Almost beyond the system limit...

MBartlett21

I did this model that was almost at the system limit.
If I made it any bigger, it didn't work.
The arcs and circles are all showing up wrong.

Why would this be happening?

Jake_Rosenfeld

@mbartlett21

Wow! One of our active priorities now (and pretty much always) is improving the fidelity of our tessellation.  Your example looks like it could be a really good test case for us.  Please make a version on your document that shows the issue and share it with us using the "Feedback" button under the "?" menu.

Sorry you're experiencing this.

MBartlett21

@Jake_Rosenfeld

Here is the document: https://cad.onshape.com/documents/6320c1fa97928fa83a4cdaf5/v/de570ed25c578ffa4dc457b2/
I have done the grey part on the right corner.

For some reason, the report a bug function isn't working. Does it need a website whitelisted??

Jake_Rosenfeld

@john_rousseau or @NeilCooke may know the answer to that.  What's the behavior you're seeing when you try to report the bug?

MBartlett21

@Jake_Rosenfeld
When I try to report a bug, nothing happens, even when I click the button again.
I do have an internet filter on, would that be stopping it working?

NeilCooke

If you have adblock on please disable it, then try again please

NeilCooke

@mbartlett21 - I took a look at the model - any reason you're trying to make a sheet metal part that is more than 500m wide? That is the Parasolid size limit for solid bodies so I would expect some errors if you are close to or over it.

michael_mcclain

NeilCooke said:

@mbartlett21 - I took a look at the model - any reason you're trying to make a sheet metal part that is more than 500m wide? That is the Parasolid size limit for solid bodies so I would expect some errors if you are close to or over it.

That's going to need one hell of a press break

Jake_Rosenfeld

The issue actually has nothing to do with the large part, it has to do with the distance of the small part from the origin.  The size of the world-space coordinates of the triangles of the tessellated part are large enough that its causing floating point resolution issues.

There are only so many numbers that can be represented in floating point format.  When your number is close to 0, the resolution from one floating point number to the next is very very very fine.  When you're out in the boonies of the real-number line there may be noticeable gaps from one number to the next.  So, the tessellation may be inexact in the first place (because of the inability to exactly represent the number we want) and then the issue compounds through a series of mathematical operations on your graphics card.  This is why the lines of the model seems to jump around; as you rotate the part you're seeing hops from one to the next discrete solution of a floating point operation.

In any case, your part looks fine if its translated closer to the origin.  Test cases are always welcome though! Submit that ticket if you got the feedback tool working, otherwise let us know if that's still being stubborn.

MBartlett21

@NeilCooke
@Jake_Rosenfeld
I was just having fun.
I also made a feature in that document which adds bend allowance to the sheet metal model.
I would find this very useful if you could specify the bend allowance per bend

MBartlett21

NeilCooke said:

If you have adblock on please disable it, then try again please

I don't have an adblocker on, but it still doesn't work.