1. Blender learning made easy...
Everything but the Kitchen Sink
www.blenderart.org Antea -by Martin Eschoyez
2. EDITOR - Gaurav Nawani
MANAGER/EDITOR - Sandra Gilbert
WEBSITE - Nam Pham
DESIGN - Gaurav Nawani
PROOFERS
Brian C. Treacy
Bruce Westfall
Daniel Hand
Daniel Mate
Henriël Veldtmann
Joshua Leung
Joshua Scotton
Kevin Braun
Mark Warren
Noah Summers
Patrick ODonnell
Phillip
Ronan Posnic
Scott Hill
Wade Bick
Valérie Hambert
WRITERS
Thomas Baron
Pedro BastosTeemu Vilén
Alexander Weide * Characters, individual body parts (i.e. feet, hands, etc).
Stephan Rodriguez
Darcy Brooker * Animals and reptiles.
Brian DeVincentis
Enrique Sahagun * Vehicles and other mechanical rigging, any other use for rigging.
Nathan Moyer
Sandra Gilbert * Add-ons and the python behind them.
Gaurav Nawani
* How to extend blender through python.
COVER ART
Antea -by Martin Eschoyez * New builds/patches/branches.
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
3. 3
"When I announced the theme for
this issue "Everything but the kitchen
sink", I was jokingly threatened with
massive numbers of sink projects."
Sandra Gilbert
Manager/Editor
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
4. 4
I
keeping a blog of the modeling work he their blessings for their open source
has finished so far. I wandered over to (non-commercial) tribute. Even more
love hearing about new blender take a look, and let me tell you, it looks fun, the original modeler is keeping an
projects. Everyone has their own pet amazing. The details and texture work is eye on Andrew's blog. No pressure
ideas on what would be cool to do or beautiful. In addition to still images, he there :P
accomplish using blender. Some also has a growing number of test
projects are by nature short and sweet videos to show off his progress. The Journeyman Project Tribute looks to
and produce cool results or new be well organized and planned out,
techniques that can be shared with Andrew has told me that one of his making it a fun project to keep an eye
others. Then there are the labors of goals is to keep the project open source on as they progress their way through
love. Projects that are destined to take and completely create the HD footage in all their goals.
months of dedicated work. Blender. If his current tests are any
indication, the finished result should be Andrew is looking for additional
The Journeyman Project Tribute is one quite impressive. volunteers to help bring this project to
such example. Not only is the completion. You can contact him for
Journeyman Project Tribute a labor of In an effort to keep the tribute as close further information
love, it is a rather ambitious one at that. as possible to the classic original as well
Andrew Curry is leading an enthusiastic as head off any legality issues, Andrew andrewcurry2@googlemail.com
Izzy Speaks team who have decided to remake the
classic game “The Journeyman Project”
has been in touch with several
members of Presto Studios, including http://andyjourneyman.blogspot.com/
by Presto Studios. The project is the president of the studio. Presto
currently split into three efforts: Studios have given Andrew and his team
* A prequel novel
* A port to Windows of “Pegasus Prime”, a
remake of Journeyman by the original
studio.
* A retelling of Journeyman using Blender
to model and render the video footage,
“Journeyman HD”.
At the moment, there are several
members hard at work on the novel and
port teams, but so far the vast majority
of the HD version has been done by
Andrew. Another team member is busy
modeling and rigging three characters
from the game. Andrew has been
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
5. 5
fully detailed interior. find a previous tutorial of mine on
how to setup blueprints here:
This project is a common example of Blueprints setup with Blender.
an amateur project. The idea came
from the 2009 Real World Racer
contest held by Scratch Made Cars The major steps are :
(SMCars). The rule were simple: “do 1. Find blueprints on the Internet.
No matter whether you are an your dream racing machine”. I choose www.SMCars.net is a very good entry
amateur or a pro, this tutorial to convert my daily ride, a Citroën C5 point, with thousands of blueprints
aims to teach the work flow and Tourer, to a Deutschen Tourenwagen available
every move you need to know Masters (DTM) coupé. During this 2. In case you don’t find blueprints, try to
to achieve a good car. It’s based project I wanted to do a “civil” version gather some side, front and rear views.
on Blender, to allow everyone as well to put in the background of the Top photographs are quite hard to find,
do it without requiring access render but a lack of time prevented but they’re not mandatory.
to high-priced or pirated me from doing so. This year, I entered 3. Collect as many reference pictures as
software. But it can apply to virtually
by - Thomas Baron
another SMC contest called Topless, you can find. Try to avoid CG pictures
any 3D package. about … topless cars. For this contest I unless they’re official, because you don’t
did a Hennessey Venom GT, which I want to reproduce the mistakes from
I based this work on my latest project, posted here, but I’ve had another idea. other CG artists.
a coupe-convertible Citroën C5. If A convertible, civil version of my C5. I 4. There are many good reference
you’re allergic to this brand, or to cars just wanted to make things a bit more picture sites listed at www.SMCars.net
in general, don’t worry as this material difficult, by doing a CC version (coupé but try to grab high resolution photos
applies to any manufactured product – convertible). So here’s my CC5. Yes I and be careful about picking the the
regardless of whether it’s big or small, know, the name isn’t great.
right model year, the right version, etc
cheap or high-end. Setup blueprints (or pictures) like I
explained in my tutorial
This tutorial is an on-going series of 5. Draw a cage by creating a mesh
posts. It starts from the very beginning without any faces. Basically, you create a
of modeling and covers every aspect: The first step is creating a concept plane, delete 3 vertices out of 4, place
hard surface modeling with a focus on drawing of my idea. I always do it in the last one at a particular spot, and
high-end quality and accuracy, best 3D, without paper. So I create what is extrude it several times to create an
meshing practices to create a nice usually called a “spline cage”, which edge. Here’s the resulting drawing.
