The Resolution Myth
Too often I see people confusing printer resolution with print accuracy. This is understandable mistake given the incremental way that 3D printers create objects; however, there is not a true relationship between accuracy and resolution. The accuracy of a print refers to how well that object conforms to its original form, while resolution is simply the smallest feature that can be printed in a particular plane. The smooth surface finish that results from printing at a higher resolution does not mean that you have an accurate product. Printers with lower, but acceptable, resolution specifications may actually deliver superior accuracy.
Dimensional Accuracy of 3D Printers
For orthodontic applications, we need 3D printers that can repeatedly deliver a high degree of accuracy. Since printer resolutions do not provide us with this information, a dimensional analysis of printed products must be performed.
To gather this data, we distributed a digital model of a patient’s lower dental arch to various orthodontic practices. They were given instructions to print the models and then scan the prints using the latest generation 3Shape Trios intra-oral scanner. This was performed for the following 3D printers: a Stratasys Objet Eden 260vs at both the High Definition (16Z, 42XY res)and High Speed print settings (30Z, 42XY res), a FormLabs Form2 at both 50 and 100 micron layer thickness (Z res), an Envisiontec Vida that was calibrated to 73 microns XY resolution (100Z res), and a Park Dental Research Juell Flash OC calibrated to 100 microns XY resolution (100Z res). Using the original .stl file of the mandibular arch and the .stl files of the scanned prints, the dimensional accuracy of the dentition and marginal gingiva on the printed models was assessed by an independent third party. To validate the consistency of the dimensional analysis, the Form2 50 micron file was submitted twice without the knowledge of the evaluator; the outcome of the analysis was the same.
Accuracy Results
All prints showed high dimensional accuracy with less than 100 micron average deviation. The differences in dimensional accuracy between printers were not clinically significant. Studies have shown that for a scan performed under extra-oral conditions the accuracy for a full dental arch is within 25microns and the minor differences between the printers were within the margin of scanning error.
Of the printers tested, the Envisiontec Vida showed the highest print accuracy; the Stratasys Objet Eden 260vs at the high speed print settings (30 micron layers) demonstrated the lowest print accuracy of the printers tested.
The overall accuracy results underscore that fact that high resolution does not mean high accuracy. This was particularly evident in the performance of the Stratasys machine. The accuracy data recorded for the Eden 260vs was in line with the stated accuracy of the printer from 3D systems, but it’s print accuracy was significantly less than the printer’s resolution and was not significantly different than more cost-effective machines.
Another interesting finding was that the most accurate prints were attained with 100 micron layer thickness (Z res) and when a finer Z resolution was used on the same machine (50 microns on the Form2 instead of 100 microns) the accuracy of the print did not improve. This would suggest that the size of features present in orthodontic models and appliances are not small enough to warrant selecting a layer thickness smaller than 100 microns. This means that the faster print speeds attained at 100 micron Z resolution can be used on these printers without degrading accuracy.
Choosing a 3D Printer
Manufacturers’ stated printer resolutions are no guarantee of accuracy. When evaluating any new printer for purchase, the dimensional accuracy of the prints must be validated and contrasted with exiting purchasing options. However, the results of this analysis demonstrate that price certainly isn’t the determining factor in print quality and accuracy. All printers tested demonstrated similar and sufficient print accuracy for orthodontic use, and the extremely cost-effective Formlabs printers were just as accurate as much more expensive printers. If you are paying more for 3D printers in the current marketplace, you aren’t doing your homework and it is time to start make better financial decisions.
Given the price point of printers like the Form2 and it’s high level of accuracy, there is no reason not to integrate 3D printing technology in your orthodontic practice. Unfortunately, some doctors who have their own small orthodontic labs are attempting to discourage orthodontists from investing in 3D printers. They jumped the gun on the 3D printer market and still need doctors to outsource their 3D printing services to them in the hopes that they can recover their investment on overpriced 3D printers. Don’t fall victim to their agenda.
3D Printer Accuracy Heatmaps & Data
Dimensional accuracy heatmaps and data for the 3D printers that were analyzed are shown below:
Stratasys Eden 260vs
High Definition Settings (16 Micron Layer Thickness)
The printed model demonstrated an average dimensional accuracy of 57.3 microns. The printed models were dimensionally accurate to within 80 microns over 78.55% of their surface area. The maximum deviation from the original model was 120 microns.
Stratasys Eden 260vs
High Speed Settings (30 Micron Layer Thickness)
The printed model demonstrated an average dimensional accuracy of 84.15 microns. The printed models were dimensionally accurate to within 80 microns over 50% of their surface area. The maximum deviation from the original model was 170 microns.
FormLabs Form2
@50 Micron Layer Thickness
The printed model demonstrated an average dimensional accuracy of 73 microns. The printed models were dimensionally accurate to within 80 microns over 64.7% of their surface area. The maximum deviation from the original model was 150 microns.
FormLabs Form2
@100 Micron Layer Thickness
The printed model demonstrated an average dimensional accuracy of 67 microns. The printed models were dimensionally accurate to within 80 microns over 65.64% of their surface area. The maximum deviation from the original model was 125 microns.
