 | | | | | |
At what angle and point do the arm/shoulder of the minifig connect to
the torso? What is the angle and offset of the bend of the minifig arm
and what is the angle of the hand with respect to the pin which connects
it to the arm?
I don't see how using the minifig parts (beyond using the figure
shortcut) is possible without having this information. Still I can't
find this information anywhere. Do I need to reverse-engineer the
minifig DAT files to find this out? Like I had to do to find out the
angle of the "2904.DAT Technic Motorcycle Pivot"?
I'm sorry for reiterating this point, but I really think we need some
kind of knowledge database for this kind of information. Without this
information, the LDraw parts database just isn't as useful as it could
have been.
What do you think?
(And can someone please answer my original question?)
Fredrik
| | | | | | | | | | | | | | | |
| |
|
In lugnet.cad.dev, Fredrik Glöckner writes:
> At what angle and point do the arm/shoulder of the minifig connect to
> the torso? What is the angle and offset of the bend of the minifig arm
> and what is the angle of the hand with respect to the pin which connects
> it to the arm?
>
> I don't see how using the minifig parts (beyond using the figure
> shortcut) is possible without having this information. Still I can't
> find this information anywhere. Do I need to reverse-engineer the
> minifig DAT files to find this out? Like I had to do to find out the
> angle of the "2904.DAT Technic Motorcycle Pivot"?
>
> I'm sorry for reiterating this point, but I really think we need some
> kind of knowledge database for this kind of information. Without this
> information, the LDraw parts database just isn't as useful as it could
> have been.
>
> What do you think?
>
> (And can someone please answer my original question?)
>
> Fredrik
The angle of the side of the minifig torso as Ldrawn is 9.782 degrees (arctan
4.655/27) from the vertical. Hope this helps.
Chris
| | | | | | | | | | | | | | | | |
"Chris Dee" <chris_w_dee@hotmail.com> writes:
> The angle of the side of the minifig torso as Ldrawn is 9.782
> degrees (arctan 4.655/27) from the vertical. Hope this helps.
Thanks. Can I safely assume that the author used 10 degrees when
modeling the part?
BTW: This is just what I find unuserfriendly about the parts library:
You had to reverse engineer the part to find the angle. I'm sure the
information I asked about in the original post was available when the
parts were modeled in the first place, but now they're "hidden".
Fredrik
| | | | | | | | | | | | | | | | | |
| |
|
In lugnet.cad.dev, Fredrik Glöckner writes:
> "Chris Dee" <chris_w_dee@hotmail.com> writes:
>
> > The angle of the side of the minifig torso as Ldrawn is 9.782
> > degrees (arctan 4.655/27) from the vertical. Hope this helps.
>
> Thanks. Can I safely assume that the author used 10 degrees when
> modeling the part?
>
> BTW: This is just what I find unuserfriendly about the parts library:
> You had to reverse engineer the part to find the angle. I'm sure the
> information I asked about in the original post was available when the
> parts were modeled in the first place, but now they're "hidden".
>
> Fredrik
No, the part is authored with an angle of 9.782. The (x,y) values of the lower
facet of the shoulder is (14.345,2) and that at the waist is (19,29) - hence
4.655/27. Or alternatively the top of the torso is (14,0), giving 5/29 - the
same ratio.
Generally, it is much easier to measure part dimensions acurately than measure
angles accurately (not that 14.345 was _measured_ directly - in fact the top of
the torso would have been measured as 14). I doubt very much that the original
author ever _knew_ the angle you are seeking, but measured the width and height
of the torso. Please correct me if I am wrong, Steve.
Chris
| | | | | | | | | | | | | | | | | |
"Chris Dee" <chris_w_dee@hotmail.com> writes:
> No, the part is authored with an angle of 9.782. The (x,y) values
> of the lower facet of the shoulder is (14.345,2) and that at the
> waist is (19,29) - hence 4.655/27. Or alternatively the top of the
> torso is (14,0), giving 5/29 - the same ratio.
Ok, thanks. I understand. It's not interesting for me (in this
setting) to know the angles and offsets of the _true_ minifig. What
is of interest is the values that were used to _model_ the minifig.
This is because what I want to do is to pose the minifig LDraw parts
in various situations, and want to make sure that the parts connect
correctly to each other. I'm sure other people have experienced a
similar problem before.
Fredrik
| | | | | | | | | | | | | | | | | | | | | |
| |
|
In lugnet.cad.dev, Fredrik Glöckner writes:
> Ok, thanks. I understand. It's not interesting for me (in this
> setting) to know the angles and offsets of the _true_ minifig. What
> is of interest is the values that were used to _model_ the minifig.
> This is because what I want to do is to pose the minifig LDraw parts
> in various situations, and want to make sure that the parts connect
> correctly to each other. I'm sure other people have experienced a
> similar problem before.
Considering the accuracy to which the torso and arm parts are modeled... is
it possible at all to do realistic poses (like an ABS minifig) without nasty
gaps and collisions? I've had trouble doing this before, but only just
realized how bad it really is when I saw the official motorcycle rider
shortcut that has huge gaps. I then tried again to get a fig to work
right... and decided it just can't be done.