polyflow and tips to solve the most means I have to draw the main curves
common issues. In addition to the of the car using Blender. However I
body, this tutorial will deal with the don’t use splines for that, I just use a
creation of other parts such as tires, regular mesh without any faces. I
rims, wheel details, windows, grilles, based the drawing on the C5 Tourer
chrome and rubber trim as well as a (station wagon) blueprints. You can
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
6. 6
The trick is to model the major surfaces first, of 3mm, and the sharpened almost-vertical edge
1. The separation lines between the various
then add details. For example, for the two is continuous from one panel to another.
panels. pieces between the bonnet and front bumper, 5. I add the thickness for the sheet metal panels
I do several steps. by extruding edges. I extrude in 2 times, first for
2. The major shapes of the car: wheel arches, the 3mm radius of curvature of the edge of the
folds, bumps, openings etc panel sheet. The second extrusion is intended to
3. Some extra curves which are useful to better have spaces between body panels rendered dark
define the shapes e.g. to show the roundness of a (7mm to a give a final of 1cm)
fender. 6. I finish the details of the body on top of this
The main point as this stage is keeping things piece (it will be behind the lower chrome trim of
as clean and accurate as possible. So I rotate a the brand logo) and add the thickness to the
bottom.
lot around my cage in perspective mode to
compare it with reference pictures. Having the 1. I start from my cage to make a simple but Here’s a perspective shot to better see details of
complete surface. The only detail at this stage is
actual car in front of my house helps a lot, so the last steps.
the almost vertical edge that I have already
you may want to consider modeling your own sharpened. At this stage you must be careful to
car (or your father’s etc…) as one of your first have very smooth surfaces because it will be
projects. If it wasn’t my car, browsing the difficult to change later. The mesh topology is
Internet is an acceptable solution. Choose important as well, try to keep it simple and well
high resolution pictures, but make sure you balanced.
get the very same version that you are 2. I add rows of vertices along the edges to
modelling. sharpen the edges of the body panel where it will
be given thickness later. You should have a Detailed modeling: the front fender
regular mesh topology based on squares from
An introduction to modeling: front the previous step, so this step will be simple. I The next step is modeling the front fender.
bumper take care about keeping constant spacing (0.005 For this step, I will try to go as detailed as I
My spline cage needs more refining but I BU or 5mm for my scale) between the edges and can in my mesh construction workflow.
the additional rows in order to have a regular
chose instead to begin actual modeling. This
curvature of the edges later. First, I the gather the edges that this fender
part won’t go into a lot of details as the goal is
to give you an overview of the workflow. The 3. I create the hole in the upper air intake, relying will have in common with its neighbors (the
next chapter will be much more detailed. on the mesh topology in place. Care must be hood and front bumper). In edit mode for
taken to have a row of faces along the edge of the each piece, I duplicate the whole edge and
hole, just to have a smooth edge. separate it. I also duplicate the relevant edges
4. I cut the bottom (hard to see on the picture, from my cage. I end up with several parts as
sorry) benefiting from having reserved a line of shown below. The pink parts are from the
vertices along the separation in my first step. That hood and bumper, note that they are already
way, I simply add two rows of vertices on either smoothed (because there is level 3 SubSurf
side of that line, at 1.5mm each. By eliminating modifier active) while the black part is from
the initial line, I find myself with a regular spacing
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
7. 7
the cage and isn’t What I did in the below picture is extrude the Before sharpening the corners, we need to
smoothed. wheel arch to create the flat side around it. As add the row of faces along the air outtake. To
stated for the bumper modeling, I don’t care do that, I add extra rows of vertices around it
I now have to yet about sharpening any edge. I also and weld vertices when needed to make sure
merge all the extruded some longitudinal edges and created these rows have the same count in the
edges together. I the transverse, relevant edges (selected in the border. In this case, I deleted the horizontal
start by first merging the corner of the two below pic, so edge going from the upper corner of the
edges from the bumper. Please note that to they appear outtake and added two rows of vertices: one
do so, the horizontal edge has been raised by yellow) I always above and one
3mm to align with the other edge from the check that all below the
blueprint, while the other edge hasn’t moved the edges are outtake. I only
yet. So there is already a gap between the smooth at had to merge
front bumper and the fender below the every time. the two front
horizontal edge, while there is no gap for the vertices to one,
other edge. We’ll Now we can create the faces. Remember to resulting in a clean flow.
take care of it always create rows of faces along all edges
later, when the and holes. For example, the below picture Now it’s to sharpen the corners. To do so, I
surface will be shows I haven’t done it yet on the right of the add more vertices near the corner ones. The
completed. selected vertices. This is why there is a pinch sharpness of the corner is determined by the
here and we will have to create a row of faces distance of these new vertices, closer vertices
During the next step, I merged all the edges, around this means sharper corners. Next we have to get
which included merging the corner vertices. I hole. This will rid of the triangles introduced by this move
removed the extra vertices at 5mm from each be the air because
corner as they were responsible for the outtake for triangles are
corners being sharp on the previous picture. brake cooling. always poorly
With them removed, the corners are handled by the
smoothed. It will be easier to build the It’s time to deal with the air outtake. First, I surface
surfaces with fewer vertices, and it won’t be roughly create the lower faces after deleting subdivision.
