A follow up - having done some testing, the desktop version of 123D is more advanced than the browser app, and Inventor Fusion is more advanced than 123D. The wrinkle with Fusion is that it will expire in April, although an update that will continue its availability has been promised.

http://www.123dapp.com
There's a browser app and a desktop version.

There's also Inventor Fusion, which apparently is more targeted to mechanical work:
http://labs.autodesk.com/technologies/fusion

It seems like there's a lot of crossover between them, though!

I've not heard of it. Say more please . . .

Rob

--------------------------------------------------
From: "eddie_walters" <eddie_walters@...>
Sent: Saturday, February 09, 2013 8:20 AM
To: <STMFC@...>
Subject: [STMFC] Re: 3D printing challenges etc.

Has anyone tried 123D from Autodesk for generating STL files? It seems to be targeted at the same market as sketchup, but given that it presumably includes some of Inventor's core, it might be less susceptible to the leak problems?

--- In STMFC@..., "Rob Kirkham" wrote:

well, I can't disagree with you Dennis.

To answer the "other software" question, I use two programs - both free - to
convert and scale a Sketchup file. So first I export a sketchup file as a
3d model. Then I import the 3d model file into Meshlab, open it and export
it as an STL file. Then in Accutrans I open the STL file and scale it. I
know there are less clunky ways but I learned this in the early days of
Shapeways services, when Rene Gourley was the only one I knew who had tried
it. He put me onto these softwares - they were specifically recommended by
Shapeways. They work, so I haven't learned a better way - which I know
others on this list have done.

Rob

Has anyone tried 123D from Autodesk for generating STL files? It seems to be targeted at the same market as sketchup, but given that it presumably includes some of Inventor's core, it might be less susceptible to the leak problems?

well, I can't disagree with you Dennis.

To answer the "other software" question, I use two programs - both free - to convert and scale a Sketchup file. So first I export a sketchup file as a 3d model. Then I import the 3d model file into Meshlab, open it and export it as an STL file. Then in Accutrans I open the STL file and scale it. I know there are less clunky ways but I learned this in the early days of Shapeways services, when Rene Gourley was the only one I knew who had tried it. He put me onto these softwares - they were specifically recommended by Shapeways. They work, so I haven't learned a better way - which I know others on this list have done.

Rob

--------------------------------------------------
From: "soolinehistory" <destorzek@...>
Sent: Friday, February 08, 2013 7:46 PM
To: <STMFC@...>
Subject: [STMFC] Re: 3D printing challenges etc.

--- In STMFC@..., "Rob Kirkham" wrote:

yes - but if using Sketchup to scale the drawing to HO, you can also run
into the 6 decimal place rounding problem if you are not careful. Dividing
most large numbers by 87.1 seldom produces neat 6 (or fewer) digit
quotients. Using other software to scale a full size drawing allows one to
get around that problem. But drawing with the rounding problem in mind
from the get-go is also worth some effort.

Rob Kirkham

I think Rob's problem lies elsewhere.

Let me preface this comment with the confession that I am by no means an expert on Sketchup... In fact, I've never used the software, and all I know I've learned from poking around in their help files and discussion lists, looking for answers to Rob's questions.

That said, the problem lies not in what scale the object is drawn in, or how many decimal places entered points are calculated to. If one were to draw the gusseted end of a hat section brace as a portion of an ellipse, or a NURBS curve, extrude that profile to give it some thickness, lay it over modeled sheathing with V groove edges, and preform the Boolean union operation to attempt to make them one solid, the software is going to preform all kinds of calculations to the limit of its precision, no matter if the original data was entered to three place precision, two place precision, or limited to whole numbers.

The problem is that while software operating with solids as the native environment will either do the operation, or choke right then and there. Sketchup, on the other hand, will happily trim all the surfaces and display them, without any indication that the different surfaces that define what you assume to be a solid do not have concurrent boundaries, and you won't learn that until thousands of operations later, when you try to convert the file to STL format. You end up with something that Sketchup calls "leaky solids", which is basically the software development team's admission that their product can't reliably calculate boundaries with all points within the internal margin of error for concurrent points. It is basically the difference between freeware, and software that will set you back multiple thousands of dollars.



Dennis



------------------------------------

Yahoo! Groups Links

Thanks for the explanation. It would indeed be frustrating to find out at the end that there is an error(or many) somewhere, and not know where to find it!


