#ACAD/3DS interface ??
46 messages in this thread
Is it just me or are others frustrated by the quality of imported meshes from
Acad to 3ds.
There are certain things done better in Acad, precision 3d construction,
radiused filleting of complex solid objects, for example. But so often when
these meshes are brought into Studio, no matter how you weld or smooth, some
faces just refuse to smooth and spoil the whole image. Re-arranging the face
structure (modify edge & collapse) will sometimes improve the shape of a face
(faces with one very small angle causes problems) but never completely solve
the problem.
Objects created entirely in 3ds rarely cause a problem which makes me perform
all sorts of contortions in shaper and lofter to avoid using Acad. This can't
be right. AutoCad is very powerful tool and AutoSurf is largely unused on my
system because I cannot even transfer the simple tutorial surfaces into 3ds and
get perfect smoothing – and that's using AutoVision's new 3ds output!
So tell me. Is it just me, or is it time the two programmes shook hands a bit
more betterer (sp,sic,bg)
JR
AutoSurf is a great modeler – you can model just about anything with it. BUT,
it is a SURFACE modeler, intented to export surface data for milling etc.
What that means is that peices within a surf model have no knowledge of what
is next to them. A cylinder for example has three surfaces – its body and two
caps, but they have no knowldege that they have such a relationship. Thus,
when you mesh them the vertices do not align, regardless of mesh settings,
and you're right – they're not going to smooth too well in 3DS (or AutoVision
for that matter).
Note that Designer is a solid modeler and does understand the overall
relationship of pieces (as long as you don't have diverse compound fillets)
and you will probably some better luck exporting to 3DS with it.
– Phil
I take your point about AutoSurf, but the thing is that the two strings of
manufacture and visualisation should be able to integrate much more. I've been
commissioned to produce a model of a new kind of engine. It is for a small
company and they would like to kill two birds with one stone and have CNC data
and product visualisation and video from the same data input.
AutoSurf, Autocad and AME all have the facility to convert surface or model
information into mesh or 3d face data. It is this conversion and then
translation to 3ds faces that I would like to see improved. It seems to me
that although Autocad and 3d Studio have leapt ahead in the last 3 years the
connectivity has not progressed much from the old DXF/3DS utility. Maybe I'm
setting my sights too high.
John
What can I say? I concur!
>>It is this conversion and then translation to 3ds faces that I would like
to see improved.
Bingo. I suggest you voice your concerns on this issue over on the ACAD
forum so it gets heard by the people that represent those products.
>>It seems to me that although Autocad and 3d Studio have leapt ahead in the
last 3 years the connectivity has not progressed much from the old DXF/3DS
utility.
Correct again <g>. Good 'ol DXF3DS.EXE is still your best bet for raw mesh
conversion. AutoVision does have some nice abilties for 3DS exhcange however.
>>Maybe I'm setting my sights too high.
No, absolutely not. Your's are valid concerns and worth airing. Express them
to your dealer(s) and, of course, of ACAD.
– Phil
Have you tried different settings of the "smoothing angle" when you import your
data?
<<Have you tried different settings of the "smoothing angle" when you import
your data?>>
Yes, with limited success. Some AME models lose faces altogether in transfer
but the worst is objects created with lots of boolean work. Some objects
created in AME with bool' will not smooth at all, but even in 3ds, objects
surfrev'ed and then bool'd can have faces that will not smooth whatever you do.
John
>>Some objects created in AME with bool' will not smooth at all, but even in
3ds, objects surfrev'ed and then bool'd can have faces that will not smooth
whatever you do.
Are you uniforming normals on conversion of solid models? This is usually not
the correct choice for solid models – they have correct normals, and results
from Boolean operations can confuse the "auto" normal facing.
Examine your meshes carefuly that need to smooth. Ensure that vertices match
at the transitions and that they are a contiguous mesh. Remember, smoothing
can only occur across welded vertices. If AutoSmooth still isn't working,
bite the bullet and assign smoothing groups yourself. You'll find you can
assign FAR fewer than Autosmooth does (making them valuable for selection
purposes later) and you can decide what constitutes an edge or not – you're
not constrained by a universal angle that may not be appropriate across the
entire mesh.