difficult to put one useless edge on the wheel arch. It’s not
them back later really detailed at this time because of the lack For example, removing the triangle shown
by adding extra of vertices in the newly created faces. Fewer above in pink is easy: I just have to add more
rows as I vertices is a good thing, as it helps controlling vertices below the outtake so that there is a
described for the the mesh flow, one-to-one relationship between vertices (i.e.
bumper but too few no triangles, only squares). I removed the
modeling. vertices won’t vertical row of vertices (in fact I merged it with
allow you to the neighbors) so I reverted to the previous
We now have a single mesh, made solely from control mesh. Then I added two rows of evenly
edges. It’s now time to prepare some surfaces. anything. distributed vertices using Ctrl+R and
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
8. 8
mousewheel. Now part. This will mode yet. So it’s
we only have to allow us to not an urban
close the faces spend more myth, if you’re like
with the selected time on a few me and a heavy
vertices shown in different issues. non-proportional
the picture. loopcut user,
The above picture shows the completed mesh, you’d better stick to 2.49b for modeling.
To sharpen the other corner, I added 2 extra but it’s far from being enough to call the
rows of vertices, as shown in the picture. fender finished. The first step is sharpening all The next picture shows the more tricky edges.
When you have to do so, remember to keep a the edges. The above picture shows the For these ones, please remember that we
well-balanced mesh by moving other vertices already hard edges in yellow. Please notice need to create a gap between the fender and
as needed. But also remember that when you how the upper edge gets wider from middle to neighboring parts (hood, bumper). So for this
move vertices, you have to check the front in order to smooth the hard edges. The reason, I created two extra rows of vertices.
smoothness of same goes for the below edge where it gets The first one is created 8mm from the actual
your mesh flow wider from the middle to the rear. On the top, border, and the second one 3mm from the
again. Yes it takes please note the polyflow. It’s easy to choose actual border. Now we’ll just have to delete
time, but it’s to achieve either a hard or smooth corner. the original border so that we have a new
required if you Here’s a detail border which is
want to achieve a for reference 3mm from the
clean job. use. neighboring parts
and which is
One of the two extra rows I added is shown by The next followed by a row
the cursor in the next picture. You will notice picture shows of vertices 5mm
that it results in adding a vertice on the edge all the still-to- away.
shared with the hood. So the curvatures of the be-sharpened
2 edges (the one on the fender and the one on edges. However, this move leads to a mistake behind
the hood) won’t the not-yet-modeled headlight. We pushed
be the same Some are quite the border backward by 3mm, so we now
unless you add easy, they are have a shift between the front edges of the
the corresponding shown below hood and fender.
vertice in the and selected in yellow. Those edges just need The next picture
hood edge. To do to be given an extra row of vertices along shows the vertices
so, the easiest them. Once again, I used loopcuts (Ctrl+R) in that need to be
way is to add a full row of vertices in the hood, non-proportional mode to be able to place the moved forward by
as shown in the picture. new row 5mm away from the original one. By 3mm.
the way, I tried to use the 2.50 Series of
I’ve speed up a little bit on mesh completion, Blender and as of writing the latest release The below picture shows how the shift is now
as it’s not really different from the previous (2.54beta) doesn’t allow the non-proportional gone. It also shows how unclean the gap is
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
9. 9
between the hood and fender. This is because different. So just like we did in a previous step,
when modeling the fender, I had to add extra we’ll add more rows on the hood which will
rows of vertices to get a clean and well- give us the same amount of vertices on both
balanced mesh. Just like we encountered borders.
earlier, this led to adding vertices on the
border of the fender which is shared with the Once done, I just delete the border from the
hood. Now that hood (and the row of sister vertices 5mm
the two borders away). This gives us a wide gap which will be
are different, they filled with faces. Select both edges, and press
don’t align the F key. The I’ll go faster on body modeling, as it’s always
anymore once the bottom option the same moves. First, remember that we had
subsurf effects in the popup to clean the area
are visible. menu will near the hood
automatically and fender. The
And we have another issue with the hood- create the faces faces we need to
fender duo. There is a hard edge on the hood, between the work on are the
another one on the fender and a hard area two rows of following.
between these vertices.
two edges where When having that
it should be Once done, I make some loopcuts in the kind of issue, I
smooth and clean middle of the gap between hood and fenders. find it more
(see the following The first efficient to
picture). loopcut is 3mm remove the
away from what middle vertices, then make sure the
It’s no big issue was previously neighboring curves are clean and then
actually. It’s just a side effect of modeling one the fender recreate the deleted vertices with a loopcut. It
surface at a time instead of modeling the border followed will give us a smoothed row of vertices that
whole car in a single piece. The way to solve it by a loopcut in we’ll move as a single item to adjust the
is pretty intuitive. We’ll merge both meshes to the middle. curvature of the surface.
create a single surface, smooth it and then
separate it back to 2 parts. Keep that middle row and remove the Once the vertices are recreated and the
neighboring rows: one from the hood, two surface smoothed, we get the following result.