Scott Haycock
Modeling Tarheel country in the Land of Enchantm ent

--- In STMFC@..., "Rob Kirkham" wrote:

yes - but if using Sketchup to scale the drawing to HO, you can also run
into the 6 decimal place rounding problem if you are not careful. Dividing
most large numbers by 87.1 seldom produces neat 6 (or fewer) digit
quotients. Using other software to scale a full size drawing allows one to
get around that problem. But drawing with the rounding problem in mind
from the get-go is also worth some effort.

Rob Kirkham

I think Rob's problem lies elsewhere.

Let me preface this comment with the confession that I am by no means an expert on Sketchup... In fact, I've never used the software, and all I know I've learned from poking around in their help files and discussion lists, looking for answers to Rob's questions.

That said, the problem lies not in what scale the object is drawn in, or how many decimal places entered points are calculated to. If one were to draw the gusseted end of a hat section brace as a portion of an ellipse, or a NURBS curve, extrude that profile to give it some thickness, lay it over modeled sheathing with V groove edges, and preform the Boolean union operation to attempt to make them one solid, the software is going to preform all kinds of calculations to the limit of its precision, no matter if the original data was entered to three place precision, two place precision, or limited to whole numbers.

The problem is that while software operating with solids as the native environment will either do the operation, or choke right then and there. Sketchup, on the other hand, will happily trim all the surfaces and display them, without any indication that the different surfaces that define what you assume to be a solid do not have concurrent boundaries, and you won't learn that until thousands of operations later, when you try to convert the file to STL format. You end up with something that Sketchup calls "leaky solids", which is basically the software development team's admission that their product can't reliably calculate boundaries with all points within the internal margin of error for concurrent points. It is basically the difference between freeware, and software that will set you back multiple thousands of dollars.



Dennis

I wrote:

I design in 1:1 because I work from prototype drawings. When I'm satisfied with the part or assembly I ** an STL file. Also 1:1. I bring that into Netfabb and scale it, and export *that* as another STL file. (I have to use the export function with Netfabb - for some reason I can't get it to ** in that program.)

I find I get a higher quality output if I scale the STL file rather than scaling the design and then creating the STL file. (I tend to lose my curiosity when I find a system that works and will just stick with that.)

Well, that doesn't make a lot of sense! Yahoo dropped two instances of "Save as" enclosed in greater than/less than brackets. They belong where I put the double asterisks above.

Tom M.

So you would output your final object to another program for scaling? What other programs would you use?

Scott Haycock

Don't know if this will pertain using Sketchup...

I design in 1:1 because I work from prototype drawings. When I'm satisfied with the part or assembly I <Save as> an STL file. Also 1:1. I bring that into Netfabb and scale it, and export *that* as another STL file. (I have to use the export function with Netfabb - for some reason I can't get it to <Save as> in that program.)

I find I get a higher quality output if I scale the STL file rather than scaling the design and then creating the STL file. (I tend to lose my curiosity when I find a system that works and will just stick with that.)

Tom Madden

So you would output your final object to another program for scaling? What other programs would you use?


Scott Haycock
Modeling Tarheel country in the Land of Enchantm ent

yes - but if using Sketchup to scale the drawing to HO, you can also run into the 6 decimal place rounding problem if you are not careful. Dividing most large numbers by 87.1 seldom produces neat 6 (or fewer) digit quotients. Using other software to scale a full size drawing allows one to get around that problem. But drawing with the rounding problem in mind from the get-go is also worth some effort.

Rob Kirkham

--------------------------------------------------
From: "Scott H. Haycock " <shhaycock@...>
Sent: Friday, February 08, 2013 6:18 PM
To: <STMFC@...>
Subject: Re: [STMFC] Re: 3D printing challenges etc.

Rob
In Sketchup, c an't you draw in full scale then reduce the final result at the end?


Scott Haycock
Modeling Tarheel country in the Land of Enchantm ent
----- Original Message -----

Rob
In Sketchup, c an't you draw in full scale then reduce the final result at the end?


Scott Haycock
Modeling Tarheel country in the Land of Enchantm ent

very nice parts Tom! The detail you are able to draw with solid works is impressive.

Rob

--------------------------------------------------
From: "pullmanboss" <pullmanboss@...>
Sent: Friday, February 08, 2013 12:50 PM
To: <STMFC@...>
Subject: [STMFC] Re: 3D printing challenges etc.

Mick Storzek wrote:

It all goes back to the mantra I've been chanting...

I can't get no
RESOLUTION,
I can't get no
RESOLUTION,
But I try, and I try, and I try...

Until the process can reliably yield .003" steps on the surface, and reliably construct smooth vertical walls, I just don't see much use for it. I keep seeing people singing the praises, but I'm not seeing any photos of the actual models, just overall photos of frosty white blobs.

