Tornado 3D v 1.5 - Summary of new features
Raytracing
Tornado3D now has full recursive Raytracing for reflections, refractions and shadows.
Raytracing is a well known technique that allows to effortlessy simulate shiny and transparent materials. It is
however very slow and its usefulness in real world production is very limited.
Tornado3D overcomes this limitation allowing to precisely control when and what to raytrace.
The Raytracing subpanel controls Recursion Depth -that is the amount of bounces between the light rays- and the
separate application of Reflection, Refraction and Shadow rays. Each surface can be set to Raytracing mode, and
entire objects can be excluded from Raytracing computations with the appropriate Unseen by Rays control in Objects
Info Panel.
Raytracing also works along with the reflection and refraction mapping techniques of Tornado3D
allowing seamless mixing of the two methods. Any kind of object can have Raytraced effects, but please note that
only polygonal models can appear in Raytraced reflections, refractions or cast Raytraced shadows. In addition to
this, Raytracing is unable to understand Interpolated Shading, and will only show full smoothed or full flat objects.
Reflection, refraction and shadow mapping can instead properly render any kind of surface.
BACK TO SUMMARY
Field Rendering
Tornado3D now allows to produce animated sequences with the so-called "field-rendering".
Field rendering is a technique that takes in account the way images are shown on tv-sets. Each image is infact
shown as two smaller, and slightly displaced, sub-images called fields. Recombining those two images to build a
full frame is called interlacing.
Interlaced pictures look jagged on computer screens and on film, but produce highly smooth
animations when reproduced on tv-sets. This happens because there are twice as much informations about scene's
movements. There is a trade-off, however, in that the vertical resolution is cut in a half.
Tornado3D switches to field rendering when the Field Rendering checkbox in Render Panel
is turned on. The Field Rendering button allows to control both Field Swapping and Field Mapping. Field swapping
inverts the two fields, there is no clear standard about how field have to be interlaced so each recording device
uses its own ordering. If rendered animations looks mangled and jittering, the swap fields checkmark has to always
be turned on when recording with that device. Field mapping controls how image sequences are to be handled. If
it is turned off, the mapping of image sequences will happen as usual, with the same picture for both the odd and
the even field. This may lead to unnatural and jerky motion because the 3D animation runs twice as smooth as the
mapped sequence. Turning field mapping on will use two images for each frame, one in the odd field and the following
in the even one. If the sequence is not already computed on fields, this will produce smooth motion but it will
also make the scene look accelerated.
BACK TO SUMMARY
Full Virgetm support
Tornado3D has been the first real-world application with support for the Virge 3D accelerator
found on the Cybervision3D graphics card manufactured by Phase5. The Virge chipset allows to free the processor
from the burden of filling and texturing polygons, thus speeding up interactive color shaded preview. In addition
to this, since the Virge chip resides on the graphic card, the amount of data to be transferred thru the slow Zorrotm
bus is drastically reduced: instead of transferring hundreds of kilobytes, just a few bytes describing the polygon
coordinates are provided to the Virge engine, resulting in an even more sensible speedup on ZorroII machines.
Following the release of "cgxvirgin3d.library" v2.0 and higher, Tornado3D's
Virge support is no more experimental. All known problems and limitations are a thing of the past. Now texture
maps are automatically managed by Tornado3D and stored in the 1MB of multimedia memory provided by the CyberVision64/3D
(this needs cgfx libraries v69e or higher). This results in another huge speedup when doing mode 6 (texture) preview:
even with texture antialias it is much faster than direct 68060 preview on non-3d gfx cards. Now the Virge screen
looks like a normal Tornado3D screen and the full-screen perspective view is shown without the scaling of software
preview at full resolution, still faster than software preview on a quarter of screen. Now particles, 3d lines,
full transparency (not just fixed like in software), backdrops, frontdrops and bilinearly antialiased textures
with mattes are shown in Virge preview much faster than software rendering. In addition to this the Virge engine
also does dithering to improve the apparent number of colors. Warning: Virge acceleration can only be used in 15
bit modes upto 1024x512 pixels; higher resolutions run out of on board memory.
