Huge STL files in X7

[RON S]

This may have been covered allready but you can't seach for STL on the forum because its to short.

Is there any way to save an STL from a verify in X7 that is not such a huge file?

 

We have been having trouble using the STL's from X7 verify so I did a little test on a simple part. The first picture is from X6 and is 798 KB using .001 tolerance. The second picture is form X7 and is 36,619 KB using .002 tolerance. Thats over 45X larger with twice the tolerance.

 

 

 

We mainly use the STL's for our Matsuura MAM72-63V 5 axis 3 machine cell and have every op in a differnt file to make things cleaner and also our simulation/post processor (CAMplete) is built off C/L of rotation and ignors any WCS other than Top so we can't use multiple WSC's to program a muilti op 5 axis part complete if we wanted to so we save a STL at the end of the op and stl Xform to the new location for the next op. All this worked fine in all previous version of Mastercam but now X7 seams to have changed the way it saves STL's making them almost totally useless. I was able to import STL files into our simulation software for reference but know it just crashes because the files are too large.

 

If any one knows of a solution I would love to hear it.

 

Thanks,

Ron

 

Intel Core [email protected]

Win 7 Pro 64 bit

32 GB RAM

NVIDIA Quadro K4000

(3) 27" Asus monitors

[rogkick]

Try changing the resolution even more, does it require .001 or .002 tolerance?

I did notice they were large myself.

[Chris Rizzo]

Other than cranking the tolerance way up,

 

This program click-to-fit has an awsome stl reducing function.

 

(yes Buck the icon was a bug )

[GIZMO 21]

All this worked fine in all previous version of Mastercam but now X7 seams to have changed the way it saves STL's making them almost totally useless.

 

I haven't loaded x7 in my box yet,but i wonder what else is broke that was working before.

[Mike Whitten CAD/CAM Contractors]

Hello Ron,

 

Please submit this to QC. I have this issue myself and also submitted it. I don't think they will correct this issue anytime soon unless they get more feedback as I believe that there are only a few of us who require STL's. I consider this a bug/severe limitation since it is totally unusable like it is. Changing the tolerance will help, but not enough to fix it. Something is wrong with the engine. As you mentioned, this is a first, all other past versions have been fine.

 

Please send them a link to this thread also. It may be good to wait a few days to see if anyone else responds to this thread.

 

Thanks

[djstedman]

The only way I have been able to use STL's from X7 at all is to open them in meshlab and simplify them as much as possible..

 

I totally concur that this is a severe step backwards from X6 and previous versions.. I have managed to work around it thankfully but there is certainly something very wrong with how STL's are generated in X7 ..

[Cannon]

I haven't loaded x7 in my box yet,but i wonder what else is broke that was working before. [/color]]

X7 sp2 is nearly unusable to me right now because of the verify. It opens painfully slow compared to old verify, the quality of verified part is a joke and trying to save an STL from the verify results is hopeless. I.ve had more crashes this week with X7 verify than what i´ve had last couple of years with previous versions. The only thing that keeps me using X7 right now is the much improved processing speed of toolpaths.

[Cannon]

We machine 3d shapes to large casting so saving STL from verify is very important. I havent tested the stock model in X7 yet because of previous bad experieces but I think I have to try it atleats..

[Tyler Robertson]

ROn - can you use the stock model for this operation instead of saving STL files?

[Tyler Robertson]

We machine 3d shapes to large casting so saving STL from verify is very important. I havent tested the stock model in X7 yet because of previous bad experieces but I think I have to try it atleats..

 

stock model has been great for me in X7, programming from huge STL files

[Tyler Robertson]

X7 sp2 is nearly unusable to me right now because of the verify. It opens painfully slow compared to old verify, the quality of verified part is a joke and trying to save an STL from the verify results is hopeless. I.ve had more crashes this week with X7 verify than what i´ve had last couple of years with previous versions. The only thing that keeps me using X7 right now is the much improved processing speed of toolpaths.

 

there's an XML file called MastercamSimulatorDefaults.xml

 

if you open that in notepad, there's a section <PrecisionFactor>X</PrecisionFactor>

 

I have mine set to 2.5 and things look great now. Why this isn't something more visible that you can change in the interface is beyond me, but hope it helps

[CamMan1]

ROn - can you use the stock model for this operation instead of saving STL files?