Good luck,
– Phil
Phil,
Now you've got me fired up I'm going to dig out couple of meshes that really
gave me trouble last year and try everything you suggest. They were both
created in AME with major boolean operations. I gave up trying then but it was
right at the beginning of R3 and difficult to sort wood from trees. I'll let
you know how I get on – but not for a couple of weeks – I'm off to over France
next week for a wee hol and the biggest problem now is whether to take the
ferry and risk it capsizing or go through a 25 mile drainpipe.
John
>>whether to take the ferry
What? And miss camping out with a hundred backpackers? How about the night
ferry, with that beautiful foggy sunrise at the end of a fitful sleep in those
unyielding chairs?
How much does the drainpipe cost? I didn't realize it was in operation yet.
Michele <personally preferring the jetfoil>
Michele,
>> How much does the drainpipe cost? I didn't realize it was in operation yet.
<<
The channel tunnel (Chunnel, to people in the know) is supposed to become
operational November 15th this year. The deadline keeps getting shifted further
back because the English haven't finished yet – the French have been ready for
some time.
Martin (personally preferring the slow boat with 1000 fellow backpackers)
Martin,
The Drain [I prefer to call it that] is running a partial pre-opening service
now, but it's not advertised. Shareholders and the like [not me!] are being
given priority but it might be possible to get pot luck. I think it's easier to
come this way from France. We might try it but I'm not sure though.
John
John,
>> The Drain [I prefer to call it that] is running a partial pre-opening
service now, but it's not advertised. <<
I'd like to walk "the Drain", but the scenery would be pretty dull and lying
flat every time a train came would be stressful <g>.
>>personally preferring the slow boat with 1000 fellow backpackers
You would, you scruffy hitchhiker you. <g> I did at first but it got old after
a while. I treated myself to the jetfoil "just once", and after that the ferry
lost its charm. I think it's okay in warm weather but pretty miserable on those
cold foggy overnight trips. And you have to admit, the food stinks.
Michele
Michele,
>> And you have to admit, the food stinks. <<
not anymore ….. Sealink has MacDonalds <G>
>> Sealink has Macdonalds <G>
As someone who is mostly vegetarian and allergic to wheat, about the only thing
I can eat at MacDonalds is the salt packets. <g>
Quite!
I was told the Chunnel opens this week, but there's been no hype. I've never
used a jetfoil – good idea – we'll try it. The best ever was the little Air
Ferry [this is really showing my age] in the late '50s and '60s. Drive down to
a small airfield, 6 cars on the plane, 20 passengers, 15-minute flight – very
civilised. There is nothing to match it now and they say we have made
progress!
John
Because the jetfoil costs L10 more per return trip, you get a different class
of people from the ferry. The passengers on the jetfoil do not bring an entire
boiled herring dinner to spread out on the floor, picnic-style, in front of the
restroom door. You also don't have to stand in line for two hours for a L5
greaseburger. Everyone gets a comfy chair with a tray table and a good view,
and a stewardess-type person comes around with duty-free items, money exchange
and food (tasty but pricier).
You can justify the expense if you strike up a conversation with the other
friendly people and give them your business card before disembarking. <g>
Michele
<< give them your business card before disembarking >>
You mean I can't get my laptop out and show them my lastest animation.
Are you by any chance suggesting a little restraint here?
Is there a remote hint that maybe I should actually forget about DXF, Gamma,
vertex welding, deadlines, the bank manager and all that stuff for 7 whole days
and just drink Champagne and Burgundy; eat all that yummy food and drive around
the french countryside in my lovely new car and not feel guilty about it?
Nah, I don't think I can – – but I'll give it my best shot.