BACK TO SUMMARY
Deformation Cages
Tornado3D always had a vast array of deformation tools, but they mainly worked at vertex
and face level, allowing the user to permanently modify the shape of an object. Deformation Cages, a technique
similar to that of Free Form Deformations, work in a completly different way. Instead of deforming the object,
a much simpler reference shape gets deformed, and than applies its changes to the real object with a smooth, spline
based, quality.
Deformation Cages are activated adding a Cage object from the Spline menu. This object
is invisible to the rendering engine, its points can not be deleted or added and has really no influence over the
scene unless it is parented to another object. After parenting it with another object using Amiga-G, its position,
size and alignment change to encompass the volume of the object it is parented to. No shape change is visible yet,
because the Cage is in its Rest Position. It can be noted, however, that redrawing and previewing of the scene
get a little slower since the Cage has to apply control each vertex of the original object everytime it is redrawn.
Selecting the Cage and entering Point Mode allows to modify the Cage's shape, and the
changes applied to it are reflected in real time as smooth changes over the parent object's surface (this only
works with move,scale and rotate; bend,taper, etc. do not allow to have realtime Cages preview but their effect
is applied anyway). Once the object has changed shape the way we want, the transformation may be fixed in place
by ungrouping it and accepting the changes. Since the deformation is independant from the objects representation,
it is even possible to deform a simple object, with few polygons, and then increase its detail with subdivision
or metameshing: the deformation will be applied to the new structure at full precision.
Cages are also useful in animation, if the deformation is not fixed. Keeping the two objects
grouped and creating auto-morphings for the Cage, allows to animate complex deformations with minimal memory requirements,
since only the 64 points of the Cage are stored even if the parent object has thousand of vertices.
BACK TO SUMMARY
Displacement Maps and Textures
Tornado3D now allows to deform objects with the so called "displacement" maps
and textures. Displacement maps and textures (Displacements, from now on) do not influence the properties of a
material, like color or brightness, but its geometry. With displacement maps an object can be made ripple, pinch,
stretch, or it can be imprinted with an image, creating for example a fractal-like landscape.
Displacements are an object modifier, and are only computed when applied to the DEFAULT
surface, the omnipresent surface any object has. Displacements are not restricted to the DEFAULT surface, however,
they extend to the entire object. Displacement can be animated, and applied with any projection method.
When dealing with bitmap images, the red component of the material represents the amount
each object's vertex gets displaced along its normal direction. This is a signed displacement by default, that
is, middle gray is considered zero displacement, black full negative and white full positive. This can be changed
with the Depth Slider, that for displacement maps does only control the displacement direction; leaving it at the
middle default provides the above beahviour, any amount towards the left produces only positive displacement and
any amount towards the right produces only negative displacement.
When dealing with procedural textures, the behaviour is similar, and it works with any
texture, not just with those that do "bumps". Intensity, however is controlled both by the per-channel
alpha value as set in texture panel and by the global intensity slider.
In order to produce continuous, non accumulating, transformations it is necessary to create
a reference geometry morphing just like for force fields.
BACK TO SUMMARY
Glow effect
The much requested ability to make the objects have an "aura" or "Glow"
is now present in Tornado3D with a superior, surface based, method that is many orders of mangitude faster than
other software's one.
Glow is a global effect, and is controlled from the World Panel; here, both the Glow Radius
and the Glow Intensity can be adjusted. Glow radius controls the area of application of the effect, and is expressed
in pixels. Increasing the radius leads to longer rendering times, but the slow-down is not exponential like in
most other implementations, but linear. This means that larger radii will make Tornado3D even faster in comparison.
Glow intensity controls the amount of lightnening each pixel is subject to. The default percentage of 50% provides
reasonable effects for things like shiny text logos, neon lamps, or warm candle lights. Higher levels can be used
for the simulation of lava, spaceships engines and explosions. Both glow radius and intensity can be animated over
time.
Glow is applied per-surface, and not with a simple color-keying. This means the real surface
colors will be glowed without any antialias problem. To activate Glow for a surface, click the appropriate checkbox
gadget in Materials Panel.
BACK TO SUMMARY
Alpha Channel
Alpha Channel support has been in Tornado3D since day zero, but Alpha information was
only used internally, and there was no way to export it in a meaningful way. Now Tornado3D allows to optionally
save the Alpha Channel during rendering. The Alpha Channel is a separate, gray, image that represents the levels
of transparency inside the scene.