Has there been any discussion regarding redoing the arm parts to more
accurately model the real thing? I'm very interested in having arms with
reasonably accurate geometry and useful rotation centers. The rotation
problem can be solved by putting the arms into subfiles. This also solves
the problem with the non-axial plane along which the arm rotates. But the
arm parts simply aren't shaped like the real things, and if you position
them where they aren't partly inside the torso then you have all kinds of
flaws- gaps around the torso shoulder holes, wide/high shoulders, and hands
that still collide with the torso and hips.
Are the arm part ID numbers taken from the arms somewhere, or are they
arbitrarily assigned?
--
Tony Hafner
www.hafhead.com
| | | | | | | | | | | | | | | | | | | | | | | |
Tony Hafner wrote:
> Has there been any discussion regarding redoing the arm parts to more
> accurately model the real thing? I'm very interested in having arms
> with reasonably accurate geometry and useful rotation centers. The
> rotation problem can be solved by putting the arms into subfiles.
> This also solves the problem with the non-axial plane along which the
> arm rotates. But the arm parts simply aren't shaped like the real
> things, and if you position them where they aren't partly inside the
> torso then you have all kinds of flaws- gaps around the torso
> shoulder holes, wide/high shoulders, and hands that still collide
> with the torso and hips.
I have had thougt about this and I also think this is worth to be
taken over. But I don't have time to realize this. If no one has a
something against this you could make realistic arm parts.
CU Bernd
| | | | | | | | | | | | | | | | | | | | | | | |
In lugnet.cad.dev, Tony Hafner writes:
> Has there been any discussion regarding redoing the arm parts to more
> accurately model the real thing?
I would *love* it if someone were to redo the arms and hand with higher
accuracy. :)
Steve
| | | | | | | | | | | | | | | | | | | | | | |
In lugnet.cad.dev, Steve Bliss writes:
> In lugnet.cad.dev, Tony Hafner writes:
>
> > Has there been any discussion regarding redoing the arm parts to more
> > accurately model the real thing?
>
> I would *love* it if someone were to redo the arms and hand with higher
> accuracy. :)
Note that it's probably not possible to make proper arms that are backwards
compatible because of the hand placement on the ends. So we'd need to
assign new id numbers to the arms... right? Either that or we break a few
thousand files.
It may be possible to use the existing hands, though, or at least have a new
hand part that is backwards compatible with the old hand. I haven't looked
that closely at the hand itself- I don't know if the hand-to-wrist angle is
accurate or if the rotation center makes sense, which are my primary concerns.
--
Tony Hafner
www.hafhead.com
| | | | | | | | | | | | | | | | | | | | | | | |
| |
|
In lugnet.cad.dev, Tony Hafner writes:
> In lugnet.cad.dev, Steve Bliss writes:
> > In lugnet.cad.dev, Tony Hafner writes:
> >
> > > Has there been any discussion regarding redoing the arm parts to more
> > > accurately model the real thing?
> >
> > I would *love* it if someone were to redo the arms and hand with higher
> > accuracy. :)
>
> Note that it's probably not possible to make proper arms that are backwards
> compatible because of the hand placement on the ends. So we'd need to
> assign new id numbers to the arms... right? Either that or we break a few
> thousand files.
>
> It may be possible to use the existing hands, though, or at least have a new
> hand part that is backwards compatible with the old hand. I haven't looked
> that closely at the hand itself- I don't know if the hand-to-wrist angle is
> accurate or if the rotation center makes sense, which are my primary concerns.
>
> --
> Tony Hafner
> www.hafhead.com
I have started working on an improved arm part - and agree this will need to
use a new number to aboid backward compatibility issues. I'll try and post
work-in-progress here when I have something worthy of comment.
Chris
| | | | | | | | | | | | | | | | | | |
In lugnet.cad.dev, Chris Dee wrote:
> No, the part is authored with an angle of 9.782. The (x,y) values of the lower
> facet of the shoulder is (14.345,2) and that at the waist is (19,29) - hence
> 4.655/27. Or alternatively the top of the torso is (14,0), giving 5/29 - the
> same ratio.
>
> Generally, it is much easier to measure part dimensions acurately than measure
> angles accurately (not that 14.345 was _measured_ directly - in fact the top of
> the torso would have been measured as 14). I doubt very much that the original
> author ever _knew_ the angle you are seeking, but measured the width and height
> of the torso. Please correct me if I am wrong, Steve.
Da, you are correct.
I'd go with the (14,0) shoulder position -- it has better (hidden)
precision. 14.345 is derived, as you pointed out, and it is rounded.
Does LGEO have the minifig torso? It'd be so nice to have truly rounded
edges and shoulders. But they'd look worse in LDraw than the sharp
edges.
Steve
| | | | | | | | | | | | | | | | |
Steve Bliss <blisses@worldnet.att.net> writes:
> Does LGEO have the minifig torso?
No, it doesn't. But it has the head, arms, hands, legs and hips.
Fredrik
| | | | | | |