OK, here's some photos of that parts plate I mentioned yesterday. It was designed 1:1 in SolidWorks 2012 and saved as an STL file. The STL file was scaled to 1:87.1 and validated in Netfabb, a free viewer available through Shapeways, and sent to ADC manufacturing. ADC created the part using 3D Systems Projet process in 29 micron layers at 750 x 750 DPI resolution. (Shapeways calls it their FUD process.) No white blobs because I hit it with a light coat of gray primer so I could make a mold. Photographed in direct sunlight to accentuate every detail, defect and dust particle. These two photos show the whole plate from two different angles:
http://pullmanproject.com/Plate1.jpg
http://pullmanproject.com/Plate2.jpg

Here's the center portion in extreme closeup:
http://pullmanproject.com/PlateCloseup.jpg

In this photo you can see a heavy wax track below the drain valve housing on the bottom of the left unit. You can also see the general roughness of the vertical surfaces. But that sheelmetal flange running the length of the middle unit is 1/2" thick - 0.006" in HO, and perfectly defined. (It projects 1".) Likewise the rivets at the top edge of that unit are 1/2" in diameter and nicely defined. But most impressive is the lack of stairstepping on the angled faces of all three housings.

I did nothing but clean and prime that plate. I polished out the wax tracks and cleaned up the first generation castings, and those will be my production masters.

Tom Madden

Dennis has basically nailed the source of my difficulties. Better software is the obvious solution, but for now I'm getting sufficiently good results with free software (Sketchup) that I am instead working to improve my drawing skills. For example, early on I was using conversions from prototype dimensions to HO scale dimensions that used all 6 decimal points available in Sketchup. "Wow - can I draw with precision" I thought. Dumb beginner mistake. As a result, the software couldn't handle dividing a space in two equal parts as it would require the 7th decimal place. Seems self-evident now, but I didn't catch that in my early on line learning efforts. Given the resolution of the printed product, nothing was being achieved with 6 decimal places that couldn't usually be achieved in 3 (and sometimes 2) places.

Rob Kirkham

--------------------------------------------------
From: "soolinehistory" <destorzek@...>
Sent: Friday, February 08, 2013 8:46 AM
To: <STMFC@...>
Subject: [STMFC] Re: 3D printing challenges etc.

--- In STMFC@..., Richard Brennan wrote:

At 09:57 PM 2/7/2013, Rob Kirkham wrote:

That too is too subtle with 3d printing.

So I've drawn and re-drawn it - and each time come to a leaky dead end.
When I get the interest back, I'm going to start from scratch and do it again.

I suspect the issue is NOT your drawing...
but the specific 3D print process/technology you are using.

Question - Were you expecting finished models to be:
1st generation: 3D print each car + paint + lettering?
2nd generation: 3D print master car car + resin-cast copies + paint +
lettering?

If the latter... then the need for 3D print quality trumps the desire
for a low cost print,
and you can find a print service that will do the resolution you
need. They do exist!

Just a thought...

--------------------
Richard Brennan - San Leandro CA

It all goes back to the mantra I've been chanting...

I can't get no
RESOLUTION,
I can't get no
RESOLUTION,
But I try, and I try, and I try...

Until the process can reliably yield .003" steps on the surface, and reliably construct smooth vertical walls, I just don't see much use for it. I keep seeing people singing the praises, but I'm not seeing any photos of the actual models, just overall photos of frosty white blobs. I'm sure one of the various additive processes will eventually yield good, usable parts... but whether we will be able to afford them is another issue.

Rob's file problems seem to trace back to the fact that Sketchup is not a solid modeler, it's a SURFACE modeler, that defines a solid by its boundary surfaces, It will therefore draw surfaces that appear to enclose a volume, but really don't. Since the STL files used by most of the "3-D printers" are extracted from solids, if the Sketchup surfaces don't actually enclose the volume, then the solid isn't there, and won't be converted. This isn't a slight of Rob's CAD abilities, the problems most likely derive from rounding errors in Sketchup's calculations as it attempts to trim the various surfaces to each other. For the web animation purposes Sketchup seems to be aimed at, these errors are insignificant; not so for our purpose. I suspect that a step up to industrial grade software would solve the problem, but not cheaply.

Here is a link to a web discussion that has several links to software plug-ins that help identify problem areas in a Sketchup model:

http://sketchucation.com/forums/viewtopic.php?f=79&p=276399

Of course, finding the problem is one thing, fixing it is a whole 'nother issue.