 

Here we are again talking about stock models for verification. This has been brought up before. They do work okay for verification as long as you don't have very many stock models in your ops manager. If you are at all human we make mistakes and if you have to change something 5 stock models back now you have to sit and wait to regen all 5 if you have done them sequentially. Don't get me wrong, stock models have there place especially for rest roughing and such but they are not effecient enough for verification. I would like to have my useable STL's back.

[Tyler Robertson]

That's a good point - I use them almost exclusively for rest roughing (which I use almost all the time), and every other toolpath I use for 3-axis stuff uses multithreading so I don't mind too much if I have to regen everything

 

If you have to change something 5-stock models back don't you have to then save out a new STL regardless after you re-verify? and while re-verifying you couldn't really be doing much else

[Mike Whitten CAD/CAM Contractors]

I agree, stock models aren't fully supported due to size etc. If you have a very complex part and must have many stock models for finish operations, they can not reside in the file. So...We need/should be able to have a normal STL export function.

 

I have been able to get by so far, but I think that's because I haven't programmed a very large and nasty part yet. I have done parts in X6 and prior that I don't think that X7 could do without screwing around. Also, this really starts to use up space on networks.

 

edit:/ I use stock models mainly for rest roughing. The problem is high resolution finishing.

[Tyler Robertson]

I get what you guys are saying

[RON S]

Wow this turned into a hot topic, I was begining to think I was the only one using STL's.

 

Stock model is ok for simpler parts but I had problems in X6 with it leaving operations out of the model so I went back to STL's even for simple parts. I have not used it a lot in X7 yet so I don't know if they fixed it but I don't believe there is any way to save a stock model and be able to shift it and save it to use in another operation which we need to do for our 5 axis machines. I have programmed several complex 5 axis parts in X7 but it's definetly a struggle. I have played with the verify xml trying to find a happy medium but its either very ugly, very slow or ugly and slow. I haven't tried it but I am guesing I would need to use a .020 STL tolerance in X7 to get the same size file X6 saves with a .001 tolerance, but that model would not be good for much. The only thing that makes it even usable is having the verify in a seperate window. I even talked the boss in to buying a 3rd monitor so I can leave veify on all the time on its own monitor. This way you can do something else while its loading and running.

 

I tried the Click to fix on the 5 axis part I am working on but it crashed several times because the file was to large. I will have to try the Meshlab but it sure would be nice if Mastercam would save an STL that you could use with out having to run it through another program.

 

I will try and submit to QC. Maybe they will fix it in X10.

[Chris Rizzo]

Hi Ron, check your messages here. I just sent you something.

[Colin Gilchrist]

Hi Everyone,

 

I'm trying to get some attention focused on this issue. Please send in examples to [email protected], and let them know that this is a big deal to you. (Also, please just reference the STL issue, no need to quote my name when sending in responses to QC). The more emails we get, the higher priority this issue becomes.

 

Thanks,

 

Colin

[Cannon]

Hi Everyone,

 

I'm trying to get some attention focused on this issue. Please send in examples to [email protected], and let them know that this is a big deal to you. (Also, please just reference the STL issue, no need to quote my name when sending in responses to QC). The more emails we get, the higher priority this issue becomes.

 

Thanks,

 

Colin

 

There is clearly some serious problems when every Stl file that you save from verify is 10x or more bigger that in X6. We just renewed maintenance but are left with unusable software. Will there be any improvements about this in X7 MU1?

[Aaron Eberhard]

Ron (or anyone else having large STL file issues) - Can you please post up one of the files you're having issues with?

 

Just a little background to anyone who didn't know this about X6's verify: The displayed model on the screen (and resultant STL) aren't necessarily water-tight, but it was lighter weight and able to be displayed nicer. When you saved an STL, it also created an LWN file that was water tight and 100% accurate to its' calculation. I remember talking to Ron Branch about some issues using the STL instead of an LWN on a project I discussed with him one time, but I digress..

 

Either way, X7's verify solution is 100% water-tight, with no reduced visibility/functionality models created.

 

I just did a test over here on an X6 sample file, I used 3D_HST_HYBRID_FINISHING, which I believe was the benchmark file. I ran the file with X7 Mu1 under both Inch and Metric to make sure it wasn't a conversion bug of some sort.