Catch up with you all in a week. Au Revoir
John
John, do you need your car in France? In my experience the best way of
crossing the Channel (and hundreds of miles of land mass) is by airplane. It's
done in one to two hours and you also get your your own seat and your own tray
table and food and drinks… and if you're flying with a nice price ticket it's
not too much more expensive than going by ferry or train (except you live in
Dover and want to go to Calais…<g>)
The two alternatives I've used so far are:
1) ferries: slow, sluggish way of crossing water; tend to sink too often for
the time being; sometimes they stink (even with McD on board…)
2) hovercraft: fast way of crossing water; especially fun when wind blows at
10+; but too loud and no good view provided
Regards,
Volker (who's now wondering if anyone cares about his two cents
(pennies, centimes) anyway…<g>)
Volker,
I have three loves in my life: my car, my wife, and rural France (not
necessarily in that order).
All three go very nicely together. Need I say more?
NOW, as the G man himself says "I'm outa here" – or as the folks around here
would say; "I'm awa the noo"!
John
John,
have a pleasant trip with and to your three loves, or, as you very
probably get this message _after_ returning home, I hope you've had a pleasant
trip with and to your three loves. Sigh! Wish I could have fresh French
baguette, cheese, wine and all these yummy things on location…
Volker
Hey John, I was reading all the messages on this topic and didn't see anything
about writing a dxf file in binary format. It is my understanding that the
solid modelers in ACAD are double precision and that 3DS is not. I may be way
off the mark here, but hey it may solve the problem you're having. Let me know.
Jeff
Jeff:
A lurker surfaces! Hi there! Do any fishing recently?
Greg Pyros
Hey Greg, Funny you should ask. Lee and I are planning our next shark fishing
expedition…an all nighter…care to join us?
Jeff
Jeff:
I'd love to! When do we embark? Have you scheduled the thunderstorm? <g>
Greg Pyros
Hey Greg,
>>have you scheduled the thunderstorm?<<
Actually Greg we thought we would make it a little more exciting, you know it
is the tail end of hurricane season? Got confirmation that one would pop up
just about the time we were done getting the bait. Actually the only
confirmation I've received is that I'm going to Autodesk University! Very
excited on this end. Can't wait to get to this class I'm taking "How to produce
excellent Animations". See you there.
Jeff
Jeff:
Glad you will make it to AU! Looking forward to seeing you again!
Greg Pyros
The DXFOUT command in ACAD has a default of exporting to 6 digit precision, I
believe.
Nik,
I posted several messages about trying to get an ATI Mach 64 w/4mb to work on
an Intel pentium 90. This message is one of the latest. I haven't gotten a
lot of response on this problem. Maybe I haven't been checking the forum often
enough. Since this message I have tried another Mach 64 w/4mb on the Intel
board. I have also tried one of the Mach 64s in a Compaq 566 XL system.
Neither of these trials worked. I also tried the SoftEngine F/X ADI drivers
that ship with the Mach64. I set up vibcfg to use ADI drivers and set the
correct rcpadi environment variables. This gave me the error: Display video
hardware did not initialize
I am out of ideas. Can you help!
<I was wondering what motherboard your using with the ATI card. I have <been
trying to get a 4Mb Mach 64 card to work with an Intel P90 <w/64Mb. It works
fine in Windoz but I get get a page fault when I try <to run 3DS3. The ATI
works fine if I remove the 2Mb upgrade, but 2Mb <won't do with a 21" monitor.
I have updated 3ds to the latest (1.0.4) <inline drivers, tried different ram,
updated the flash bios on the <motherboard and reinstalling 3DS. I am about
out of ideas. I may <have to go back to a Matrox card.
thanks
SSH -Graphic Realities-
Hi Jeff,
you are 110% right on cue here,
(about using the binary format as output.
M.Schneider
Hey Mark,
I have had problems in the past with complex geometry created in ACAD
and couldn't understand why. I tried writing out the file as binary and bingo
it worked! Thanks for the input.
Jeff
>>solid modelers in ACAD are double precision and that 3DS is not>>
I thought I was going to be able to sneak through life without anyone knowing I
haven't a clue what 'double precision' means exactly.