Dennis




------------------------------------

Yahoo! Groups Links

--- In STMFC@..., "pullmanboss" wrote:

Dennis asked:

Cool, Tom. Care to share some "fer instance" pricing, to compare them to Shapeways?

The part from Shapeways was $53.60 plus shipping. From ADC it was $120 including shipping. Probably not justifiable if you were making a part that size to use as-is, but cheap for a good pattern. (I love doing the 3D CAD work and, as with model building, place no value on the time spent.)

Tom Madden

Thanks Tom. Just a reality check on those who say that new technologies will make everything else obsolete.

There is certainly something usable here, but not the revolution that will make everything that came before obsolete. As the "masses" (those who haven't a clue as to what they are talking about) flock to the "better" service bureaus, look for the fees to increase, not decrease, due to the amount of hand holding time required.

Dennus

Dennis asked:

Cool, Tom. Care to share some "fer instance" pricing, to compare them to Shapeways?

The part from Shapeways was $53.60 plus shipping. From ADC it was $120 including shipping. Probably not justifiable if you were making a part that size to use as-is, but cheap for a good pattern. (I love doing the 3D CAD work and, as with model building, place no value on the time spent.)

Tom Madden

At 01:34 PM 2/8/2013, soolinehistory wrote:

So, the part is being built with little balls of material just slightly larger than one thousandth of an inch in diameter. And, I suspect they are balls, not completely circular, but not sharp cornered, either. That would account for lack of stair stepping on mid-range angular surfaces, say, 30 to 60 deg. from horizontal. The rounded edges of the balls conform to the angle better than sharp corners would.

I'm not sure the material is even "solid" when it is being laid down...
In some technologies, the build is from a pool of liquid, and only the parts a laser is focused on will fuse into a solid. AFAIK... These are limited by the size/intensity of the laser beam. For example, look at http://formlabs.com/pages/tech-specs

Unfortunately, Shapeways is promoting the "gee-whiz" aspect of 3D modeling, rather than thoughtful use of the process, so rather than better models, we are going to see more and more stuff like this:
http://www.shapeways.com/model/711147/40-stockcar-body.html?li=productBox-search

Garbage-In = Garbage-Out.
True in every technology we use, -especially- in some of our favorite CNC-cut die injection-molded plastic foobies!
(...where the cost to correct is many times higher)


--------------------
Richard Brennan - San Leandro CA
--------------------

--- In STMFC@..., "pullmanboss" wrote:

OK, here's some photos of that parts plate I mentioned yesterday. It was designed 1:1 in SolidWorks 2012 and saved as an STL file. The STL file was scaled to 1:87.1 and validated in Netfabb, a free viewer available through Shapeways, and sent to ADC manufacturing. ADC created the part using 3D Systems Projet process in 29 micron layers at 750 x 750 DPI resolution. (Shapeways calls it their FUD process.) No white blobs because I hit it with a light coat of gray primer so I could make a mold. Photographed in direct sunlight to accentuate every detail, defect and dust particle. These two photos show the whole plate from two different angles:
http://pullmanproject.com/Plate1.jpg
http://pullmanproject.com/Plate2.jpg

Here's the center portion in extreme closeup:
http://pullmanproject.com/PlateCloseup.jpg

In this photo you can see a heavy wax track below the drain valve housing on the bottom of the left unit. You can also see the general roughness of the vertical surfaces. But that sheelmetal flange running the length of the middle unit is 1/2" thick - 0.006" in HO, and perfectly defined. (It projects 1".) Likewise the rivets at the top edge of that unit are 1/2" in diameter and nicely defined. But most impressive is the lack of stairstepping on the angled faces of all three housings.

I did nothing but clean and prime that plate. I polished out the wax tracks and cleaned up the first generation castings, and those will be my production masters.

Tom Madden

Cool, Tom. Care to share some "fer instance" pricing, to compare them to Shapeways?

29 microns is .0011", which is what I've figured will be needed to make the stair steps no more than surface noise. 750 DPI is .0013". So, the part is being built with little balls of material just slightly larger than one thousandth of an inch in diameter. And, I suspect they are balls, not completely circular, but not sharp cornered, either. That would account for lack of stair stepping on mid-range angular surfaces, say, 30 to 60 deg. from horizontal. The rounded edges of the balls conform to the angle better than sharp corners would.

It's interesting that the machine is from 3-D Systems... that's the outfit that gobbled up the 3M/Keltool sintered metal process a number of years ago. Using that process Tom's patterns could be turned into copper-tungston EDM electrodes, and used to sink injection mold cavities. For an example, Accurail's "Bettendorf" and roller bearing truck frames were done using electrodes made by that process, back when 3M would work with hand crafted patterns.