 

As you can see, with comparable quality (middle vs. max on the sliders), there's not too much of a difference over here, the worst is the metric one with a 46 Mb to 58 Mb difference..

 

Another thing I did was create a stock model with a .002 finish in inches, which weighed in at 18 Mb when saved as an STL, putting it a touch above the medium precision slider (which is expected).. Note that the SM generated in 30 seconds, as opposed to 2.5 minutes on "medium" quality X6's verify..

 

The metric equivalent went to 15Mb set to .05mm, compared to the medium STL of 16, that's pretty close. Setting it to .025mm, increased my SM processing time from 27 seconds to 35.29, with a save size of 18.6 Mb, so still not too bad.

 

I also used this opportunity to grab a screen shot of the same area and compare the visual quality, which I've attached here:

 

Cheers,

 

 

Edit: I should add that I had a co-worker test a few of these on an X7 SP2 machine and his results matched mine.

[?Mark]

As you can see, with comparable quality (middle vs. max on the sliders), there's not too much of a difference over here, the worst is the metric one with a 46 Mb to 58 Mb difference..

 

I was under the impression that this slider was for visual reference only and had nothing to do with accuracy of the stl (slider at the bottom left of verify). What exactly is "medium"? Not a very technical way of describing anything. What is "save stl tolarance" than?

 

Yes, the files are 2-4 times bigger than before...I'm hitting 300mb daily....(

[Colin Gilchrist]

Something that is really important to note is what Aaron mentioned about the STL files produced from X6. These models are not water-tight. This means you would often have gaps or other problems with the STL model being produced, prior to X7. That means if you take a STL file from X6 and use it in a "Rest" operation, there is a chance the toolpath calculation would fail, due to the STL model quality, even though it might "visually" look fine when inspecting it or bringing it back into Verify for another session. In X6 and prior versions, saving the STL model from Verify (not water-tight) and then bringing it back into Verify as stock for different operations could also cause Verify to crash. Again, due to the models not being water-tight.

 

We could give users the option to create a much smaller STL file from the Verify in X7, but you would run the risk of having a non water-tight STL file that might cause you problems. Do you want that option? Would a smaller model that carries some potentially bad side-effects be of use to you?

[RON S]

I am not sure what the file you are using looks like but the sample I posted in the original post was just a fairly simple box with some holes. X6 STL file size was 798K and the X6 LWN was 805K with .001 tolerance. The X7 STL file was 36,619K with a .002 STL tolerance which is over 45 times larger. If you are testing a completely surfaced part they will probably be closer because even X6 has tons of triangles to display a surfaced part. If you test a simple part with large flat faces I believe you will get a larger difference because X6 would only use a few triangles to display a flat face while X7 will use hundreds of triangles to display a flat face.

[Colin Gilchrist]

I was under the impression that this slider was for visual reference only and had nothing to do with accuracy of the stl (slider at the bottom left of verify). What exactly is "medium"? Not a very technical way of describing anything. What is "save stl tolarance" than?

 

Yes, the files are 2-4 times bigger than before...I'm hitting 300mb daily....(

 

Hi Mark,

 

Again, the STL files produced in X6 were not water-tight. So yes, in X7 you might have a 300 Mb STL vs. a 75 Mb STL from X6, but that X6 STL File is almost certain to have some errors. These errors may or may not cause problems with both "Rest" style paths, and trying to run further sessions of Verify.

 

If you look in your old directories for STL models from prior versions of X7, you'll sometimes see a ".lwn" file. These are the native files produced from X6 verify, before the STL conversion.

[Colin Gilchrist]

I am not sure what the file you are using looks like but the sample I posted in the original post was just a fairly simple box with some holes. X6 STL file size was 798K and the X6 LWN was 805K with .001 tolerance. The X7 STL file was 36,619K with a .002 STL tolerance which is over 45 times larger. If you are testing a completely surfaced part they will probably be closer because even X6 has tons of triangles to display a surfaced part. If you test a simple part with large flat faces I believe you will get a larger difference because X6 would only use a few triangles to display a flat face while X7 will use hundreds of triangles to display a flat face.

 

Hi Ron,

 

I sent several people links to this thread, so we are aware of the issue and looking into it. Thanks for bringing it up. We are working on a solution.