Can someone please provide an idiot's guide. But more important, how does it
affect the translation of data and does it matter to 3ds. In other words, is
it just a matter of the accuracy of the defining of a point in space – and
surely 6 decimal places is enough – or is our problem a case of information
being lost in conversion of format.
For a start there is a fundamental difference in the two programs in that Acad
makes 3-d faces with 4 sides and 3ds makes them with 3. Or is Acad just
displaying that way?
John
There are, in the PC architecture, 3 types of floating point numbers. The
difference is only on the amount of precision.
float (or single precision) are 32 bits in length from which 8 bits are used
for the Exponent and 23 for the Mantissa (1 for sign). It can represent a
number from 1.18×10^-30 (1.18E-30) to 3.40×10^38. That's 30 decimal places for
a negative number and 38 for a positive number.
double (or double precision) are 64 bits in length from which 11 bits are used
for the Exponent and 52 for the Mantissa. It can represent a number from
2.23×10^-308 to 1.80×10^308 (308 decimal places).
long double (or neurotic precision) are 80 bits in length from which 15 bits
are used for the Exponent and 64 for the Mantissa. It can represent a number
from 3.36×10^-4932 to 1.19×10^4932 (4,932 decimal places).
Errors in conversion are accumulative but I doubt it very much it would make
any difference even with single precision. With a double precision you could
split an atom at the edge of the solar system (provided you had what to aim
with).
—
What are the cons of double-precision? Basically that it forces the system
to use twice the amount of memory for all geometry. Eventually, when RAM
is cheap enough, this won't be an issue. But for now…
– G
<< (or neurotic precision >>
:->
Beautiful description, Gus. Your elucidation is always refreshing!
– G
Thanks for the explanation, Gus.
You confirm my feeling that I didn't need to worry about the level of accuracy
but now at least I know what it all means.
John
The accuracy reffered to is for vertex coordinate locations (the primpary
thing read by 3DS).
AutoCAD exports at its full precision (16 decimal places I believe) when
exporting a DXF file using the Binary option. 3DS has 10 decimal places of
accuracy so there is a bit of round off that occurs upon import. The
important thing to realize is that this roundoff is consistant, so exporting
multiple DXFs will allign as they did in AutoCAD – on the nose.
Nick was correct in saying that AutoCAD defaults to 6 levels of accuracy for
DXF export, but that's only if you let it use that low level default. Your
best bet is to not enter a number at the prompt but a "b" for binary instead
– full accuracy and a smaller, quicker file.
– Phil
I'm going to do a few experiments on this when I get back from my break. I'm
sure there is more to the conversion problems than the accuracy of vertex
coordinates.
John
Hey John,
I'll try to shed some more light on this subject. The way I understand
this is that because 3D Studio does not work to the accuracy of lets say AME
when you dxf a model to 6 decimal places it only will see the first two.
However by using binary in writing out the dxf file you bring the geometry to
the the same level of accuracy which is native to 3D Studio and when welding is
applied no holes are left in the geometry. I'm sure the subject is deeper then
that and I have not spent the time to do in-depth research on it, but I do know
I had problems with complex meshes created with AME and when I used binary on
my dxf files they seem to have disappeared. Hope this helps.
Jeff
To calculate precision (especially the difference between 3DS and
Acad), you need to understand the differences between single and
double floats. Here's a good posting from Gus on the subject:
: 126684 S2/Autodesk 3D Studio
02-Oct-94 05:32:13
There are, in the PC architecture, 3 types of floating point numbers.
The difference is only on the amount of precision.
float (or single precision) are 32 bits in length from which 8 bits
are used for the Exponent and 23 for the Mantissa (1 for sign). It
can represent a number from 1.18×10^-30 (1.18E-30) to 3.40×10^38.
