Sketchup is Inacurrate???

Wo3Dan

@wo3dan said:

Or (same thing) when loading a file saved with precision with more digits?

Did this - set my model to maximum decimal digits, closed and reopened the model.

What template can I try?

The one that you (or better, SU) currently use(s). What if you re-new that template, but this time with precision set to max. Be sure to make it the default one. What happens then.
For me both methods (loading file including its saved precision / start with correct template) work for me in SU7. I still have to try in SU8, but I'm almost convinced that I have seen correct display of arc lengths in SU8 in the past.

jeff hammond

just chiming in to say that, on mac, there's no tricky steps to take to get those boxes to update.. change the units in model info and they update accordingly in entity info.. as you'd expect should happen.. so yeah, there's something wrong on the windows side of things pertaining to this.

and i fear, out of all of this discussion, this will be the only thing that gets fixed

driven

I support improvements for arcs and circles, wholeheartedly, and thank Jeff for his diligent pursuit of enhancement.

On reading this entire thread, yet again, I believe there is a solution available.

Create a new tool by exposing, existing, functionality of arcs and circles and cardinal points.

Each know where there centre point lay, both know there own radius, and are capable of changing the number of sides after an initial commit. So, iff...

one: show the radial vectors and the centre point and use the end vectors as a handles for radial positioning (dividing the arc segments to fit between two, user selected, end points [with inference?])
two: when 'scaling' an arc from either end, allow a modifier key to scale from arc centre point (like a full circle does)
three: scale a copy from centre point, as a bonus modifier key...

then we could at least draw them quicker.
leave the rest alone... for now!

john

DesertRaven

So here is another annoying short cumming of Sketchup.
I know the follow me tool is not supposed to be a real revolve tool, but why can't this be made to work to create a geometry one would expect?

result?!

To me this also defies any logic in an alleged simple to learn design software.

@unknownuser said:

@ Jeff, there's something wrong on the windows side of things pertaining to this.

No there isn't, on windows entity info shows the correct units and digits.

jeff hammond

@desertraven said:

So here is another annoying short cumming of Sketchup.

that's the same error.. there are lots of ways to see it (though this way makes it pretty obvious )

TIG

The affects of an Arc path's segment's angle to the face to be extruded with native FollowMe have been discussed many times over...

Here are the three variants.
An Arc path that does not have its start segment perpendicular to the face: radius sized base, but odd start/end facets.
An Arc path that does have its start [half] segment perpendicular to the face: non-radius sized base, half start/end facets.
EEbyLathe, which produces the desired result: radius sized base, all full facets matching Arc's segmentation etc...

Alan Fraser

I'm not really understanding what the 'error' is in Olav's example. The path appears to be set up to produce two very short, on-axis side segments, joined by a 4 segment arc. Isn't this exactly what Follow Me has done?

I hate to sound like I'm defending SU yet again, but the quadrant below was produced in a single action (using Loft junctions along 2 paths).
You can't realistically expect SU to guess your intentions. If you want a quadrant with two end profiles sharing a common centre of rotation and at right angles to each other, then draw them that way; don't provide SU with a path with which it doesn't stand a cat-in-hells chance of achieving your aim and expect it to magically solve your geometry errors for you.
The same goes for drawing an extrude path using a simple arc...like when you round off the corner of a rectangle, In that situation, neither end segment will be on axis...so you can't realistically expect the resulting end profiles to be either.

I'm not saying that the devs or a Ruby scripter couldn't produce a specific tool for producing such a quadrant (leaving the default extrude action exactly how it is...which works perfectly well in most cases) But there are plenty of ways of achieving the same result at present, including the plugin I used, which took little longer than using Follow me.

jeff hammond

@alan fraser said:

I'm not really understanding what the 'error' is in Olav's example. The path appears to be set up to produce two very short, on-axis side segments, joined by a 4 segment arc. Isn't this exactly what Follow Me has done?

actually, in that exact situation, i think the measurements are correct.. (i can't measure right now but this is probably right.. and RoundCorner would draw it accurately in a similar situation)
if you put a little flat space prior to the corner, the measurements will start going bad too.
http://sketchucation.com/forums/viewtopic.php?p=451105#p451105

(visually, olav's example is not correct.. you'd have to get the sides back to 90º but upon doing so, i'm pretty sure the numbers will be right)

.

@alan fraser said:

You can't realistically expect SU to guess your intentions.

correct, if there's an arc in the path then it acts one way.. if the path is not arcs/curves then it offsets accordingly.. no guesswork required.