Because of issue from the fender. Now the mesh flow is clean,
#1, the two parts but only to better show how the surfaces
can’t be merged aren’t clean.
cleanly because
the number of
vertices are
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
10. 10
It’s now time to split the hood and the fender 4. Start building your mesh The lower part show the fully modeled
back to two separate pieces. I won’t detail it About the side skirt, I’ve been rushing a bit on surface.
with pictures as it’s always the same move. Do the fender. The bottom rear side need to be
a loopcut at a given distance to specify the extruded to form the side skirt. So I deleted
width of the gap, then separate them by the thickness vertices from fender on that
removing the gap faces, add extra rows on area. The following pictures show the vertices
each part to from the fender which will be extruded to the
prepare a side skirt.
nice edge for Please notice
when we will Now it’s the usual trick: sharpen borders and
that the
add edges, and add thickness.
vertices on the
thickness. right are from
the side skirt
Now with
cage.
thickness added, the fender can be called
finished. At least for now, it’s definitely Once connected, the whole part looks like the
possible to following picture. In terms of industrial
notice bumps feasibility, it’s definitely not realistic. The front
later, so we and rear fender and the side skirt can’t be
could We’re done with the fender. Let’s take care of
produced as a single part, it would cost too
possibly have the doors. Workflow is the same. The higher
much to produce or to repair. But as I’m only
to get back to part of the following picture shows in yellow
making a concept car which will never leave
work on the the vertices from the cage, while unselected
the Blender bytes, who cares?
fender – or vertices are from existing neighboring parts.
any other part. The lower part of the picture is more
interesting.
Now it’s time to deal with side skirts. When
starting a new part of the body, remember it’s The selected vertices are the first I create to
always the same workflow: ensure the straightness of the doors’ shape.
Yes, the shapes are more simple on the doors
1. Duplicate borders from already modeled than every other part we already modeled.
neighboring parts Building the
Time to actually create the mesh. As usual I do
2. Duplicate every other curve from the cage mesh that way
the easy parts first, where the polyflow is
(including the borders from non-yet-modeled is the best way
intuitively built from the existing mesh and
parts) to make sure
the cage elements. This is the higher part of
3. Gather all the borders and curves in a single the surfaces will
the following picture.
object, and merge the vertices if needed be very clean.
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
11. 11
Some pictures (trunk is easy, rear bumper a bit more
show the usual difficult) Here’s my very current stage of
rhythm: sharpen modeling for this project; the main body is
edges, borders. almost done
The only
difference is to
take care of the
sharpness of the fold, as it should not be arazo
r blade.
Another thing worth noticing is how the upper
corners of the doors are sharp while the lower
corners are rounded. The following picture
shows the detail
of the mesh
structure to
achieve those 2
different type of
corners.
I won’t detail
separating doors and adding thickness, it’s
exactly like I explained it before. So we’ll move
on directly to the finished doors, presented
with the rest of the model.
Now you should have reached a significant
level of knowledge to know by yourself how to
complete the roof, roof arches, windshield
frame (really simple parts) and the rear end
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
12. 12
take them yourself with your digital picture. It's the notion of projection of
camera, or you can grab tons of them the real world on the picture.
from the Internet. Some people can
use them to create a model by eye, insight3d works using this projection,
but it requires some serious attention by reverse computing the projection.
due to the nature of perspective. Give it the pictures, a set of specific
Perspective makes closer things points on your model and it will be
Modeling an object is the art of bigger, so the proportions can't be able to compute the location of the
recreating it in digital, 3D space. trusted without correcting for different cameras that shot those
No matter how accurate you perspective. Here comes the approach pictures.
want your model, no matter I wanted to introduce; using dedicated
how fast you'll be working, it software to make this correction. Once the camera is located, it can
will always be about these compute the 3D location of the
famous 3 dimensions. specific points, by throwing lines from
How does it work the camera location to the location of
Some objects are so simple you can each point on each picture. For a
by - Thomas Baron create them from scratch just by Consider our 3D space, where the
given specific point, all the lines will
studying it and taking measurements. object stands, and a set of cameras
around it. Each camera takes a converge onto the real place of this
Take a look around and you'll see reference point on the model. As a
plenty of these kinds of objects. But picture. Put each picture in front of its
camera on a transparent sheet. When result you're able to compute a 3D
there are also many difficult objects model from a set of pictures. Of
with elaborate shapes. Put your ruler you look from the location of the
camera, the picture will perfectly course it's a simplified 3D model. e.g.
back into your schoolbag, it won't be I only used 171 specific points. But it
of use this time. match the object behind it (if you
don't consider the optical distortions would be basic to draw a better spline
induced by camera defects). cage.
The main solution to modeling a
difficult object is blueprints. These
aren't blue anymore like architect Now consider a specific point on your Using insight3d
drawings from 50 years ago but are object. E.g. a corner of the windshield,
or the center of a rim. You can see this insight3d is Free Software, available
usually orthogonal (front, rear, side, for free at
and top) views of an object. With them point from some of your cameras, so
it shows up on some of your pictures. http://insight3d.sourceforge.net for
you can locate any specific point of the both Windows and Linux. Once
object in 3D space. You can draw a straight line going
from a camera to this point and it will installed and executed, it shows up
pass through the related picture with its unusual interface.
But for some objects, there is just no
blueprint. So what do we have left? (standing in front of the camera on its
Well start straight away. I assume you
Pictures. It's easy to have pictures of transparent sheet) precisely on the
have already gathered the reference
the object you want to model. You can spot where this point is located in the
pictures of your object. Mine is a
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
13. 13
When moving the mouse over a cross, the
popup shows again to display the same point
in all the previous pictures. For example, the
previous picture shows two thumbnails in the
popup: the location in the previous picture of
the reference point we're about to place in
this picture and the zoomed portion of this
picture under the mouse cursor. When you
have placed this point, the popup window will
auto jump to the next reference point. Nice
concept car only shown at a couple of motor enough. Nice attention from the developer. attention from the developer again.