Alpha Channel is primarily used to allow for seamless composition of rendered and real-world
scenes, but it can also be used for multipass rendering and layering. Tornado3D's Alpha Channel represents black
as full transparency and White as full opacity, that is "solid".
Please note that no Alpha Channel information is saved for Lens Flares or Glows. Regardless
of the incorrect behaviour of other softwares, Alpha Channel must not be computed for additive effects, since those
are summed over and not composed with existing data. The correct way to matte-in a Flare or a Glow is to render
them on black and then use an Absolute Add image processing operator.
BACK TO SUMMARY
Blobby Particles
Particle Systems can now be turned into Blobby objects with the Blobby checkbox inside
Particles Panel. The particles system behaviour remains unchanged, but the particles's points are turned into metaballs
at render time, and this allow to produce fully solid and three dimensional effects.
Blobby particles do of course slow down rendering, but the number of particles required
for a realistic effect is also much lower than with single-point particles. Effects such as Lava, Dense Smoke,
Water Flows and Jelly are much easier and require much less memory.
BACK TO SUMMARY
Broadcast video encoding
Broadcasted video has to comply with several restrictions defined by the video norm in
use. Failure in doing so produces artifacts, distortions, and generally unpleasant pictures. If your TV set shows
a programme with hot colors, washed out areas and rippling effects near bright objects, it has not been properly
encoded for broadcasting.
Now Tornado3D allows to optionally encode and make broadcast-safe, any image before saving
it, in a completly automatic way. Images can be made safe both for PAL and NTSC video norms, and the Chroma and
Composite level thresholds are completly controllable. It is also possible to force the encoder to act on Luminance
or Chrominance, thus allowing to preserve the overall brightness of the scene or the individual tints. Broadcast
encoding assumes a Gamma setting of 1.0 in Tornado3D and compensates for PAL and NTSC Gamma norms. Broadcast encoding
values are expressed in IREs.
BACK TO SUMMARY
Particles Ageing
Particles can now optionally be "aged" when rendering, making them more and
more transparent as time passes from their initial appearance on screen. This is turned on with the "Fade"
checkbox in particles panel, and allows to do more easily explosions or other dampening effects, without fiddling
too much with Gradient and Firball textures.
BACK TO SUMMARY
Extended Automorphing for keyframeless animation
One of the most important features in Tornado3D has always been the Automorphing. This
allowed to make changes during an animation and have those changes remembered without complex "spreadsheet
like" setups in dedicated panels. It was still necessary to record keyframes and use the (R) and (C) gadgets
in panels like Lights, World or Materials.
Now those keyframe gadgets do not exist anymore, and every changed performed in World,
Material, Camera or Lights panel, as well as object tracking is automagically remembered and stored without the
need for an explicit user command. The only thing that still needs to be keyframed manually are changes to objects
position, alignment or size.
BACK TO SUMMARY
Smart Hierarchies and Poses
A new kind of automorphing is now possible: Hierarchical. In previous releases Hierarchies
(or Groups, a term that is being phased out) were more a convenience to the user than anything else. Now complex
linked structures, like human bodies or sophisticated robot machinery are automatically handled by Tornado3D with
a new class of morphings, the Poses.
Creating a Pose is as simple as going into animation mode and selecting the child node
of a group: trying to move or rotate any children will pop-up a requester similar to that for geometric morphings
that will allow to create a Pose. Any modification done to the hierarchy will be remembered and smoothly animated,
just like a geometric morphing. Of course this also works along with Inverse Kinematics and poses created with
it are no different than forward (standard) kinematics ones.
Unlike other programs, where creating hierarchies resets the transformation of children
nodes and requires lenghty adjustments or carefull initial planning of relative positions, Tornado3D will smartly
determine the wanted relationship between objects and enforce it.
BACK TO SUMMARY
Pose Manager with built-in Over and Under Shoot control
As powerful as the automorphing concept is, there are some situation when some bookeeping
is needed. We may have created many poses (here we use the term poses loosely, referring to Properties, Geometry
and Hierarchy morphings) that are named as the frames they were created in, and want to assign more descriptive
names. Or, we may want to delete some old poses not useful anymore. Enter the Pose Manager: this panel operates
on the three fundamental kind of poses allowed in Tornado3D and provides complete control over creation, removal,
renaming and replacement of poses.