Envisiontec Vida
@73 Microns XY Resolution
The printed model demonstrated an average dimensional accuracy of 49.4 microns. The printed models were dimensionally accurate to within 80 microns over 83.58% of their surface area. The maximum deviation from the original model was 146 microns.
Park Dental Research Juell 3D-2
@100 Microns XY Resolution
The printed model demonstrated an average dimensional accuracy of 57.7 microns. The printed models were dimensionally accurate to within 80 microns over 68% of their surface area. The maximum deviation from the original model was 130 microns.
Structo DentaForm
@96 Microns XY Resolution
The printed model demonstrated an average dimensional accuracy of 57.5 microns. The printed models were dimensionally accurate to within 80 microns over 66% of their surface area. The maximum deviation from the original model was 108.7 microns.
Uniz Slash+
@75 Microns XY Resolution
The printed model demonstrated an average dimensional accuracy of 90 microns. The printed models were dimensionally accurate to within 80 microns over 40% of their surface area. The maximum deviation from the original model was 177 microns.
Great article and wonderful study. I am a DTS for Henry Schein so I sell the 3Shape TRIOS scanner and it’s a fantastic product, and recently I have been researching 3D printers. Which of the tested models would be your suggestion for the best value purchase? If I understand the study correctly, it seems that it’s not necessary to select a 3D printer with finer that 100 microns resolution since the finer resolution doesn’t translate to greater accuracy. I am encountering a lot of people in the field who are high on the Formlabs 2 (mostly for printing surgical guides) so where personally would you rank this printer? Thanks again for the post.
All of the printers tested had a high level of accuracy, so they are all great 3D printers. Using cost-effective printers like the Form2 (laser-SLA) to slowly scale a small print farm is the best option for people to get started in the current market. However, as their needs evolve, they may find that they require additional printing technologies (i.e. DLP, Polyjet, MSLA, etc.) in their workflow.
Great article overall, my only concern after review is how some of the data was analyzed. Some data in the posterior of the models below the CEJ on the base is selected, ie. the EnvisionTec&Form50um analysis, which will lower the overall accuracy result. On the 100um Form labs model and Stratasys model, that same data is not selected which will artificially increase the overall accuracy. For a true comparison the exact same data points should be analyzed. I suggest that either the entire model is analyzed or only above the CEJ is analyzed for a consistent comparison. In my opinion the true data we care about is the accuracy of the teeth, and that should be the only data analyzed to keep the study consistent.
The visual representation here is simply to show people people an easy to see a heatmap. The analysis was done exactly how you are suggesting, though I requested 1mm of marginal gingiva be included to have a consistent means of measuring some soft tissue. The gingival accuracy is important for Hawley retainers and other appliances. To be sure about the method, I attempted to trick the independent examiner by sending a modified file under a different name and it yielded an almost identical result. Accuracy is accuracy, and the bottomline is that all the printers are equivalent from a clinical standpoint given the average variance from scans. I am not sure what your agenda is (though I see your email is from an Enivisiontec domain), but I my interest was finding out the real numbers. If you are interested in furthering that end, please send me the files you promised in our conversation in LinkedIn. I am working on having these files provided by other means, but it would be far simpler if you already have some.
Interesting experiment. We need more of this coming to not make predictably irrational decisions 😉
Would you be willing to share the mesh files (scan of 3d models printed by different printer and the original stl. file) that you did the analysis on?
What are your thoughts on the operator bias during printing? How much could printing setup influence the accuracy? Based on what I can get from your post, your N=1 for each printer.
I had different operators print and email the files, so they were all pretty self-interested in getting good prints.
I would love to get a large sample to demonstrate the repeatability of each printer, though I doubt it would be much different. I was paying the engineer to do the evaluation based on the area I wanted measured, so that was one reason the sample size was small haha.
If you want to design another larger test, I would be happy to assist. I will email you.
Merhabalar, oldukça güzel bir karşılaştırma olmuş.
Üretim hızı olarakta karşılaştırmasını oldukça merak ediyorum.
Google Translate:
Hi, it’s a pretty good comparison.
I am curious about the speed of production according to the degree of forwarding.
Thanks! At scale speed is not really an issue. You can purchase several “slower,” more cost-effective printers and produce more per hour for less money.
Thank you very much for this great article : I was really looking for this type of summary !
I’m fairly new to this field and I was wondering which 3D printer would you recommend to start with…
Everyone is talking about the Formlab 2…, but how about Ultimaker 3 : could it work to print dental models for aligners ?
Which resolutions should I choose if I only print models for aligners : 20-50-100 microns ?
Thanks for your help
The Z-res should be 100 microns. Nothing to gain with dental models with a smaller Z resolution.
The Ultimaker could certainly print dental models as well. I have not verified it’s accuracy, but as long as it is properly calibrated it should yield quality prints (FDM printers can be finicky)
What units are the scales in the gradient diagrams?
They are in millimeters. The written summary of the data provides better detail than the visual representation.
Hi, I wonder wich resin did you use on the Form2 because I print dental models with Dental Model V2 (brown-yellow) and I got always smaller models than the original file…
The testing was done with the standard grey resin. What you are describing is a printer calibration issue.
Hi, what’s the software you used to compared?
thanks
I paid a mechanical engineer to do the dimensional accuracy testing. Off hand I am not sure what software they used.