3D Printer/Scanner/Modeling Software on Kickstarter - Looking for feedback

Hey everyone,

Before I write anything else, we are looking for feedback on our printer, scanner, and modeling software combination.

I work at Radiant Fabrication and we just launched our 3D printer & scanner, the Lionhead, that connects directly to our 3D modeling software, Li.

[www.kickstarter.com]

[radiantfabrication.com]

The Lionhead Bunny, our beta machine, features 4 parallel printing printheads and the production version has 8 printheads.

Li is a voxel based modeling software that has a similar interface to Minecraft, but has tools to allow you to create large objects/structures quickly.

What does everyone think?

Thanks in advance,
Nate

I don't like to be "that guy", but you know RepRap is primarily about open source projects don't you? Sure we entertain some topics about proprietary devices, but the focus is on building and making, rather than buying off the shelf and using.

So unless you Lion thingy is open source, I would say a) your posting just looks like an advert, and b) you haven't done your market research very well, if you think that RepRappers will be very interested.

So it does puzzle me why people post this sort of thing here, unless it is simply the "spam everyone!" approach to marketing.

Hey bobc,

Fair points. I'm posting here not to advertise, but to get feedback from a community I used to take part in. I owned an early Makerbot and used to work with people in my area to create various versions of Repraps by printing them out on my Makerbot, that would always break, or a high end Stratasys that I also had access to.

I helped to start Radiant because even though I could help people get 3D printers up and running or help repair them, lack of 3D modeling, design, and technical training ultimately resulted in people not being able to use their printers.

I fully understand that the Reprap community focuses on research and development of open source solutions, and I posted in hopes of getting honest feedback from people who use, assemble, and develop 3D printers themselves.

Thanks,
Nate

If you truly wanted feedback and not eyeballs on your KS page, I think you'd have posted about it before going live with your Kickstarter, no?

What sort of feedback - technical, aesthetic, business, legal or something else?

I suppose it is nice of old friends to drop by and say how well they are doing for themselves, but to give something back to RepRap would involve providing some open source designs we can study, build, improve, rather than just the opportunity to buy your printer thingy. Which looks very nice btw, hope it does well.

Hey bobc and greenman100,

I had previously avoided posting here because of how I thought a non-open source solution would be received. We respect the work that people have been doing on the RepRap project, which is why I'm posting. We were hoping for an honest opinion on our offerings for people who are interested in 3D printing but haven't decided to use a RepRap.

Specifically:
1) Our software uses voxels instead of boundary representations, trading off continuous faces for what we believe is a more intuitive modeling interface. Could you see yourselves using a voxel-based modeling tool? Or do you have sufficient background in boundary modeling that you're not interested in using a different interface?

2) We print with multiple printheads and have tried to make the results of our prints as close as possible to the voxel based models. Initially the voxels in our software are 0.5mm on a side, meaning that is the smallest feature size that you can create. Is this prohibitive to the types of objects that you print on a regular basis? If so, what scale would be small enough? 0.25mm, 0.1mm?

Thanks,
Nate

I think your software is an interesting idea but not something I would personally use. While it may be true that most printers cannot make an extruded path less than about 0.4 mm, the positioning they can achieve is much finer than that, and using normal "boundary representations" works out great. Often times I need to make a bracket or some other fixture that will fit onto something else; this precise fit will never be possible using your voxels. Also, the production of smooth faces is fairly easy for a 3d printer to achieve and I feel like restricting to 0.5 mm cubes would be limiting it too much. I will agree that in some cases, such as a shallow slope on the top face, it may be difficult for an FDM approach to produce a smooth face but with layer heights of 0.1mm it's still way smoother than anything your voxels can make.

In short, I don't see the voxels as useful for anything besides making toys, and fairly limiting even for that.

Maybe if you make the voxels instead represent the finest positioning the printer can achieve, they would make a bit more sense in my opinion, but then they would be far too small.

