Creating 3d Model Building - Evermotion
Creating 3d Model Building - Evermotion
org
T HI S I S O L D T UT O RI A L . I T CA N B E DI S P L A Y E D WRO NG O R CO RRUP T E D - I F S O P L E A S E CO NT A CT US
in this tutorial, w e w ill make a 3d model building from still images(non-orthographic). w e can so easliy make 3d model if w e have any information about measurements. in first step i'll show you if w e
don't have any information and measurement of that building, and w e just have minimum reference image.
We have tw o pictures as reference. These picture show a train station. It's located in Semarang, w ich w ere built
during dutch colonial. Exactly I don't know any information about its measurement definitely.
Reference image
Picture1
Picture2
Open MAX,
Match Bitmap
background aspect ratio.
Vanishing Point
Look at the edge of the object from ref image, find VP in picture area and draw a line by continuing from
edge.
See pic. All red line from edges, parallel to y axis(w orld coordinate system).
We have had a vanishing point/point perspective. Its location must be arround there.
For the next step, w e should look for other vanishing point/point perspective. But for this ref image, it w ill
be so difficult to find them.
Approximately, there are 3 point perspective on that ref image.
You should know about point perspective system ( one point perspective - six point perspective ). see
links n references in the end of tutorial.
To solve problem of seeking other point perspective, draw line from that building basic form.
see picture
Horizon
Approximate location of horizon.(because w e cant see the horizon in ref image).
Horizon w ill be parallel to x axis (w orld coord. system). Most photographer w ill keep a horizon
horizontal.
Create a box as a line (automatically box is horizontal). We estimate the position of horizon. Position of
horizon and vanishing point not in a line. My estimation horizon w rong?
Reset MAX
Repeat prev steps. there is such different, point perspective or vanishing point is out
of the scene.
use mousescroll to enlarge canvas.
Look for the horizon position by using box. Lengthen box untill passin VP. horizon cut
the VP, that is good forerunner.
- Camera Matching
to match the perspective of a camera to a photograph. We w ill need accurate scene measurements.
but if w e don't have any information about measurment w e can still make an estimate and matching the perspective to ref image.
Measurements
Reset MAX
Analyze ref image. find any object can be use as measure comparator. i use that oldman in the ref image.
w e estimate height of that oldman is 155cm(too short..no problem)
Create Box w ith height as high as building(line guide). Rotate box corresponding to Z axis perspective.
add segs box, height of segs as according to height of the people.
w e get 7 segs.
155cm x 7 = 158.7V
10.85m = 158.7V
V = 0.0678
304.8 x V = w idth.
304.8 x 0.0678 = 20.665m
w e get w = 20.665m
Camera Matching1
Reset MAX
Length
right click on 3d snap, check only vertex. then click snap to activate (S).
Create>helper>Camera Match>Campoint.
Open box object properties, uncheck backface cull. and check see-trough.
so w e can see box backface.
w arning,"background image ratio should be set to that of the renderer, fix?" yes, ok.
Create camera.
w e can see, box not match w ith w hich w e expect. that mean estimation of our length is w rong or
position of vanishing point is not accurate.
Fixing
change position of 2d point and assign new position. w e just need to assigning new position of
campoint05 and campoint06.
a good camera error range is about 0 - 1.5
Until w e have satisfied w ith match result. then w e see camera position w hich have been made.
the photographer height is about 165cm and camera position is about 160cm..
Camera matching 2
on other view port (say a left view port), change to perspective view .
Create campoint.
see pic
Assign position.
before w e create camera, uncheck "use this point" at campoint 1-6, because w e dont need
now .
utilities>camera match>uncheck use this point.
Create camera,
error, not enough point, ok w e need 2 additional point.
the problem is, w here w e can determine position of 2 additional point.
w orkflow :
John R. Moreland, " Reference Photo Tutorial for Purdue Virtual Campus Project ", ;
this page about creating 3d from ref image, using orthogonal photo reference.
Olle Bergman, " 3d model from a photo ", : this paper about creating 3d from ref
image using SketchUp.
Point Perspective:
w w w .termespheres.com
w w w .olejarz.com
Other:
add comment
© E v e r m o t i o n 2 0 0 9 - a l l r i g h t s r e se r v e d - c o p y i n g a n d p u b l i sh i n g w i t h o u t p e r m i ssi o n i s f o r b i d d e n .
EV ERMO T IO N s . c . , u l . Pr z e d z a l n i a n a 8 , 1 5 - 6 8 8 B i a l y s t o k , p h . : + 4 8 8 5 6 7 4 5 0 0 5 , f a x : + 4 8 8 5 6 5 3 3 6 2 0 PO L A ND. O u r o f f i c e i s o p e n f r o m 8 a m t o
4 p m ( + 1 . 0 0 G MT )