shows so there is no blueprint available. I'll
add my first picture, using the menu item “File Add another picture now. It will appear but It's now time to place more points to get
> Add Image”. Tip: you can zoom the image using the the reference points don’t show up as you enough data so that insight3d can compute
scrollwheel and pan it using the middle button. have to place them. insight3d must also be the camera locations and triangulate the
able to do the connection with the ones you points. You can try that whenever you want by
placed in first picture. using the menu item "Calibration >Automatic
calibration", and then the item "Modelling >
To do so, use the PageUp and PageDown keys Triangulate user vertices".
to go through the list of existing points. They
will appear on the left of the mouse cursor, Once these two steps are performed,
conveniently allowing you to see all the insight3d should display green dots near the
occurrences of a given point in all the pictures crosses, to indicate where it computed the
you have already added. The picture below position of our reference points from the
shows two things on the popup: the zoom and reconstructed model.
the sample of the previous picture showing
this point. You may have noticed that the reference point
on the right (near the popup) has no green
Use the "Points Creator" button on the left
allows you to place the reference points on
our first picture which are represented by
little crosses.
You can see on the picture that a little popup
will appear with the zoomed-in portion of the
image under the mouse cursor to help you
place the points more precisely. It will
conveniently auto-disappear if you zoom in
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
14. 14
dot. This means that the current amount of perfectly is not possible, many pictures leads have to align it properly. For that purpose, I
data (number of pictures and number of to a beam of lines to compute an average used the yellow edge in the following picture.
points) wasn't enough for insight3d to position which is much more reliable because
compute the position of this reference point the errors are minimized.
in 3D space. This is easy to fix by simply
adding more points on more pictures. As an
example, my final set is using 171 points on
15 pictures for the real project about this First, you export the model from insight3d. I
concept car. used the VRML format, by using the menu
item "File > Export VRML". Then, in Blender,
use the menu "File > Import > X3D & VRML 97"
(the last entry This edge goes from the center of a wheel toth
in Blender e other one. So it defines horizontal and rear-
2.49b) Do not end orientation. I use this property to align
use the "File" > the model. Also, pay attention to the middle
Import > VRML vertices (shown in the white line below), they
1.0", it won't should be aligned as much as possible.
work. When
done, a cloud Here's the full model (171 dots) once properly
of vertices shows up in Blender. orientated, untouched from insight3d except
The green dots may also be slightly off the for a mirror modifier. You can clearly see the
cross. This means insight3d discovered an Unfortunately insight3d didn’t export the proportions look right and the shapes are
incoherent location of the reference point on edges, so you'll have to recreate them. Enter accurate. It's up to you to use more points to
this picture. The green dot shows the place Edit mode (Tab key) and create edges by extract more details from the reference
where the reference point should be selecting two vertices and pressing the F key. pictures and to bring them to your 3D model.
according to insight3d. When done, you'll have the following (this For example, here's the current look of my
picture shows a project (as of writing). I just used some more
It may be right or wrong, because the work-in-
computation performed depends on how progress
precisely you placed points on all the pictures. version of my
Don't hesitate to zoom in to locate points project, so
really carefully as it will help insight3d to there is less
reconstruct a more accurate model. Another than 171
good thing is to use many pictures. points).
Theoretically, 2 pictures are enough to
perfectly locate a point in 3D space because Because insight3d doesn't know about the
there will be only one intersection of the 2 ground and the natural orientation of our
lines. But as placing a point on a picture model, it appears randomly oriented. So we
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
15. 15
points, and added details into Blender. is the "Polygon creator" button on the left One critic about insight3d, about feedback &
side. When activated, you can select a number support. During my project, I encountered
of reference points, and a polygon is shown several bugs. An annoying one is that the
using the points you selected. You can confirm project file doesn't support the use of spaces
this polygon and move on to creating the next in a folder names (at least in the Windows
one by pressing the Enter key. At any time, version I used). I tried to reach the developer
you can cycle backwards through all the about that, but I got no reply from him. It
created polygons using the BackSpace key. looks like he doesn't reply to incoming emails
Use the menu item "Edit > Erase current about his software and a comment on a
polygon" to delete the polygon shown in pink. BlenderNation article about insight3d
confirmed my fear. Let's hope this article can
. cause him to change his mind!
Some extra features within insight3d
insight3d also supports camera export, but http://sourceforge.net/projects/insight3dng/
unfortunately I didn't achieve a useful result.
The exported data for cameras (location,
Picture credits
insight3d features the ability to discover rotation) doesn't match the imported model.
reference points automatically based upon insight3d does support exporting to other file The first picture of a Lamborghini Sesto
picture pattern recognition. This works better formats, for example the .rzi files used by Elemento (shown twice in this article) was
with pictures that are similar and have few Image Modeler, a similar program by found in Wikipedia, the author is Alainrx8. The
changes between them. The tutorial available AutoDesk. Opening the .rzi file in other two pictures of the Lamborghini Sesto
on the insight3d website demonstrates this ImageModeler (export by insight3d) didn't Elemento were shot by Thomas Durand,
feature for buildings, and there it works well. work either. I guess there are different known in the modeling scene as AMV12.
For my example, it didn’t work as I was using coordinate systems for the different exports Congratulations to both of them for these nice
various points of views and several available from insight3d, but I didn't try to pictures
backgrounds behind the model (different read the source code to understand the
motor shows). So I had to create all my relationships between these coordinate
reference points by hand. Make sure to take a systems. Save often.
look at the tutorial on the insight3d website if
you want to try it. There are some minor bugs with the current
version of insight3d (0.3.2). Even though it's a
small version number, it's already quite
stable. However don't forget to frequently
insight3d allows you to create polygons to be save your project.
exported to Blender. I didn't use this feature
that much, except to create lines (2 vertices
polygons) to follow the body lines of the car I
was working on. The key tool for that purpose
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
16. 16
of the character’s mesh. What Each of the five panels in the facial UI
happens is the controls in the facial UI shown in Figure 1 has a controller the
manipulate the character rig structure animator can freely manipulate within
which in turn deforms the mesh. the boundaries of the corresponding
panel. The default location of each
The number of panels in the facial UI controller is in the center of the
usually corresponds to the number of corresponding panel.
