Three Point Circle does not intersect the points selected to create it.

[drew]

I'm experiencing an odd behaviour of the three-point circle in a sketch.

Three points are chosen (at maximum zoom), yet the circle that results does not intersect the three points chosen, it is created just 'inside' the points.

Please explain if this is intended behaviour.

If not an intended behaviour, please fix it.

More detail for understanding is in the screencast.

[masa.minohara]

Hi @drew,

Thank you for posting about the issue. I was able to reproduce something similar and I think this is a bug. If you use Measure and select the point and 3-point-circle, what distance do you get? 

There is clearly a gap between these but I got 0mm. Are are you seeing the same? 

If not, could you share a public link to the design with me so I could take a closer look?

[drew]

Hi,

Hmm. Not sure if I should be proud or bummed out to have found a bug. Beginner's luck?

I can measure .043 between the circle and the arc up at the top of the design.

I get 0.00 when I measure from one of the points used to create the circle to the circle itself.

I get that .043 measurement when I measure at the _largest visible gap_ and when I measure at the _smallest visible gap_ (the outer arc is faceted, it's not a true arc).

I'm happy to share a link directly with you, but the artwork is copyrighted (and I'm on thin ice posting the screencast as it is) - so I do not wish to post that link here in the open for anyone to access.

Sent the link in a PM, please look for it.

[drew]

Also - Time zones may make this sporadic. I'm GMT/UTC + 12:00, and have no clue where you are. Have just been to a movie with the family and come back to find your reply here.

So just be aware that my replies will be made when not tending to family, working, or sleeping... and may have a time lag... 

[drew]

A new development:

It is not just the three-point circle.

Please see new screencast showing that if you create a new center point circle off to the side and attempt to make that circle coincident with the two points used previously, it still has an offset fro the points and does not close the profile.

[drew]

Edited/deleted repost of screencast

[drew]

Another development:

Started just messing around with it to see what happened, and discovered that you cannot make a point visually coincident with the outer curve in this design. Ends up with the same odd offset as the circle, and an unmeasurable but visible distance. New screencast shows phenomenon.

[TrippyLighting]

Can you share your design ?

Was that sketch imported from a DXF or SVG ?

[drew]

Sent you both a shared link to the design yesterday - in that PM.

Does the link not work?

______________

And yes, the art has been imported. Graphic designer supplied a DWG file which I had trouble opening. Think it was from Adobe Illustrator but am not certain.

I posted about my difficulties importing it into Fusion 360 here:

https://forums.autodesk.com/t5/fusion-360-support/why-this-dwg-file-not-uploading-and-causing/td-p/8...

@jhackney1972 was very kind to investigate what was up with the file, and to import it into AutoCAD for a fix, then posting his own export from that application. So it is his 'fixed' f3d file that has been imported into my posted design.

[drew]

Ah to heck with it- here's the link to the design:

https://a360.co/2Oq9p1g

[TrippyLighting]

I've had a look at this and I am not experiencing the snap problems you show in you screencasts. I can snap the 3rd point to "something".

I usually have snapping and the sketch grid turned off as I find both annoying.

As to the precision issues you are seeing I believe the geometry that is created is actually OK but there is definitely a viewport display glitch at work here. When you are snapping it "looks like" it is not snapping to the accrual geometry when it actually does. It just isn't displayed that way.

This should definitely be fixed as it's definitely not the first time it's caused confusion!

[drew]

I'll reiterate to make sure I understand:

The REAL line is INVISIBLE.

The VISIBLE line is NOT REALLY THERE.

You can snap to the INVISIBLE line.

You cannot snap to the VISIBLE line.

(OMFG, like it's not hard enough to learn how to use CAD as it is!)

Correct?

[TrippyLighting]

Yes, but only if you zoom in to about max level. What sort of precision you think you are loosing at that zoom level ?

Should this display issue be improved. Probably.

Does it have any detrimental effect on your design ? Probably not!

Also tying to improve the curvature on the existing circle by creating a new one does not really achieve anything.

The shape extrudes form and filleting the ages is no problem.

[drew]

1) "Does it have any detrimental effect on your design ? Probably not!"

Well, if I cannot determine if things are connected via the on-screen representation, then I can never really be sure that any two things I think are connected are indeed connected.

I understand that given the current circumstances I am meant to 'trust' in the constraint symbols that appear nearby to feel comfortable with the asserted connection when it isn't visible, but this is not an easy thing for a person like me who SEES that they are NOT connected... 

So it affects my designs in that I will henceforth be quadruple-quintuple-bazziliantuple-checking everything and spending a lot of time to verify that things behave as if they are connected - so my 'design process', already a snail's pace, will slow even more.

2) "What sort of precision you think you are loosing at that zoom level ?"

I am losing the precision of my understanding. That .043mm distance (line to line, visually) is irrelevant to my design.

The 'big deal' is that I now cannot trust what I see.

How far I'm zoomed is irrelevant, since the onscreen line won't actually hit the onscreen point at any zoom level, it's just that you can't tell anymore from a 'distance'.

Knowing it's a screen-rendering thing doesn't actually help to alleviate the uncertainty that the flaw introduces in my head. EG of thought process: "If I can't trust it to show me connected lines can I really trust it to do anything?"

3) "Also tying to improve the curvature on the existing circle by creating a new one does not really achieve anything."

Use the curvature comb tool on the original arc, then use it on the circle.

There is a clear difference. The original curve is NOT a semicircle (if I can trust the comb, that is).

The end product will be reproducing that outer circle at a 340mm diameter, and that weird wobble in the original will be evident at that scale.

Just for info/context:

The internal lines have all been through *heavy* modification prior to my discovering this issue. That work was done to 'fair' all the curves for the same 'scaling up' reason, adding lots of tangency constraints, adding new splines to replace unusable existing curves, and more - the thing has to be very smooth-curved for the inlay to look nice.

I never encountered this 'not visibly connected' issue with any of those lines, and I've also never run into this on other Fusion stuff I've done (only a few things ever though so not like I've done a ton). Maybe it was happening all along and I never noticed... but it's still pretty messed-up that it's so.

Hope they take my commentary onboard and get to work fixing this.

Making connected lines look connected seems like a pretty fundamental thing for a program that you draw with, and having it be messed up like this aught to raise red flags!

No further replies needed, BTW - nothing more to say about it other than the above.

[bndrcr82a08e349g]

Hi, I'm having the same issue, 3 point circle  does not match tre selected points

[TrippyLighting]

If you're looking for help, you'll need to provide more detail, preferably by attaching your design to a post.

[bndrcr82a08e349g]

I need to construct a circle parallel to that arc segment, distance 0.2mm larger. I tried both drawing the circumference through three points and constructing a generic circumference and adapting it using snaps, the result is always the same, the shown circumference does not intersect the chosen points, however even during the construction phase it seems that the mouse pointer remains detached from the circumference

[TrippyLighting]

Please attach your design in .f3d format to your last post.

[bndrcr82a08e349g]

Here

[TrippyLighting]

Please see the video (once it is converted) and provide feedback if that's what you need. Looks trivial to me.