Come on, those guys are not “scientists”; they are university students.
Why do people do research? A few reasons:
1. As a student, if you want to get into a master’s program or pursue a PhD, having research papers helps with acceptance.
2. To start or support a business.
3. Why not both #1 and #2?
Let me entertain this idea for a moment.
One of those guys is in the medical field. The image you showed in your initial post illustrates Prusa being “the best” slicer. Coincidence? I don’t think so.
Let’s create our own “dental” slicer (based on open-source Prusa) and sell it for $5,000—or even $10,000—based on a scientific research paper we wrote that proves the superiority of the product. No dentist will say no to that, and $10,000 is actually a small price to pay. Considering the dental boom in Canada due to government subsidies, it sounds like a great business idea.
I’m being intentionally vague and somewhat entertaining here because I don’t have anything against those kids doing this. If they find a buyer, good for them. Let them succeed. And I’m sure that with solid “scientific” proof, they probably will.
But don’t come to a forum where people print slinky dragons on specific hardware and tell us this is for our benefit and that it will change the world, because it isn’t and it won’t.
I’m not claiming I’m smarter, so please don’t twist what I said. I’ve been in this business for the last 30 years. I started by writing G-code on a piece of paper with a pen and manually entering it into CNC machines through keypads because computers didn’t yet have interfaces for CNC machines. I also spent four years teaching the subject in academia, although that’s irrelevant here. They are probably smarter than I am.
What I am saying is this: don’t pick PrusaSlicer to generate G-code because a paper claims it achieves higher accuracy. This paper is scientific slop written for a particular personal objective (good for them if it helps them succeed). It is not a paper intended to help the broader 3D-printing community achieve better accuracy.
Please don’t gaslight me by claiming they are doing this for us.
No one is gaslighting you - you’re doing a great job all on your own of badmouthing good projects for the community. There’s no need to reply - I’m not going to respond to your flat-earth theories anymore since you’re just trying to derail the conversation.
Strong words from someone who’s flooding Makerworld with AI slop. I went ahead and blocked your Makerworld account too.
You had your chance, but you didn’t take it. Now after you’ve doubled down last time, you want to go all in? You should have asked your slop AI what the video is about.
Not with me - there’s the door.
Agreed. I would’ve been more inclined to have found his results with the scanner to be more valuable if he had performed more than one scan to create a dataset across slicers. A separate video would give a better run time to dedicate to performing a proper experiment.
I can’t remember what it was that I used a similar software for, but I was able to resurrect a model that I only had the gcode for. I felt a little like the Jurassic park scientists who grew dinos from a mosquito’s last meal. It was good for what it was, but also it wasn’t a precision part and it certainly wouldn’t have escaped the enclosure and eaten the lawyer.
I can see the use case for laboratories and such where a “poopsie, it’s off by 0.05mm!” can lead to catastrophic results. Any part of the printing process can introduce errors. If this research leads to improved accuracy whether by elimination (doubtful) or refinement to bring the margin of error down to a quantifiable (and thus predicable) amount, then all the better.
As for me, while my models require precision, they’re also kit cards that make miniature things. Nobody gets hurt if the accuracy is off and the model doesn’t fit together 
While FDM 3D printers tend to have minor inaccuracies they tend to have great repeatability. For instance, if a part is designed with 3mm diameter holes, when printed the holes will have a diameter of around 2.9mm. One can tweak their slicer settings all day long and never achieve a hole with a diameter of 3mm. In order to print a hole with a diameter of 3mm, one has to design the object with a 3.2mm hole.
I come from the aerospace industry and have a habit of designing things with tight tolerances. Many of the models that I have on MakerWorld rely on tight tolerances for the joints to hold tightly. As long as I can get a joint to hold tightly, or move smoothly, as the case might be, the absolute dimensions become secondary to the relationship from one part to the next. Once I have a model fully designed, even though the exact dimensions may be slightly off, the relationships hold regardless of which printer the model is printed on.
A great deal of technology has been developed at universities. Many of the larger ones are called research universities. A student spends the first four years learning what is known. Doing original research is central to pursuing a PhD, even though that original research often aligns with whatever the professor in charge is doing. Linus Torvalds was in college when he created Linux. Study after study has shown that people tend to come up with their best ideas much moreso when one is young as compared to when one is old.
There is a setting for Bambu fillaments that will get you really close to the hole sizes, although they are still slightly off and not repeatable from the X, P series to the H series. The H series always produces smaller diameter holes for me, even when using the new setting that I can’t remember its name.