Character facial animation is facial regions of the character that
difficult because the face of a need to be animated. Usually the
character assumes many controls are grouped based on the Idea of dynamic facial UIs
complex expressions. To anatomical areas of the face. There The facial UI method presented in this
achieve convincing visual are a number of references to help the article is dynamic because it extends
results for animation, digital character rigger do this better, such as the usability of the common facial UIs
artists need to prepare their Paul Ekman’s book, “Facial Action used in character facial animation. It
characters with sophisticated Coding System.” Figure 1 shows an is a method developed with Blender’s
control structures. One of the most example of common panels placed in potential allowing an animator to
by - Pedro Bastos important and efficient techniques to a facial UI. easily relocate and re-dimension a
achieve good facial animation is the facial UI. The techniques explained in
use of facial control interfaces, also this article allow the animator to
called facial user interfaces, or facial customize the limits and the layout of
UIs. the facial UI and therefore have more
control over the animation of the
Common facial UIs several regions of the face of a
character.
But what exactly is a facial UI? Facial
UIs are GUIs (graphical user Please notice that this article doesn’t
interfaces) with panels and controls to go into precise detail on how to do
facilitate the animation of a common tasks in the Blender
character’s face. Facial UIs are usually software. It is expected that the
placed side-by-side with the face of reader is experienced in Blender 2.5
the character in order for the but if further help is needed just send
animator to have easy access to them. me an email to ptbbastos@gmail.com.
The panels in the facial UI move in 2D So let’s start! First off, get a character’s
space, as well as the controllers placed face model. Figure 2 shows the front
inside each panel. The spatial and side views of the character used
transformation of a control in a given in this article, a funny cartoon
panel will affect the transformation of character called Blue.
a corresponding part in the skeleton Figure 1 – A common facial UI
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
17. 17
After doing the necessary setup to skin the UI example for Blue in Figure 5.
mesh to the armature (you only have to skin
bone A), the time comes to build the facial UI.
We are going to build more bones and place
them directly in the facial UI. But first we need
to build a facial UI with panels so we know
exactly where to place the new bones to allow
the relocation and re-dimension of Blue’s
facial UI. Figure 4
shows the facial UI
Figure 2 – The character, “Blue” for Blue’s jaw, built Figure 5 – The final layout of the facial UI to controlBlu
with two mesh e’s jaw
We will build a dynamic facial UI example for objects and one text
The objects that make up the facial UI have
Blue’s jaw. Once you understand the method object.
been slightly changed since Figure 4. The
used for the jaw, you can extend it to all other
Figure 4 – The objects of outside panel (global panel) is now vertically
facial regions and build a complete, dynamic
the facial UI to control bigger to better suit the proportions of Blue’s
facial UI.
Blue’s jaw face and I’ve added one more text to better
Begin by adding an armature to the scene and identify the global panel. The next step is to
place two bones to gain control over the jaw. We only need a few objects for the facial UI. add six more bones to
Have a look at Figure 3 to get an idea of how Objects A and B in Figure 4 are planes the armature and
to place the bones. Bone A is the jaw bone configured so the user can only see their place them as shown
itself and bone B is a controller for bone A. wireframe from the front view, the view we in Figure 6. You can
Now assign a Stretch To constraint to bone A are going to be using to manipulate the facial use the 3D Cursor to
using bone B as the target. This will allow you UI. The other object is a text placed above facilitate this process.
to control the jaw with a lot of stretching and panel A to easily identify that panel, which is
the one we will use to restrict the jaw Figure 6 – A six bone setup
squashing abilities. to control the jaw panel
controller. The outside panel, or B, is a global
panel to which the movement of panel A will
be restricted to, in order to better keep track Bone A, the highlighted smallest one in the
of the facial UI. The objective is for the user to center, is the one meant to drag the entire jaw
be able to relocate and re-dimension both panel. Bone B is the controller of the jaw’s
these panels. Although in this article we only animation. The movement of this bone will be
illustrate the process for the jaw panel, you restricted to the jaw panel. The other four
can later extend it to the global panel as well. bones are placed in the corners of the panel
So go ahead and place the facial UI side by and you guessed it, each will control one
side to the face of the character. It doesn’t corner of the panel. Make sure that bone A is
need to be very close, nor too far. See the final the parent of bones 1 to 4 and also of B
Figure 3 – Jaw bones for Blue
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
18. 18
(keeping the offset between them as We can’t forget to restrict the movement of link the jaw bone in the facial UI to the jaw
illustrated by the relationship lines in Figure bone B to the jaw panel. So let’s assign bone in Blue’s face mesh, so we get some
6). another Limit Location constraint and one motion of the jaw. Figure 9 shows the bones
more Shrink-wrap constraint to bone B. Notice we need to link.