DesertRaven

@unknownuser said:

@ Alan,You can't realistically expect SU to guess your intention

Really? What alternative intentions could I of had in this example? Why would anyone want the result as shown in my diagram?

Why would I, as a user, care about weather or not the SU needs to read the beginning of the path perpendicular? or why would I want to know how the dog on thing is programmed?

I use a path, that starts and ends in a square, an arch just as SU draws it and that is all I got using the ntive SU tools.

I don't have an alternative arch tool that draws an arch with a half segment at the start and the end.

Also talking a bout guessing, when I draw a rectangular shape with the rectangle tool, SU guesses that I may be drawing a square or a golden section. Hmmm,,

And obviously the geometry produced by Sketchup in my example is useless to say the best and completely unacceptable to make it clear.

@ TIG, I haven't expected anything else but that this issue had been discussed many times before, and there is a obvious reason why.
But quite on the contrary to my amazement the developers of Sketchup have not found it necessary to even attempt to make this work. (I guess they are just thinking "good enough")

An arch does not end perpendicular it is dependent on the center and all segments need to be equal.

I think these issues reside in Sketchup because someone keeps denying what is really needed on the user side; and because of that stubbornness, we, the users, have to put up with work arounds and a cluttered up, extensive plug in library.

jeff hammond

@desertraven said:

But quite on the contrary to my amazement the developers of Sketchup have not found it necessary to even attempt to make this work. (I guess they are just thinking "good enough")

dunno, i think it's best you start seeking alternatives.. i mean here in this very thead, one of the developers has said "oh.. that would be neat to have in sketchup"…

neat?

not absolutely necessary?

fighting up to this point was an uphill battle.. after john's response, that hill just turned into a vertical wall..

personally, i'm over it.. can't believe i lasted this long..

gilles

N.E.A.T: never ever ask this.

jeff hammond

DesertRaven

@unknownuser said:

@desertraven said:

But quite on the contrary to my amazement the developers of Sketchup have not found it necessary to even attempt to make this work. (I guess they are just thinking "good enough")

dunno, i think it's best you start seeking alternatives.. i mean here in this very thead, one of the developers has said "oh.. that would be neat to have in sketchup"…

neat?

not absolutely necessary?

fighting up to this point was an uphill battle.. after john's response, that hill just turned into a vertical wall..

personally, i'm over it.. can't believe i lasted this long..

I'm quite amazed you stuck it out fighting so long and also you never seem to sleep? I'll have to look up Johns respond I may have missed it.
I'm not sure if posting here will ever do any change, but I just think spreading awareness is a good thing.

I've spent so many hours working my ass off using Sketchup on various large scale projects, remembering those moments where I wished I had one of the developers at hand to just give them a good shake to get them back to the reality of what we are doing here.

But despite all frustration so far SU has served me well, if I look on the bright side.
I am able to create vast scenes with tons of buildings and still am able to work on a detail over there and not have the whole thing freeze like I would have had using 3D Max or AutoCAD.
Usually I save each building or building segment in it's own file.

But anyway then again, it get's me when I see all this potential and only some glitches standing in the way of excellence. I have spent hours, and this is not just me, to hunt down a minor little issue that has suck in because either me or one of my guys hasn't payed explicit attention to what Sketchup thinks is the right way. As soon as you leave the square you are in trouble.
Many times I also experience that after opening a drawing after a day or two some minute things have changed.
I don't have a concrete example at hand, but it happened more then once and it always has to do with geometry that is on any angle that is not square. e.g. 2 or more buildings at an 12.5 degree angle to each other.

One time I did an organic shaped thatched roof, every time I attempted to save the drawing Sketchup threatened to "fix" my geometry, only to add odd, twisted faces in a so far clean solid model.

I see there is lots of work to do and a long road ahead if the Dev. is serious about this product.

Anyway, I hope to see you around here some time longer, I appreciate you input and your eagerness to fight for the better of Sketchup, thanks,

Olav

Alan Fraser

I wouldn't want anyone to imagine that I think SU is 'good enough' either. One of the areas of contention here is down to a basic misunderstanding in terms...I think.

Take the example of Olav's quarter hemisphere. My viewpoint is why would you think that the Follow Me Tool would lathe a quarter turn under those circumstances. In order to do that, it would need to be a Rotate Tool; it isn't, it's an extrude tool...as its name suggests.

