modeling resolution

[liquidguitars]

I reading it now so lets say I have a 24x24 job size and I place a 8x8 stl in the center I set the project to high this will give me a flawed carving? can you show me this as a JPG?

[Rocketman]

Read the tip I posted in #161. You'll be fine.

EDIT: Thinking about the violin design...there are no small details on that model and if you boost your resolution in your layout and place your models in an area just big enough to fit the models (and allow for your hold-down screws or clamps), you shouldn't experience any resolution issues to worry about. I think you probably sand the heck out of the violins after they come off the machine anyway and your vector cutout profiles around the outer perimeter are "pure"...there are no resolution concerns with vectors at all. If you want me to help you with this, please send me the model and I can lay it out for a great result. It will be my pleasure to do this for you, if you want me too.

What your saying in post 161; that's exactly what most nesting software does anyway. The file sizes would be huge if they didn't.

What really shocks me is your telling us many of the big players take perfectly good STL files and screw them up by converting them to raster. There are several ways you could do that and still keep the resolution high but your telling us that's not the case? I guess it all boils down to keeping files small so programs don't crash trying to crunch the numbers. Or lazy programmers might be another reason. From what your saying I don't think it would even matter how complex the original STL file is, it's going to get crunched anyway. I can have two STL files of a 1" cube for example and one could have ten times the triangulation as the other. In most software I play with the one with more triangles would have a higher resolution and you would not notice the effect unless you enlarged it to a great extent. Some of my cam programs do have a harder time loading STL files with a high triangle count and I have to lower resolution to import them. Sometimes that more of a limitation with my computer rather than the cam program I'm running. Sometimes it's a graphics card limit more than anything as a lot of those processes are handed off to the graphics cards these days. Also it may be a problem with either Open GL or Direct X depending on which platform the program designer decided to use. Lots of variables but like most programmers one thing remains the same; the saying among programmers used to be "garbage in = garbage out!"

Brad

[mtylerfl]

Hi Brad,

You are stating many of the concerns/questions I've heard countless times before. People are perplexed why pixel-based CAM software behaves in this way. Unfortunately, this is the norm. I wish it were not! We, as designers, must be aware of the resolution limitations of pixel-based software to assure we achieve the best result possible. I hope that someday, ALL CAM software can somehow be resolution independent, but this is not the case today.

As long as you have at least a reasonable understanding of the resolution tips put forth previously, it is unlikely to cause you any major problems. Here is yet another tip if you are creating models from scratch yourself...when you make the model, make it twice the size you anticipate using it in an actual job layout. This way you can "pack it" with more data points from the get-go. The PDF link posted previously explains this in more detail. I often follow this procedure (but not always, even though I probably should).

Just an aside, after I create a model and export it as STL, I choose the triangulation tolerance setting to 0.0001". As you can imagine, this can result in a very large file size. But, I often use the free MeshLab software to remesh at 67% or sometimes 50% reduction. This does a good job of reducing the file size without too great a hit on the detail (usually no noticeable hit at all). Having said that, I usually use the original STL (the larger file) when I import the STL into the CarveWright STL importer for conversion to a PTN file. It just depends on whether a file is going to be "too large" for the importer to handle or not. I have found a file size larger than about 99 MB can sometimes crash the importer, so if I run into that, I'll use the remeshed file to import for conversion instead.

I hope this is helpful!!

[liquidguitars]

when you make the model, make it twice the size you anticipate using it in an actual job layout. This way you can "pack it" with more data points from the get-go.

Hi MT are you saying that I should scale the 3D model 2x bigger" not talking about tessellation" and then scale the raster down 50% in Aspire? This sound like a good thing to try.

[mtylerfl]

I reading it now so lets say I have a 24x24 job size and I place a 8x8 stl in the center I set the project to high this will give me a flawed carving? can you show me this as a JPG?

I just saw your post! You can easily do this yourself (if you have access to the Aspire software), but I will create some examples for you at various resolution quality settings.

[mtylerfl]

Hi MT are you saying that I should scale the 3D model 2x bigger" not talking about tessellation" and then scale the raster down 50% in Aspire? This sound like a good thing to try.

If you have the original model and software it was created in (are you using LightWave?), it might have some benefit as described in the PDF. But, the PDF is actually making the recommendation if you are creating the model in a pixel-based modeling program to begin with, I believe. I don't think you are using pixel-based modeling programs to create your models. But you are right...this would be a good thing to try.

[liquidguitars]

Yes Lightwave...yes we are OT.

[fwharris]

Attention FW...may I suggest moving the 'resolution portion' of this discussion to a separate thread, if possible?...maybe entitled "modeling resolution" in the third party software section. I'm still feeling badly about the hijack I caused.

I think I did it right!!

[mtylerfl]

FW, thank you for moving the thread to here!

[liquidguitars]

Thanks Floyd!