The Pose Manager is accessed thru the Animation Menu and tries to auto-detect the most
likely kind of pose you are about to work with. The kind of pose can always be changed by hand using the topmost
cycle-gadget labeled "Act On". Immediatly below, the "Current Pose" cycle gadget shows which
pose is active on the current frame (if any). The associated "Pose Name" string gadget allows to rename
the pose. This gadget is also employed when creating new poses to define the new pose name. The "Remove"
button erases the current pose from memory and the cycle gadget on its side determines how this removal affects
existing keys, removing them or leaving them but setting the Pose to the Original one.
The area below the separator bar is related to the most important feature of Pose Manager,
the Pose Mixer. Given two different poses, the Pose Mixer allows to create a third one either interpolating or
extrapolating the first two. The first pose is always the "Current Pose" and the second is the one specified
with the "Mix With" gadget. The row immediatly below controls how the mixing is performed. The cycle
gadget controls "Mix","Undershoot" and "Overshoot". Mix is nothing more than a simple
linear interpolation between the poses. Under and Over shoot allow to extrapolate either before the first pose
or after the second. We use the before and after terms because it is common practice to use this kind of transformation
over animation poses, for example to create realistic anticipation ad exageration effects for character animation.
The following gadgets control the exact amount of interpolation, and three preset modes, corresponding to 1/4,
1/2 and 3/4 are provided. One quarter and three quarters are commnly used for Over and Undershooting. Once the
desired amount of transformation is set, the Pose can be created with the Make button; once again, the cycle gadget
provides control over the kind of creation. If set to "Pose & Key" both a Pose and a Keyframe set
to it will be produced. If it is set to "Pose only" no key will be produced. This second mode is expecially
useful when producing several poses in a row to be spread at a later time.
BACK TO SUMMARY
Spline Ease In-Out control over animation
Traditional animators often use the "Ease" term referring to two quite common
concepts: acceleration and deceleration. In real life, things do never start at full speed or stop in place. Friction,
gravity and many other forces controls the behaviour of objects. Easing is a quite established method to simulate
this behaviour in animation. By Easing In a keyframe, the object will start slowly and reach full speed at the
ed of the key. By Easing Out the same key the reverse will happen.
Easing is controlled with the two new "Ease" controls in Keyframe Panel. By
making both values the same it is possible to make an object reach full speed in the middle of a keyframe, or conversly
to make it slow down and then gain speed again. Easing works spacing closer and closer the keyframes for the object,
and has a dramatic impact over motion blurred animation.
BACK TO SUMMARY
Real-world camera model
The simplyfied camera model used in 3D graphics usually only allows to define the Field
Of View or Zoom, and ignores completly any other real-world parameter. Tornado3D makes one more step further towards
the full integration of real world and computer generated imagery, allowing to define a syntetich camera that acts
and can be controlled exactly like a real world one.
The currently defined real world camera in Tornado3D is a 35mm SLR, Single-Lens Reflex.
It is one of the most popular cameras around and uses 36x24mm film. Every calculation is done with respect to this
camera, and more camera definitions will be added in the future. Thanks to the real world camera model, Tornado3D
now understands and makes use of concepts like "Focal lenght", "Hyperfocal distance", "Far
and near focus distances", "Aperture size" and "circle of confusion". A good book about
photography, which is mandatory for succesful computer graphics anyway, is surely a better place where to understand
the exact meaning of those terms.
BACK TO SUMMARY
Startup tips to ease learning
Startup tips are a great way to learn how to use a software without having to always wade
thru a boring manual. Upon startup, Tornado3D will present you with a Tip, a short suggestion on how something
can be done with Tornado3D. If you don't want to see tips at startup, simply click on the Show tips at startup
checkbox gadget to turn it off. You can turn it on later reopening the Tips Panel with the HELP key. You can also
browse thru the tips with the forward and backward gadgets. More tips files will be made available on http://www.tornado3D.com.
BACK TO SUMMARY
New spline primitives
Five new spline path primitives are provided, Square, Circular, 8-Gon, Bulge (flower)
and Star. Those are all closed paths with the same number of points and make for great building blocks for the
new spline skinning function. They also show quite clearly how spline Tension, Continuity and Bias can produce
straight or curved segments making so called Edges, or Lines completly obsolete.