The next step is to restrict the movement of that this time the target of the Shrinkwrap
bones 1 to 4 and also of bone A to the large constraint is the jaw panel, not the global
panel, the one with the text “blue UI”, seen in panel. At this point, if you move the bones in
Figure 5. Remember that the main purpose of the system, the panel doesn’t follow along. In
this panel is to hold all the panels of the order for that to happen, we can use a
different facial regions in order to keep track number of techniques, such as Vertex Groups.
of the entire facial UI. So, no panel should go But I find it to work better using the Hooks
outside this one. We do this using two bone modifier.
constraints in Pose Mode for each of the
bones previously mentioned. Be aware of the So go ahead and assign four Hooks to the jaw Figure 9 – In this case bone A is simply going to mimicb
difference between object constraints and panel object. Do the necessary configuration one B
bone constraints in Blender. The first of the Hooks modifiers keeping in mind that
constraint is a Limit Location, to prevent the each Hook should point to the correspondent The purpose is for the jaw controller bone in
bones from moving in the Y axis (you can see bone. Figure 8 shows the Hook modifier’s the mesh (bone A in Figure 9) to mimic the
the spatial orientation for Blue’s facial UI by options. Send movements of the control bone in the facial
looking at the mini axis in me an email UI (bone B in Figure 9). We can do this by
Figure 6 but you have to be if you have assigning a Copy Location constraint to bone
aware that the orientation trouble A and using bone B as the target.
in your scene may be setting this
To wrap-up the process, you can help the
different from this one). modifier.
animator by hiding the unnecessary bones
The second constraint is a
Figure 8 – The and assigning custom shapes to the bones the
Shrinkwrap targeting to the
Hook modifier animator is supposed to control. This will
global panel object.
improve their appearance and make the facial
Remember to repeat this
Now you can drag the panel in Pose Mode UI a lot more user-friendly.
process for bones 1 to 4
and also A, so each of them using the facial UI bone system (using bone A
has the two types of in Figure 6). But there’s one object being left Conclusion
constraints mentioned. behind: the jaw text. For the text object to
follow along, you can simply make it a child of This method adds a lot of flexibility to
Figure 7 shows these common facial UIs, especially the UIs using
constraints types. bone A. But if you want you can also make it
dynamic using another bone to control it. lots of panels for every facial region. And even
Figure 7 – The types of Notice how now you can also change the more functionality can be added to it. Figure
constraints to use position of the corners of the panel using the 10 shows a final configuration result where
corner bones in Pose Mode. Next we have to the jaw panel was first moved up and then its
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
20. 20
really just 2D pictures, they are
projected onto hidden low poly 3D
geometry. The resulting character
masking and shadows add to the
illusion very effectively. With hidden
3D geometry masking the pre-
rendered background, a realtime
Alpha Polaris is a point-and- character can walk around an object
click horror adventure game as if it was really in 3D space. Typical
developed by Turmoil Games for adventure games, we also
and set to be released in June think this is one of the things we have
been very good at. Throughout the rendered separate front masks for
2011. We have been working on scenes.
the project for roughly two project, Lassi has been very effective
years now, using Blender for in training people to use Blender. With
visuals and Wintermute as a a limited initial skill set and a team
game engine. The core of the game is with no game industry background,
by - Teemu Vilén the titular Alpha Polaris, a distant oil we've also approached the project
research station in the icy wastes of very iteratively, improving the graphics
Greenland. Our team of six people set and re-doing things when needed.
out to produce a consistent, high-
Our key art assets in Blender are the
quality gaming experience with a
scenes. With indoor locations, one
special focus on the place itself. Our
Blender scene corresponds to one
setting wasn't going to be a "slog
game screen. Outdoors, there might
through different levels" world, but an
be several screens rendered from one
isolated setting full of little details. In
Blender scene. For example, our
this article, I'll discuss some of the
station from the outside is a full 3D
visual challenges we have tackled.
environment. One of the more
Making adventure games is a very art important observations we made
heavy business. To pull one off with regarding the pre-rendered scenes is
1280x800 wide-screen graphics, pre- that including dynamic elements will
rendered backgrounds and real-time make them more interesting. We don't
characters, we needed a lot of art use parallax scrolling, so we employ
different sprite animations like One would initially think that working
resources. This meant we all had to with such a well defined setting is
learn how to do modeling and snowing overlay, glow from aurora
borealis, subtle flickering on computer relatively simple when it comes to
texturing in Blender. In the beginning, rendering. However, our scenes
only our art director Lassi had screens and so on. Illusion of depth is
also vital. Since the backgrounds are contain a high number of interactive
significant modeling experience. I
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
21. 21
batch tools so post processing the sprite Character animations were pretty
animations was also possible. straightforward to do. We made one action
strip per animation in Blender and combined
The same scenes were also used for cut scene the animations in Fragmotion. Our characters
videos. The challenge was to build the are semi-realistic so the animations had to be
suspense while adding visually dynamic close to a real person’s movement. Animating
content to the game. Instead of elaborate the walk cycles was the hardest part. We
storyboards, the team prepared test animatics videoed our team walking on a treadmill for
for the cut scenes phase of the project which reference at a local gym which helped a bit.
and changing details, and are revisited in meant we already had a lot of the scenes
different parts of the game. On top of that, we done. This sped up the process considerably.
have three totally different lighting conditions: The cut scenes in the game are not very long,
morning, evening and night. Combined with but combined they total at over 12 minutes,
relatively long rendering times, it has been an so they were a big part of the project for a
arduous process. For example, if an artist small team.
forgot to render a night version for objects on
With characters, the most significant problem
the kitchen table, the error would be painfully
was getting them to Wintermute. It uses the
obvious.