I am however pretty intrigued in your 8-nozzled printer and how well that can actually print in multiple materials when other people are struggling with only two. I could not view the videos on your kickstarter page but your lack of example multicolor models suggests to me that I am correct to be hesitant of this design. Why does the picture show four plastic filament entry ports but all four of them using the same color of filament?

Hi destroyer2012,

Thanks for the feedback.

While the initial launch was intended to match prints to voxels, we are working on averaging/smoothing algorithms so that the surfaces have better finishes. Do you have any STLs of objects that you have fit together in the past that you would be willing to share? If it doesn't work immediately it will prove to be a great test for our upcoming modifications.

Our goal with Li was to create an intuitive modeling interface that people with no modeling background would be capable of using. We realize that it is not perfect, but we are hoping that with some more user input we can make it powerful enough for both technical and non-technical users.

Our printers operate all of the printheads simultaneously and Li is currently set up to model with a single material, that's why the printheads all use the same color in the photos. However, we have created the code base in Li and set up our hardware to be capable of printing with multiple colors. Our first stretch goal for the campaign is to allow for user-customizable colors in Li and our printers. This goal would also allow us to provide the ability to use PVA as well for the 8 printhead Lionhead. By user-customizable I mean that you would tell Li what colors you had loaded, create voxel-based models with the same colors, and then print them out. So if you want to stick with speed, you could do 2 different colors loaded into 4 printheads each, or you could pick 8 different colors to print into a single object. We will get to this change eventually, but if we are lucky enough to hit the stretch goal we would be able to hire some additional help to either work on that or pick up the slack elsewhere while we make the necessary code changes in the near term.

Here are Youtube versions of the videos if Vimeo was giving you trouble:
[www.youtube.com]
[www.youtube.com]
[www.youtube.com]

The first link includes a time lapse of a smaller print and you can see how the object is passed around between the multiple printheads.

Thanks again,
Nate

Interesting idea. Also, as an owner of an actual lionhead, I like the name.

Looking at it as a product intended for non-technical users, here is what jumps out at me:

Firstly, one of the first things laypeople ask me when seeing a 3D printed object for the first time is "Cool! Can you get rid of these lines, though?" Even with fairly low layer height prints (0.2mm) the roughness of the surface bothers people used to seeing plastic stuff that was injection molded. If you're marketing to people who are interested in the finished printed piece, rather than the technology and the printer itself, the surface finish would stand out to them as noticeably rough, much more than standard "perimeter and infill" type prints. I see your point about the voxel approach reducing the learning curve to almost nothing, but if the printer is designed solely for printing cool stuff for the sake of looking cool, that pixellation (voxellation?) may be a big point against you for some potential customers. It looks like it could end up being like Microsoft Paint--the absolute simplest thing in the world to use, but almost no one uses it for anything even remotely complex because it simply lacks the tools.

Next, to stretch the metaphor, if this is Paint, will you let users upgrade to Photoshop (or, since this is an open source community, upgrade to Gimp)? If your software lets people learn how fun and easy 3D printing is, and they want to start using Blender or OpenSCAD or some other heavier package, will they be able to import that? Or will you let the printer accept traditional Gcode? Because if you sell a printer that's marketed as being designed for beginners, people won't want to get something for two grand if they might outgrow it in a few months and have to spend a few more grand on another fully-featured printer. Also, if you sell a reliable printer with four or eight print heads for around two grand that's not crippled in any way, I suspect it will catch the eyes of a lot of non-beginners, since "how many printheads can I add" seems to be a popular topic around here lately.

Speaking of reliable printheads, if you're going to offer any sort of technical support, I'd take a good, long look at multiple printheads. Many print problems revolve around the hot end--ooze, clogs, stringing, stripped filament, temperature fluctuations etc., and hotend failures can end up catastrophic (e.g. a thermistor fails and the hotend keeps getting hotter until the lining melts or something catches on fire). It seems like adding 3 to 7 more hotends than other printers is making the possibility of a customer having to call you about problems increase by 4 to 8 times. By adding all these printheads, what are you achieving? Faster print times? If so, by how much? Is it so much faster that it overshadows the increased complexity and chance of failure? I know the eventual goal is multi-color or multi-material prints, but you're adding a ton of complexity and failure points for a feature that isn't implemented yet.