BACK TO SUMMARY
Spline Skinning
Spline skinning is a funamental tool for building so-called cross-sectional models. A
cross-sectional model is one that can be easily decomposed into sctions, or slices. A good example of a cross-sectional
model is a rowboat, where wood strips are skinned over a cross sectional structure.
Spline skinning allows to select any number of splines, open or closed, of any desired
shape as long as all the splines share the same number of points. When the desired number of splines is selected,
Multiply/Skin builds a polygonal skin around those splines, connetcing them into a single smooth surface. The polygonal
resolution is taken from the first selected spline, and the skinning order is the same as the selection order.
Note also that the splines are not modified in any way and can be immediatly reused to build another object.
Using splines with the same number of points but different Tension, Continuity and Bias
values, allows to smoothly produce bulges, dents and even to morph straight flat surfaces into curved or inflated
shapes.
BACK TO SUMMARY
Imaginetm & Lightwavetm objects import
The ability to exchange data with other applications is one of the most critical tasks
for any computer application. 3D software data exchange is even more difficult because only a small portion of
data retains a meaning when translated from the creator application to another one. The industry-standard exchange
file format is DXF - 3DFACE as popularized by Autodesk's AutoCad and Tornado3D supports it since release 1.0.
Many users asked for more file-format exchange facilities, so both Imagine and Lightwave
import functionalities were added. The reader software is accessed with the Import menu, and automagically detects
which reader is appropriate for the requested file. It also tries to translate as much data as possible, retaining
a meaningful representation but this is not always possible. Tornado3D loads Imagine Groups and Objects, preserving
shape, size, alignment, and the basic surface properties. It also loads Ligthwave objects preserving shape, most
surface properties, and surfaces assignment. The limitations are detailed below.
Imagine objects are composed by unstructured triangles. This means that it is unlikely
that the normals will be correctly oriented once they are loaded into T3D (or any other application that uses structured
normals). T3D tries to correctly align surface normals, but this is not always possible: if objects appears to
miss faces, those faces are probably backfacing, and a Flip Normals command is necessary to display them properly.
In addition to this, Imagine allows for so-called T-edges, that is, it does not need properly strutured polygonal
meshes. Some commands like Metameshing and Booleans will fail if applied to objects with T-edges.
Lightwave objects import is mostly immune from the above mentioned problems, as its objects
are required to have both structured normals and no T-edges, but it allows for so-called Concave N-sided polys.
Those are polygons with an arbitrary number of vertices, eventually with holes shown as Bridges, that is as thin
lines that cut a side of a poly to allow for an hole, like in a letter O. Tornado3D can not read in this kind of
polygons and will ignore them. If your object appears to miss some faces, load it into Lightwave's Modeler and
use the Triple function to enforce it is made entirely by triangles. LW objects can also have more than the 32000
points, edges or faces allowed in Tornado3D: if so conversion will abort early.
BACK TO SUMMARY
Object auto-saving for projects
Object auto-saving is a significant improvement in the way objects and projects are stored
by Tornado3D. Now the Objects directory is no more the only place where objects are stored, instead it has become
a repository of useful re-usable objects. For each project, a system directory is created into the Projects directory,
with the same name as the project andd $TP appended to the end. This directory holds all the objects needed for
the current project, and everytime the project is saved, all the objects are saved along it. The Camera and all
the Lights are also stored as objects, allowing to save things like projector lights settings.
This way it is much easier to keep projects organized, and to store them away once the
project has been completed. It is also impossible to mistakenly save a project loosing some objects because they
were not saved. Tornado3D tries not to allow to objects with the same name to appear in the project, but this is
not always possible, so a rename check is performed everytime the project is saved, and adds numerical extensions
to the names of already existing objects. If Tornado3D warns that some objects were renemed when a project is saved,
it would be a good idea to check for named look-at constraints, as those may become invalid after such a renaming.
BACK TO SUMMARY
Fast menus
Fast menus are another great timesaver, expecially when dealing with the first, experimental
steps in the creation of an animation or image. The six most recently loaded objects and projects are shown in
a submenu next to the Open Project and Load Object menuitems, and this allows to reload them immediatly without
having to browse thru a file requester. Along with autosaving, fast menus should let the user to concentrate on
creative flow instead of having to do bookeeping of directories and paths.