age-old DirectX format for models, and the
All rendering was done with Blender’s own Blender .X-exporter didn't seem to work
renderer. We went through several other correctly. As usual, this meant some semi- I'm happy to say that there's not much to do
possibilities, but at the time none of them blind pipeline testing and frustration. If we got anymore. At the moment I'm writing this, we
were integrated well enough into Blender to to the point we could see the characters in are implementing the localization kits - the
use in a heavy project without problems. For Wintermute, they always had some lighting game is to be published in several languages
example we needed to render lots of sprite and animation related glitches in them. After by a major publisher. Our long, iterative
animations for the scenes with render layers about two weeks of testing, we ended up process seems to be coming to conclusion.
and alpha. The scene renders were all post using Gandaldf's excellent .B3D exporter, and We will definitely continue to work with
processed in Photoshop. Luckily it has good then converting them to .X in FragMotion. Blender in the future, as it's proven to have
outstanding features and flexibility for indie
The characters in the game are realtime with game development, not to mention the
3000-4000 triangles. Wintermute only community support
supports diffuse maps, so we had to bake and
paint some lighting effects on the character Writer: Teemu Vilén, Project manager of Alpha Polaris
textures. They are 2048x2048 resolution and and CEO of Turmoil Games LTD
were mostly painted in Photoshop, but
Blender’s texture paint was also used as it is
excellent for clearing texture seams with the
clone stamp brush.
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
22. 22
or other MMOs. So it’s possible to now I will not share any blender files,
build an RPG game alone with all that scripts, textures assets, so don't ask
stuff too. All Blender users know the for it. All files are done by myself so I
game Yo Frankie!, and it was the first have a lot of things to do in the next
big step in my opinion. couple of months.
Thanks to the hours they spent to If my concept works in a browser with
Hi, I am Alexander from Leipzig build such a good game like Yo the webplugin Burster for Blender I
and I had the idea to build a Frankie! with Blender, it gave me the will present it as Browser game in 3D.
game in that sunny place of energy to start my own project. As I I got a lot of ideas. I hope the
Germany. But how? First I have started it was not easy to find a good community brings the feedback and
to say I was working in the film story, so I began with drawing. After a support I need for that.
industry for one year along with couple of days in January 2011 I got
a couple of years in the direction, a short story and Deadline is around Fall of 2011.
commercials and other 3D personally it’s what I want to spend
animation stuff using tools like time on. Information about it is on Facebook
by - Alexander Weide Houdini or Maya. But one day I
thought: it’s not what I want. We live in The creative part of it is very similar to
a world with perfect animations and creating 3D animations. I played a lot
“perfect” pictures, but is that what we of games, I saw a lot of movies and I
want? We want to have entertainment, was looking at what the community
we want to do anything, we want to does. And the community does a lot,
build our own worlds. For only one but most don't reach the end of their
person it is an impossible task to build projects. Why? Is it not possible to do
effects driven movies like District 9 or it? Is Blender not the right tool? I don't
Battle L.A in a short period of time, know, but I will complete the end of
without thinking of budgets. So it was my experiment and I hope it will
my decision to build one world - inspire other artists around the world
alone. I will see how far I can get, how to do same with Blender!
much I can reach, and how fast I can
reach my target. In the next few days I will start a
Facebook site to share the project with
I am an artist rather than a the community. If the community
programmer so Blender was the best stands behind me, I will gather more
choice to help me meet my goals. The power to go forward. If the
new version has a very high potential community is not interested in that, it
to build games with all that graphic could be a hard way for me maybe. So
stuff we know from World of Warcraft I hope I will get a lot of feedback. For
www.blenderart.org Issue 33 - Everything but a Kitchen Sink
23. 23
In 2005, after learning some to take it back ...
rigging/skinning stuff, I began to work
on shapes (called rvk's at that time) for The short will be (hopefully)
facial animation of my characters. humorous, which is why I decided to
Today this is still the area that model some toony and quite ugly
provides me the most pleasure. I love characters, especially the young girl
the way it gives life to characters and who is supposed to be pretty (she's a
Hello, my name is Stephan this is where the acting job begins. Beauty Miss).
Rodriguez and I'm a a French
3D artist. I live in a very So that year I produced my first short
attractive and sunny town in animation with a lipsync job to
the south of France called celebrate the
Montpellier with my wife and 2
children. 10th Blender anniversary. This was
called "Happy Birthday Mr Blender"
Since I began 3D back in 2001 when I and is visible on my Youtube Channel.
by - Stephan Rodriguez discovered Blender, my goal has Tony Mullen, the author of
always been to do some animation "Introducing Character Animation
one day. At that time I was amazed by With Blender" even gave me the
every 3D short film I could see but I pleasure of including it on his DVD and
never imagined I could do the same a I was very proud of this. One of the first things to do besides
few years later. I began like every modeling was the storyboard. This is a
other Blenderhead, reading every During all these years, I had a couple very important thing because you
tutorial I could find on the web, asking of story ideas running in my head. constantly refer to it to know what
questions on forums like Elysiun (the Then in 2009, after improving my scene to do, what the shot will require
ancestor of Blenderartists) or a French animation skills, I decided to step in terms of animation, characters, set,
forum called zoo-blender where I forward and lay down my story on props etc. I created a personal
used to spend a lot of time. paper. On February 2nd of 2009, my
short animation project officially
I'd say it took me about 4 years to feel started.
really comfortable with Blender, just
enough to make some decent looking "The Crown" tells the story of a young
still pictures. I had learned to model, girl who has been revered as a Mini
texture, light and was beginning to Miss in her town. What she doesn't
produce some very basic animation know is that the crown on her head,
along with some particle and camera found 50 years ago in the countryside,
work. was actually lost by some distant
visitors and they are here again today
www.blenderart.org Issue 33 - Everything but a Kitchen Sink