The wax tracking for support of overhangs is still a problem, unless you can design around it. This is going to be a problem with parts like that tank car center sill, it's all overhangs and is going to have the tracks all over the finished part. It would have been better if it had been done as multiple parts; each "show" surface should have been done as a separate part, so it could have the best orientation.

Unfortunately, Shapeways is promoting the "gee-whiz" aspect of 3D modeling, rather than thoughtful use of the process, so rather than better models, we are going to see more and more stuff like this:

http://www.shapeways.com/model/711147/40-stockcar-body.html?li=productBox-search

Dennis

Mick Storzek wrote:

It all goes back to the mantra I've been chanting...

I can't get no
RESOLUTION,
I can't get no
RESOLUTION,
But I try, and I try, and I try...

Until the process can reliably yield .003" steps on the surface, and reliably construct smooth vertical walls, I just don't see much use for it. I keep seeing people singing the praises, but I'm not seeing any photos of the actual models, just overall photos of frosty white blobs.

OK, here's some photos of that parts plate I mentioned yesterday. It was designed 1:1 in SolidWorks 2012 and saved as an STL file. The STL file was scaled to 1:87.1 and validated in Netfabb, a free viewer available through Shapeways, and sent to ADC manufacturing. ADC created the part using 3D Systems Projet process in 29 micron layers at 750 x 750 DPI resolution. (Shapeways calls it their FUD process.) No white blobs because I hit it with a light coat of gray primer so I could make a mold. Photographed in direct sunlight to accentuate every detail, defect and dust particle. These two photos show the whole plate from two different angles:
http://pullmanproject.com/Plate1.jpg
http://pullmanproject.com/Plate2.jpg

Here's the center portion in extreme closeup:
http://pullmanproject.com/PlateCloseup.jpg

In this photo you can see a heavy wax track below the drain valve housing on the bottom of the left unit. You can also see the general roughness of the vertical surfaces. But that sheelmetal flange running the length of the middle unit is 1/2" thick - 0.006" in HO, and perfectly defined. (It projects 1".) Likewise the rivets at the top edge of that unit are 1/2" in diameter and nicely defined. But most impressive is the lack of stairstepping on the angled faces of all three housings.

I did nothing but clean and prime that plate. I polished out the wax tracks and cleaned up the first generation castings, and those will be my production masters.

Tom Madden

--- In STMFC@..., Richard Brennan wrote:

At 09:57 PM 2/7/2013, Rob Kirkham wrote:

That too is too subtle with 3d printing.

So I've drawn and re-drawn it - and each time come to a leaky dead end.
When I get the interest back, I'm going to start from scratch and do it again.

I suspect the issue is NOT your drawing...
but the specific 3D print process/technology you are using.

Question - Were you expecting finished models to be:
1st generation: 3D print each car + paint + lettering?
2nd generation: 3D print master car car + resin-cast copies + paint +
lettering?

If the latter... then the need for 3D print quality trumps the desire
for a low cost print,
and you can find a print service that will do the resolution you
need. They do exist!

Just a thought...

--------------------
Richard Brennan - San Leandro CA

It all goes back to the mantra I've been chanting...

I can't get no
RESOLUTION,
I can't get no
RESOLUTION,
But I try, and I try, and I try...

Until the process can reliably yield .003" steps on the surface, and reliably construct smooth vertical walls, I just don't see much use for it. I keep seeing people singing the praises, but I'm not seeing any photos of the actual models, just overall photos of frosty white blobs. I'm sure one of the various additive processes will eventually yield good, usable parts... but whether we will be able to afford them is another issue.

Rob's file problems seem to trace back to the fact that Sketchup is not a solid modeler, it's a SURFACE modeler, that defines a solid by its boundary surfaces, It will therefore draw surfaces that appear to enclose a volume, but really don't. Since the STL files used by most of the "3-D printers" are extracted from solids, if the Sketchup surfaces don't actually enclose the volume, then the solid isn't there, and won't be converted. This isn't a slight of Rob's CAD abilities, the problems most likely derive from rounding errors in Sketchup's calculations as it attempts to trim the various surfaces to each other. For the web animation purposes Sketchup seems to be aimed at, these errors are insignificant; not so for our purpose. I suspect that a step up to industrial grade software would solve the problem, but not cheaply.

Here is a link to a web discussion that has several links to software plug-ins that help identify problem areas in a Sketchup model:

http://sketchucation.com/forums/viewtopic.php?f=79&p=276399

Of course, finding the problem is one thing, fixing it is a whole 'nother issue.

Dennis