That's 30 decimal places for a negative number and 38 for a positive
number.
double (or double precision) are 64 bits in length from which 11 bits
are used for the Exponent and 52 for the Mantissa. It can represent a
number from 2.23×10^-308 to 1.80×10^308 (308 decimal places).
long double (or neurotic precision) are 80 bits in length from which
15 bits are used for the Exponent and 64 for the Mantissa. It can
represent a number from 3.36×10^-4932 to 1.19×10^4932 (4,932 decimal
places).
Errors in conversion are accumulative but I doubt it very much it
would make any difference even with single precision. With a double
precision you could split an atom at the edge of the solar system
(provided you had what to aim with).
Thanks, that helps. One last question
if I might- How does Ray Trace Bias
affect the position of the edge of shadows?
Does it affect them only where shadow casting
objects are near shadow receiving objects, or
all along the edge of the shadow?
Thank again for your help
KJG [LFA]
Hi Kevin,
<< One last question if I might- How does Ray Trace Bias affect the
position of the edge of shadows? >>
Please see pg. 7-163 in the 3DS Ref. Man.
>> Please see pg. 7-163 in the 3DS Ref. Man.
Thanks, but you obviously jumped in the
middle of this or I was passed on to you
for asking too many questions…
I'll try once more. I'm comparing a 'shadow'
cast by an object onto a surface by a ray
traced light source with a calculated shadow
created in AutoCAD. I was able to work out
the precision of the ACAD method because the
math involved was relatively straightforward.
I like using 3DS images to describe shadow effects
as they are much more clear. So… I am trying
to determine the precision of the ray-tracing
algorithms used by 3DS so that I might account
for any differences in results.
I realize this is probably not a simple question,
so all I will ask for is the best reference which
might describe the method used by 3DS for ray-traced
shadows, (Vince doesn't really say much about it,
Foley/Van Dam describe several).
I asked specifically about the bias setting because
I thought it might be a simple answer: does the
setting affect the shadow placement along the entire
shadow, or just where it is close to the casting
object.
Thanks again for any light you can shed…
KJG [LFA]
-Kevin Gilson
Hi Kevin,
<< I asked specifically about the bias setting because
I thought it might be a simple answer: does the
setting affect the shadow placement along the entire
shadow, or just where it is close to the casting
object. >>
This is an interesting question. I haven't seen any threads on this over the
last 2 years. Since this behavior is not documented in the Reference manual or
in any of my 3DS API docs, I would suggest that you create a controlled test
and post your empirical results here on the forum for further discussion.
jonas[adesk]
>> I asked specifically about the bias setting because
I thought it might be a simple answer: does the
setting affect the shadow placement along the entire
shadow, or just where it is close to the casting
object.
From my experience, the Bias setting seems to affect where the edge of the
shadow engages the casting object and not the extents of the shadow itself.
Any clarifications on this Gary or Jack?
– Phil
Yup, David's right. Here's an old post from Gus regarding the difference
between single, double and long precision. Note that David's comment
re reorienting the scene around the origin ought to help quite a bit.
– G
: 126684 S2/Autodesk 3D Studio
02-Oct-94 05:32:13
There are, in the PC architecture, 3 types of floating point numbers. The
difference is only on the amount of precision.
float (or single precision) are 32 bits in length from which 8 bits are used
for the Exponent and 23 for the Mantissa (1 for sign). It can represent a
number from 1.18×10^-30 (1.18E-30) to 3.40×10^38. That's 30 decimal places for
a negative number and 38 for a positive number.
double (or double precision) are 64 bits in length from which 11 bits are used
for the Exponent and 52 for the Mantissa. It can represent a number from
2.23×10^-308 to 1.80×10^308 (308 decimal places).
long double (or neurotic precision) are 80 bits in length from which 15 bits
are used for the Exponent and 64 for the Mantissa. It can represent a number
from 3.36×10^-4932 to 1.19×10^4932 (4,932 decimal places).
Errors in conversion are accumulative but I doubt it very much it would make
any difference even with single precision. With a double precision you could
split an atom at the edge of the solar system (provided you had what to aim
with).