| |
In lugnet.build, Paul Mison writes:
> "Selçuk " wrote:
> >
> > Jonathan Perret <jperret@cybercable.fr> wrote in message
> >
> > <snip>
> >
> > > The only reasonable explanation I have found so far is that
> > > they are not using a 3-D package to compose the instructions at
> > > all. I'm still having a hard time believing this but I can't explain
> > > the mistakes otherwise.
> > >
> > > Cheers,
> > > --Jonathan
> >
> >
> > Yes, it is seemingly impossible at the beginning but I've seen a 8880
> > instruction error related to a 1x1 technic brick with hole, and I still
> > don't have any clue how a 3D CAD system can do this (especially by using a
> > ready to use 3D library of bricks)..:-)
The errors I see are definitely impossible with a 3D CAD system.
Except an _extremely_ buggy one.
On the other hand, such errors are also highly unlikely to appear
in hand-drawn graphics. Glaring perspective errors would have
been noticed by the draughtsman.
See 8448 p.203 for example (wish I had a scanner), look around
the bottommost black rack in the back of the car : this picture
is wrecked...
The evidence I've collected so far points to a 2D CAD program.
That would explain how partial pieces could be duplicated and/or
moved as in this example. A mouse accident, probably : you
know how sometimes you go to move an object and forget that
you had a bunch of other objects selected.
So, one possibility is that they have a library of 2D parts
renderings (in a vector format) and they simply pick and
arrange these together to create the illusion of a 3D model.
The first implication of this would be that they need to
create a different picture for each view angle. For example
in 8448 there are at least 3 views (looking at the front,
looking at the back and looking at the bottom of the car -
not really front, back and bottom views but you see what
I mean). I have an example that supports this hypothesis.
Look at p.125 of 8448 : two brown "axle with stud" parts
are added. Note how the car is seen from the back. Up to
page 139 the viewpoint is the same and the brown parts
are still OK. Now on page 141, the car is turned to be
seen from the front and suddenly the brown "axle with stud"
has been replaced with a regular black axle. Jump to p.169
where the car turns again and presto! The brown part is
back.
So there clearly seem to be two separate files used
for the front view and the back view. Mistakes in one
don't migrate to the other.
Now that we are almost certain they use a 2D system for
the model assembly (why would they need separate files
for separate views otherwise?), one might ask how the
individual parts are rendered. Well, it's hard to believe
they go 2D all the way. While the models need to be drawn
once for each view there are not many views so it's still
practical. On the other hand parts come at all sorts of
weird angles so it seems a lot of work to draw them from
scratch for each angle (the articulated joint, for instance,
must be quite a bitch).
Yet, there are things that do not seem compatible with
a 3D rendering solution. One example is the beam hole.
In general the holes are pictured as pitch black. Even
on thin beams or plates where there should be some visibility
through the hole at most angles. But worse than that, on some
parts not all the holes are represented in the same way.
For example in the 8448 instructions, the black 6+3 liftarms have
all their holes black _except_ for the one that is at the bend, which
is filled with dark gray instead, although nothing in the geometry
of the part warrants this change in color. Interestingly, the same
liftarms that appear in the 8480 Space Shuttle in white are drawn
with all the holes in black. Another weirdness : the regular studded
beams usually have black holes; but when viewed from one particular
angle the holes are gray. See 8448 p.15 for an example. This is just
shading so there's nothing to be alarmed about. _Except_ that none of
the studless beams (liftarms) have gray holes even in the exact same
angle. For example in 8448 p.31 you can see that the liftarms have
black holes (except for the hole-at-bend peculiarity mentioned above)
but the 2x1 two-holed bricks that are parallel have gray holes.
Funnily, in 8480 all the holes are black regardless of the angle or
part considered. So the gray holes could be a shading improvement
that has not yet been propagated to all the parts (possibly because
of aesthetic considerations).
Finally, 8448 p.183 has a liftarm that begins with black holes but
suddenly these become transparent in step 2, when a studded part is
inserted in the holes. This one is really weird.
All these inconsistencies indicate that at least some tweaking of the
colors has taken place. One possibility is that they render the 3D
parts to a 2D vector format _then_ tweak them manually. Or their
3D renderer has some sophisticated directives to render the holes
in particular colors (or not at all). I think the most likely
is that they have a huge library of 2D parts renderings (huge
because they need every part in every possible angle), and
quite possibly no 3D library at all. I pity the instruction book
designers if that is the case...
I'd like to hear what other people think of my "findings"...
The funny thing is that if I'm right, we hobbyists have a much
better tool than LEGO in the form of the Ldraw family !
> The consistency in instruction styles from the mid to late 70's onwards
> through until now makes me wonder about this. They couldn't have been
> using a CAD system them (surely? far too expensive) so they'd have been
> done by draughts[wo]men, and it seems reasonable that until (and unless)
> they could get the same quality via a CAD system it would still be done
> manually today.
I'm not a CAD professional but I'm not sure about CAD being too
expensive in the 70's. But consistency is an important point indeed.
As for the quality, I think it's a safe bet to say they could easily
get the quality they have now from a 3D package. Plus they would
get rid of the errors !
The only problem I see with a full 3D solution is that they'd often
need to add 2D elements to the resulting picture. For instance, the
red circles that occasionnally appear. Cuts would probably be harder,
too (though if you look closely at the cut on p.309 of 8448, you'll
see that they have some trouble doing it in 2D as well).
> I do wonder what role CAD/CAM played in the surge of pieces around
> 86/87, with all the new 'diagonal' pieces, tiles, small L-shaped (2x2
> minus 1) bricks and grilles that started appearing then, and the
> increase in printing (the first non-standard fig heads were in 89,
> weren't they?) That seems more likely than a change in the instruction
> scans side.
What I'm seeing now is rather a decrease in printing and an annoying
number of stickers (yuck)...
Cheers,
--Jonathan
|
|
Message has 2 Replies:
Message is in Reply To:
 | | Re: More studs in holes...
|
| (...) The consistency in instruction styles from the mid to late 70's onwards through until now makes me wonder about this. They couldn't have been using a CAD system them (surely? far too expensive) so they'd have been done by draughts[wo]men, and (...) (25 years ago, 17-Dec-99, to lugnet.build)
|
7 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