Other than that, all I can say is it's a beautiful printer, and best of luck!

Here is an example of the kinds of objects I print: [www.thingiverse.com]

If you can reproduce that with your software I would be quite impressed.

This object is an example of why a lot of people would want 3d printers, and something your software cannot accomplish (please prove me wrong ). What I have done is obtained a 3d model of a protein structure, then slightly modified it (union and difference with the 'peg' parts) so it can be locked together with other structures.

I agree that developing a 3d modeling software for the masses is badly needed, and this software need not and will not look anything like solidworks. I think smoothed voxels are a good idea but I still think you need some kind of "mating surface" where the dimensions are accurate.

Here is an example of what I mean [www.thingiverse.com]

In this object there are two circular holes which fit snugly around the ultimaker drive shafts. Between them there is an organic-shaped arm which could be done with voxels. A lot of my 3d printing (besides the proteins) is making various clamps and brackets to hold some spare part up against something else as part of an assembly. For example, a mount for a stepper motor on an aluminum extrusion, or together with some bearings and axles in an extruder. My approach to these models is to first represent the physical objects in my modeling software, then create some mass and difference the shapes out of that mass to yield the finished product.

Here is my vision for your software:
1) you scan your spare parts
2) the software represents the physical objects in the build space
3) you position the physical objects, duplicating and rotating them as you need to. Maybe you could introduce hinge mates or whatnot to test out the motion of your design, or produce solids of partial revolution.
4) Use a voxel-based builder to create the 3d printed mass that will hold the parts together.

The finished model is then composed of mating surfaces (a difference operation between the voxels and scanned physical objects) fused with the bulk mass which is voxel-based. Maybe the user could also introduce some simple shapes like cylinders and cubes as "physical objects" to create specific voids in a shape where for example some filament would pass through in the case of an extruder body.

If you could make that software I would buy it in a heartbeat.

Personally, I think some people including me, would like to be able to scan an object, and print it out in its smooth form (not all "blocky").

Ah ok, I "get" the voxel thing now. I saw the blocky chess piece and wondered why anyone would print like that, unless it is supposed to be lego effect.

Kinda of ingenious way to utilise multiple extruders, but as mentioned before the gramophone effect is not that popular, so I am not sure lego effect will be preferred.

Thanks for all the feedback everyone. I'll answer them in order...

Hey Topherman,

We agree that the current version of our surface finish could use some work and it is being actively worked on each day.

The analogy to Paint is good initially, however, Li isn't meant to be just a starter tool... just a different interface than nearly every other modeling software. Sure we don't have a full feature set built out yet, but we are hoping to get enough user feedback to help shape future additions and modifications. We could make some type of export, but we would rather add the features that the user is looking to use than send them off in search of another program.

As for printheads, we are likely going to do a teardown video of one of our extruder drives + heater blocks as a Kickstarter update. The quick version: for 4 printheads we have 2 pieces for the extruder drive (total, but not including the four motors/gears), 6 small tubes to bridge and mount from the drive to the heater block, and a single custom heater block for all 4 nozzles. This means that we can get the placement of the four nozzles exactly where we need them AND the extruder drives work without any springs or plungers. This also means 1 thermistor, for which we have code that detects if it has failed. Our code also prevents heating if the heater is on but the temperature is not changing.



Hey destroyer2012,

Which orientation do you normally print the object and do you normally use any support structures? I should have some 'free time' to test it out next week.

I agree mating surfaces are an important thing to be able to print. Lately my printed and reprinted objects to use to test our algorithm changes has been a nut and bolt set.

You should try Li, it has many of the features you are mentioning and its free:

[radiantfabrication.com]

Li is our single interface 3D scanning, modeling, and printing software. Li starts with 5 basic patterns (groups of voxels) in your inventory that can be rotated and scaled, a pyramid, a cone, a cube, a sphere, and cylinder. You can create custom patterns or import STLs from other locations to use as patterns. With the patterns (hotkey'd to numbers or can be re-arranged through your inventory - press the E button to bring it up) you can place or remove voxels. We also have face extrude/sink, measuring, copying, and paint bucket tools.