And on that, I have some stands that use dove tails to slide together and hold. I have 3 X1Cs that will print the same files in various tolerances. It is actually quite annoying, for quite awhile I kept adjusting the sizes in CAD software thiking that it was random, sometimes the parts were too tight, sometimes too loose. Then I realized it was dependent on the printer. I haven’t started sizing for my other printers yet, but I know which printer prints the part the best on, and use that one when it is free.
I really should try these on my H printers and the X2D. I have the calibration plates, I wonder if that will help…
Mind you, this isn’t complaining, I long ago came to accept that 3d printers are not perfect. These are just observations.
I’ve had the same experience where my A1 mini would produce looser parts than my two X1Cs. It was consistent across projects and filaments. I suspect it has to do with how each printer type handles flow ratio or possibly the inherent differences in coreXY vs bedslinger. I don’t have anything to back that up. Just idle musings.
At any rate, I no longer have the A1 mini (sold it to a friend so I can rope more people into the hobby) but have been eyeing the H2C for awhile now. I’d be curious to know how it compares.
When I first got into 3D printing, I watched YouTube videos where the presenters would go through their slicer settings, at times selecting a couple dozen different settings, before every print. To me it seemed like such a waste of time. When I decided to make models to put up on MakerWorld I made the decision that everything I publish on MakerWorld can run on standard configured Bambu Lab printers. Which includes the standard Bambu Studio slicer settings. That is to say that I design to the machine. As such, anyone with a standard configured printer, should be able to reproduce what I have created. This is the point that the open-source crowd seems to miss. That the sharing of 3D models across different printers from around the world works best when those doing the sharing have similarly configured machines.
As I like to point out to newbies. The ease of printing is often a function of the quality of the underlying model. With a good quality model, printing is a breeze. But with a poorly designed model, 3D printing becomes a nightmare.
I think maybe the real takeaway here is that if you’re manufacturing many parts that have to fit together, you are heavily incentivized to use the same printer slicer software for every part of the assembly, or suffer some integration ■■■■.
Some of my projects bridge orca, Bambu, and qidi slicer and I think in retrospect this might explain the one or two extra drafts I had to produce to get holes super lined up. Or I’m just bad at design which is a given
Not everyone can use arachne, as it is not backwards compatible. I myself have models from long ago that slice without it, and when arachne is enabled, things go wrong.
As long as the mode was the same across all slicers, or was stated for the slicer where different, it’s fine.
As for the merit of the paper, it’s known that there are accuracy and precision issues with 3d printing, and this seems like a reasonable attempt to measure parts of it. Knowing where things go wrong will help developers tune the programs.
I think this is reason to justify using the bambu vision encoder plate. When you have models that will print across multiple machines and you need them all calibrated the same.
I agree with you.
Another issue is which file format to import into the slicer - this has been a topic of discussion for years and just adds to the list of problems. That’s why I don’t want to go into too much detail about it; I’ll just mention it.
STEP vs STL
A topic that never seems to settle down, but also plays a part. Or, to put it another way, the entire workflow. This thread alone has been going on for several years:
But before I stray from the topic, here’s the key point: It’s important to consider the entire workflow - regardless of the printer - because anything that happens along the way can lead to inaccuracies.
It would be interesting to see what would happen if they had conducted the same experiment with different file formats - to see if that makes a difference with large, complex objects.
Hi There,
i am the “dude” that created the Video and also one of the “dudes” that wrote that research paper.
So now u got my contact details 
Calling this “dishonest science” is quite a step — and not one I expected someone to casually take on a Monday morning.
You are aware that scientific publications go through an extensive peer-review process involving multiple independent specialists, right?
More importantly, if you had actually read the paper, you would know that this research was not performed on arbitrary datasets. It investigates a specific real-world use case, geometry, and application where these deviations matter.
We specifically used mandibular CT segmentations because hundreds of anatomical models like these are printed every year for patient treatment in our facility alone. They are used for reconstructive surgery, surgical approach planning, pre-bending of reconstruction plates, and patient information.
So saying that the problem could simply be “optimized away” or designed around entirely misses the point. In anatomical 3D printing, you cannot redesign the patient’s anatomy to make it more convenient for the slicer. The dataset represents a real anatomical structure.
This is also precisely why we did not optimize individual path planners, compensate geometries, tune extrusion parameters around the effect, or use other specialist workarounds. The research is aimed at users of 3D printing, not at demonstrating what an expert can achieve after extensively optimizing a specific dataset.
