scanning Targets

Note how the ID tags simply clip into the scanning target for correct identification.  The targets are printed with 100% Rectilinear infill and the insert portion is ironed using a setting of  60mm for speed and 30% for flow.

To ensure that the capture is done at the correct angle and precise positioning, a special holder for each colorimeter was 3D printed that included a stand for the smartphone.

LS173 Colorimeter Stand

 

The stand has a slot for the target that places it at the same exact spot for each scan.  Both the LS173 and LS170 can be rotated in 90 degree angles as necessary.

The same tests that were performed using the LS170 were completed using the LS173.  In both cases, repeated scans were consistent and within a point or so.  In this regard, the LS170, is just as consistent as the LS173.  At half the price, it's probably the best value.  Although, the screen of the LS173 is very useful.

Here are the results of the LS173 color captures. (Click for full size)

The lighting system of the LS173 is diffused and it is the camera that is offset at 8 degrees.  Based on what I see here, the rotations made little or no difference and there is actually little difference between the ironed and un-ironed surfaces.  Generally the difference is less than 2 points!  But, with black, the difference between the published value and the tested values was significantly greater than that of the LS150 by almost 20 points.  With White, the two colorimeters were in agreement.  

None of the colorimeters in the price range with which I am working deal with 'sheen'  as a separately tested quality.  I have a feeling that 'sheen' plays a factor in testing 3D printed parts.

Now that we know, for sure, that we are not going to be able to match ANY published color spec with an affordable colorimeter, we need to determine if and how the information they do give us can help in our work.  We also want to explore whether HEX & RGB() or LAB values are where we should put our focus. 

 

Coming up with a Colorimeter Strategy for 3D Prints

 I used the Behr swatches as a staring place, because paint comparison is exactly the application the manufacturers had in mind when designing their products.  Paint swatches are professionally printed uniformly on smooth surfaces.  Thus, they provide a good starting place with which to compare the performance of the colorimeters I own.

3D prints, on the other hand, are NOT uniform surfaces.  They are, by the very nature of the way they are produced, corrugated in form, reflecting light in lines of brightness and darkness in a photo.  Here, for instance, is the bottom of our test print we are using for our colorimeter work.

Bottom of 3D Print Sample

 

The corrugated nature of the 3D print is plain to see,  And, it's easy to imagine that the very nature of the surface would make at least SOME difference in our colorimeter readings.

We can minimize this effect by smoothing, to some extend, by "Ironing" the top surfaces of our test objects.  Here is the top of our test object that has been 'Ironed" to increase smoothness.

"Ironed" top of 3D Print Sample

 

While not as perfect as the surfaces of our Behr color samples, it is at least somewhat improved.  Given the differences in surface quality I decide to test whether these two surfaces return measurably different results.

The LS170 was selected for this test because of the lighting/capture system it uses. 

The LS170 relies on a 45°/0° capture system, meaning the lighting hits the target at an angle of  45° and the camera capture directly from above at 0°.  

 I wondered if the lighting angle might reflect differently in one orientation as apposed to another due to the artifacts in the 3D print.  So, I also decided to see if rotating the LS170 90 had any affect on the outcome.

This meant each sample would be tested 4 times.  The top would be sampled twice, at 0° and 90° as would be the bottom.  The LS170 lends itself to this reorientation due to it's square shape. 

The differences (Delta) between the published Bambu color specs were then calculated for each test. 

THE RESULTS

Five different Bambu Labs filaments were selected for testing.  The goal was to compare the results between the ironed top and raw bottom of the print sample.  Each of these were tested twice by rotating the LS170 by 90°. The differences (Delta) between the published Bambu color specs were then calculated for each test.  (Click on image for the full size)

While these results haven't been fully analyzed, it seems clear that the ironed top surface generally produces smaller differences than the bottom counterparts.  What is obviously clear is that LS170 ALWAYS returns a darker shade than the specs of the target color.  It's not enormous, in most cases; but, visible in the samples.

I was actually quite surprised that rotating the LS170 didn't seem to make all that much difference in the results. 

We plan to run the same test using the LS173, which relies on a different lighting/capture system, before scanning all of our current inventory.  I suspect our Delta values will be much smaller overall.  But, I do NOT expect to ever match the target specs.  These best we might be able to do is come up with a uniform offset that can applied.  More importantly, as the library of color samples continues to grow on FilamentColors.xyz, we will be building our own library of colors for entering into their search engine.  And, we will have the tools to locate a closely matching color should we need to search for one.

UPDATE 

After my initial posting of this article, I realized that it was a major oversight not to have included the color "Basic Black: in my tests.  So, here it is:

Why this is an important inclusion is that it gives us the bandwith of the colorimeter.  By sight we can see that Bambu Black is as black as one could expect.  Yet it does NOT register as uniform zeros across Red, Green and Blue by the colorimeter.  The light source of the colorimeter is bright enough, and the surface reflective enough,  to register almost 50 units above zeros across the spectrum.

When we combine this with the readings of around 20 points LOWER for white than we might expect, it's clear that, at least when measuring rough surface plastics, the bandwidth of the colorimeter is less than the full spectrum.  This is common in all digital photography.  Professional photographers use a histogram to pull blacks down and whites up to achieve the proper contrast.  

It doesn't matter all that much if our goal is simply to compare a specific color of one brand against another as long as we use the same colorimeter using samples printed in the very same way. 

 

 

Testing the ColorReader EZ, LS170 and LS173 with Behr Paint Swatches

The most professional approach to testing a colorimeter is probably using Pantone swatches.  But, my budget wouldn't allow that.  So, I took the poor man's approach.

I made the assumption that paint manufacturers have to pay very close attention to make sure their color swatches precisely match there published color values.  Fortunately, every HomeDepot has  a broad selection of Behr paint swatches and the color values of every one of those swatches is published on Behr.com.

(By the way, matching paint colors is precisely the application for which these colorimeters were designed.)

I gathered several swatches, looked up their color HEX and RGB() values on the we site and began testing with each of my colorimeters.  Pay close attention to the DELTA columns, as they document just how far off the reading are for Red, Green and Blue of each colorimeter. (Click for full size)

 

The first thing that stood out to me is that there is no standard of deviation.  No colorimeter is consistently off by a certain direction by a set amount. That is, none of the colorimeters consistently measure darker or lighter.

But, the tests also verified that the LS173, with the D/8° lighting system is generally the closest match.  The LS170 uses a 45°/0° lighting/capture system.  The first value, 45° indicates that lighting is cast at a 45° angle and the second value indicates the camera takes an image at 0°.

The LS173, on the other hand, uses a sphere diffusing (D) lighting system and captures the image at 8° offset.  this probably helps reduce lighting artifacts from specular surfaces. 

It's important to remember that these professionally printed swatches are smooth and our 3D prints are relatively rough and uneven, even if we use ironing.  So, it is no surprise that the numbers we get with testing never match the published numbers from the filament manufacturers.  The best we can expect is that our colorimeters will provide us with a very, very close match that we can use to find other brands of filaments having a nearly identical profile.