While it is true that it can be 'tricked' into being a Rotate Tool in one, very specific circumstance...lathing a profile around a full circle, it can only do that because (a) that extrude path has no beginning or end; and (b) the profile's inner edge just happens to be at the circle's centre. You have to remember, however, that it isn't actually rotating around that central axis...it's extruding along the exterior, circular path.

Like I said, this is one single, specific circumstance. It is not downscaleable to lathing a quarter or half turn...because such a path is not an infinite loop...it has ends...and at those ends Follow me will revert to its true nature as an Extrude Tool, exactly as in Olav's illustration. This is the reason that I claim such a result is neither an error nor is it illogical. It is entirely logical...for an Extrude Tool.

Why SU doesn't actually possess a proper Rotate/Lathe tool after all this time is quite another matter.

DesertRaven

@alan fraser said:

I wouldn't want anyone to imagine that I think SU is 'good enough' either. One of the areas of contention here is down to a basic misunderstanding in terms...I think.

Take the example of Olav's quarter hemisphere. My viewpoint is why would you think that the Follow Me Tool would lathe a quarter turn under those circumstances. In order to do that, it would need to be a Rotate Tool; it isn't, it's an extrude tool...

Why SU doesn't actually possess a proper Rotate/Lathe tool after all this time is quite another matter.

But Alan, that is what they are selling it for or else they would of provided the revolve tool long ago.
I've watched those tutorials on SU6 when it was introduced as the architectural tool for Google Earth.
I will never forget the sentence "trust Sketchup it is smart and easy enough to know what you are doing." okay that was pertaining to the inference engine.
And that again proves their position of "it" being good enough.

With the given native tools there is no alternative to creating the undesired result, and then fixing it manually.

So it still boils down to the arch and circle tool where unexpected inconsistency occurs.

as Jeff already elaborated on the "J" shows what anybody would expect of a offset tool to do when dealing with arches.

By failing to deliver a clean solution, there are pitfalls that can affect your work on the computer using SU down the road; if one does not know or pay explicit attention to the shortcomings of SU.

this example shows how inconsistent and deceptive the workflow is with SU.:

The above diagram shows how a circle is created using the inference engines suggestion.

In creating the circle I am following the red direction which creates the circle according to it's radius on a vertice (not at a half segment)
So going from there I'm going to create the half circle for my arch.

Next i'm using the offset tool. Now if I know it all and am paying attention I see there is a inconsistency at both ends of the arch/ half circle. Also I notice SU breaks up the arch / half circle and inserts two end points. Why I don't know.

So let's say I want my offset half circle to be an actual half circle then what do I do??
I go in there and add two small pieces using the inference engine.

Now I realize I've just gained two more end points in my geometry, I can not get rid of, unless I explode the curve.

If i was to close the shape of my two arches I would be able to extrude my 3d arch but then would see that I need to smooth the two lines appearing where the extra endpoints are.

At this point I would rather draw a brand new arch.

Anyway In case I say who cares it's only SU, good enough, I'll take the thing as it comes.

So the above diagram shows the whole dilemma, either the post will not be consistent with the width of the arch, not being 2' but 1' 11 13/16" and that added up in an array can cause trouble just like that,
or I fix it manually, gaining the before mentioned extra endpoints.

So the extra endpoints can lead to a problem in SU, that is if you don't pay explicit attention, it will be a pit fall.
The inference engine could, because it is so close to the desired endpoint, pick the wrong one and then your faces won't close on some wall down the road when you're not thinking of the extra endpoint any more.
This is one of those minute little things that can get you down the road. And then you got to get back and find that little bugger.

I think the circle and arches really need to follow what one expects them to do when offsetting, copying, or extruding along them as a path. They are circles and arches and should not act as if they were segments.

jeff hammond

@alan fraser said:

Why SU doesn't actually possess a proper Rotate/Lathe tool after all this time is quite another matter.

that reminded me of something from when sketchup8 first came out.. they used this picture as promo for the new solid tools..

this is a screenshot of a video.. don't try to push play.. it won't work ;)

but you can't draw that with sketchup tools and have it be recognized as solid.. try that with followme and you'll have all sorts of problems

http://sketchucation.com/forums/viewtopic.php?p=268160#p268160

Alan Fraser

Olav, in the case of the arches, wouldn't it make more sense not to bisect the circle until after you have offset it? And when you offset it, just ignore the value in the Measurement box, but offset by inferencing across the top of the established pillar?
That way, you not only avoid all the nonsense of having to deal with little line stubs, but you also get an entirely consistent vertex to vertex distance up the pillar and right around the arch.
Or am I missing something?