The objective was to quantify and communicate an effect relevant to a specific field of additive manufacturing.
You would know that if you had read the paper’s Aim of Study before accusing its authors of scientific dishonesty.
Disagreeing with methodology, scope, or conclusions is completely legitimate. That’s how scientific discussion works. Calling research “dishonest” because you disagree with its scope is something very different.
And making an accusation of scientific dishonesty toward a research team whose members each have 10–20 years of additive-manufacturing experience and actively work in AM research is quite a claim — particularly when it comes from a ragebaiting forum post that apparently didn’t get as far as reading the stated aim of the paper.
As for my Source for all those informations:
Im the author of the Video as well as one of the authors of the Study…
Wow.
You are quite a piece of work, hahaha.
Let’s start with this:
“Those guys are not scientists; they are university students.”
If you had spent even two minutes looking into the authors before writing all of this, you would have realized that there wasn’t even a Master’s or PhD student involved in this particular paper.
The authors are established professionals, well advanced in their careers, working in additive manufacturing, medical 3D printing and research. Some of them hold PhDs, doctorates and professorships in these specific fields.
This research was done by people who use 3D printing to help patients on a daily basis.
So your first theory — that this is some student research done to get into a Master’s program, pursue a PhD or improve someone’s chances of getting hired — already falls apart there.
I started with pen and paper too. I’ve been doing this for roughly 25 years, including CNC machining and 3D printing. And since you mentioned your four years of teaching in academia: I’ve been teaching additive manufacturing myself since 2017.
Now to your rather entertaining Prusa conspiracy theory.
You suggest that because PrusaSlicer performed well in the study and one of us works in the medical field, perhaps we’re planning to create a $5,000–$10,000 “dental slicer” based on Prusa and use this paper to sell it.
I’m going to tell you a little secret:
We don’t use a single Prusa printer or PrusaSlicer in our hospital lab. At all.
And here’s another one:
We don’t develop or sell machines or slicers.
There is no dental slicer. There is no product. There is no company waiting to commercialize these results.
What we do is research. That’s our full-time job.
We conduct research at a university and translate it directly into clinical practice at a university hospital, where we use additive manufacturing to benefit patient care.
We don’t earn money from publishing this research either. We publish our work Open Access — which actually costs the university a considerable amount of money — specifically so that everyone can access and read it for free.
So you started with the assumption that we’re “university students,” invented a hypothetical commercial product, invented a price for that product, invented a target market, invented a business model, and then used that entirely fictional scenario to speculate about our “particular personal objective.”
That’s certainly creative. It just has absolutely nothing to do with us or the research.
And this part is particularly ironic:
“I also spent four years teaching the subject in academia.”
Yet in the same post you demonstrate a remarkable ignorance of who actually conducts academic research, how university research works, and apparently even of the very basic concept of checking the authors and their affiliations before making claims about their motivations.
And finally, calling the authors “those kids” is quite something.
Half of the authors are well into their 30s, 40s and 50s. Some have PhDs, doctorates and professor titles in these exact fields and have spent significant parts of their careers working professionally in additive manufacturing and medical research.
Calling them “kids” while simultaneously explaining how much more experience you have is perhaps the pinnacle of self-importance in this entire post.
You don’t have to agree with the paper. You don’t have to use PrusaSlicer. And nobody is asking you to change how you print your slinky dragons.
But perhaps before constructing an entire story about who the authors are, why they conducted their research and what imaginary product they’re supposedly planning to sell, you could simply look them up first.
Source: me. an author of the study. (now it is on you: what do i try to sell you in this post? Why am i deishonest here again? One more round of Sebo’s conspiracy storys?)
For a Kid you are doing well! 
Interesting video, haven’t finished it yet but I think very well presented. I’m not the sharpest hammer in the kitchen but did they do the equivalent of say middlemanning a print job sent from a slicer through a cloud, and then looking at what the gcode was actually going to print?
Thank you very much for your detailed comments.
I have pinned one of them to the top of the thread so that people can see them instantly. That means even if someone is new to the topic, they’ll see it right away, and your comment won’t get overlooked.
Unfortunately, I can only pin one. Ideally, I would pin all three, but at least this way people can see it right away. 
I’m struggling a bit with section 2.3 of your paper, are PC1, PC2, … PCn the same as PC1, PC2, … PCn?
Thanks.