It is still in dire need of a tutorial, but the F1 button will bring up all of the tool/movement/scaling hotkeys.



Hey Jo1212,

The voxels in Li are currently 0.5mm on a side, but we are testing smaller ones as well. Also, the dimensions of the scans we take are within +-2mm of the actual objects, just as good as a number of other 3D scanners available. Is it the surface finish of the prints that concern you?


Hey bobc,

Yeah, the surface finish is a top priority. We'll be posting updates to the campaign with pictures of improved prints.


Thanks again everyone,
Nate

The orientation I provide is correct.

I've tried your software now and I gotta say I don't understand this trend to interpret minecraft as a 3d design tool. Minecraft is a game where building stuff is supposed to be addictive, not fun or fast or precise.

In other words, building things from a first person perspective, placing block by block is designed to be time consuming and laborious. Sure, people have made really complex things in minecraft, but they spent WEEKS doing it when any CAD package could have accomplished the same task in minutes.

Instead of looking at minecraft you should look at other games where building IS designed to be fast and easy and intuitive. For example, Spore and Kerbal Space Program.

Both games provide an understandable "studio" environment so the user can get oriented. (your software does this too so yay)

Both games have a "point and click" interface which allows users to easily and quickly get views from all directions.

Both games have an extensive list of "off the shelf" parts which are suggestive of how they should be put together.

Neither game has a speed limit on camera positioning or a maximum range at which you can apply parts.

What these games really need is an intuitive and useful method of measuring and aligning things to make soundly engineered parts. This is the area in which you should be making your innovation.

Hey destroyer2012,

Thanks for the response.

I understand your point of view. It is indeed time consuming to build structures in the main game mode for Minecraft because you have to collect resources and removing blocks can take forever depending on the ore type and your selected tool. However, creative mode provides you with unlimited supply of all block types and allows you to add or remove them with a single click. If notch had purposefully designed Minecraft to be time consuming and laborious then creative mode would be in direct opposition to that goal. I'm not claiming creative mode is fast by any means, but it is meant to allow you to create complex structures faster (and more safely) than in game mode.

I agree that some structures could be created in CAD much faster than a Minecraft setting, but how long would it take someone to learn that CAD package in the first place? I've seen people pick up Minecraft and Li controls in less than 5 minutes, but I took multiple semesters of CAD courses in college with hundreds of other people and I still wouldn't say that they had 'learned' the software. Also, after learning the CAD package, how many additional months of practice with the software would it take someone to prepare to tackle something like a 1:1 recreation of the Enterprise:
[www.youtube.com]

CAD comes easy for some people, typically those who are better at spatial reasoning, but for the vast majority of people it isn't a good solution. It seems like we are in agreement on this, but not on what the interface should look/feel like.

"which allows users to easily and quickly get views from all directions"
In Li you can either move around the object to get different views OR use the zoom mode via the Tab key. This allows you to rotate around what you have created very quickly and view it from any angle or distance.

"have an extensive list of 'off the shelf' parts which are suggestive of how they should be put together"
In Li we have created two tools which allow you to use whatever parts that you can find or create. 1) You can import any STL file and use it to place or remove voxels. 2) You can use the Copy/Selection tool to make new patterns/tools from groups of voxels to place or remove voxels. You can also scale, rotate, and move these shapes anywhere in the build area.

"speed limit on camera positioning or a maximum range at which you can apply parts"
In Li, we have three speeds for camera positioning. 1) Normal movement, 2) Running via the Shift key, and 3) Zoom mode via the Tab key allows you to move around your objects quickly from any distance using your mouse scroll.
These games do have a maximum range at which you can apply parts. You are given a space to modify your creature/building/spaceship/UFO/etc and all components must be connected in some way.
In Li we currently limit users to the usable print volume as a way to help people creating new objects understand the available space constraints.

Thanks again,
Nate