BACK TO SUMMARY
Graphic preview of animation splines
The Keyframe Info panel now allows to see a preview of the Tension, Continuity, Bias and
Ease parameters and to manipulate it interactively with Increment Buttons. Once clicked, the increment Buttons
add or subtract a value proportional to the amount of mouse movement to the left or to the right, and this change
is also shown in the preview-area. The dots represents frames of the animation, so a closer spacing means that
the object travels slower and an high curvature means that the object will approach the keyframe along a loose
curve.
BACK TO SUMMARY
Smart object selector
When many objects are visible in the scene, it becomes harder to find the exact center
of the one we want to select, and this slows down any operation. The smart object selector turns the mouse pointer
into a selection cross-hair when it is over a selectable object. After a few days of use, you'll become confident
that you can actually click when the mouse pointer changes and obtain the desired result. It is very hard to explain
how faster this allows to work, so just try it and you'll find that your mouse movements will become more gentle,
relaxed and work will be much easier.
BACK TO SUMMARY
Improved preferences panel
The Preferences Panel has been expanded and now allows to define all the ToolTypes that
previously had to be set by hand. It also allows to change all the default paths to other temporary ones, simply
by changing the path and then clicking on Cancel and not on Save. The changed paths will be in effect until T3D
is restarted.
BACK TO SUMMARY
Rendering needs less memory
Rendering in accumulative modes now requires two bytes less per-pixel. This results in
a saving of over 600 KB for a 640x480 image, without any slowdown or loss of image precision.
BACK TO SUMMARY
Color shaded boxes on preview
Bounding boxes for boxdrawn objects are now computed on the fly and shown in color shaded
preview modes (modes 3 thru 6). This allows to speedup even more animation playback, and on Virge-equipped machines
allows to preview most animations in full 25/30 fps.
BACK TO SUMMARY
More controllable motion-blur
Motion-blur is now more controllable, with the addition of the Motion Blur Panel in the
Render Panel. The Max blur lenght setting determines the amount of streaking the object will produce and allows
to fine tune the motion blur effect, with long streaks even when there is not much movement in the scene.
BACK TO SUMMARY
Detail control over rendering
There is a new Image Integration control in Darkroom Panel: Detail. Increasing it produces
sharper images, and decreasing it makes them more blurred. This is not a simple Image Processing control, but works
inside the integration phase to allow more control over the final result. Personal tastes aside, the default value
is appropriate for animations to be recorded on video, while higher vslues may be useful for images to be printed
at high DPI resolutions.
BACK TO SUMMARY
Improved reflection & refraction mapping
Reflection and Refraction mapping now share the same code and are much improved in terms
of final rendering quality. It is often impossible to distinguish between a Raytraced and a Raymapped scene, but
care must be taken in that Raymapping can not be applied to flat mirrors, that is to objects without thickness
in any of the three axes.
Refractions will also appear to be computed with a much higher refraction index, so it
would be wise to lower the IOR setting for refractions to be Raymapped. It is also possible to control the amount
of memory and time spent Raymapping, lowering the MapSiwe gadget found in Materials Editor right below the Map/View/Trace
cycle gadget. The default value of 256 produces six maps, and each one requires 256*256*4 bytes. Lowering the size
produces coarser maps that are computed much faster. As a general rule, Map Size should be at least as large as
the projected size of the reflective/refractive object when rendered.
BACK TO SUMMARY
Improved lights panel
The lights panel has been reworked and now automagically defaults to the first selected
light, speeding up editing in most circumstances. It also allows to set Shadow casting, volumetric and projective
attributes for point lights.
BACK TO SUMMARY
Improved and more controllable lens flares
Lens Flares are now much more controllable, and the Red Ring, Red Glow and Noisy Streaks
attributes can be turned on or off separately. It is also possible to directly set Flare Size in the panel, and
to animate this size value for explosions or photon torpedo effects.
Lens Flare now respect Alpha Channel in any rendering mode, becoming dimmer and dimmer
as they are rendered behind transparent objects. Lens Reflections code has been toned down to allow for a less
colorful appearence.
BACK TO SUMMARY
Auto-convergence of Inverse Kinematics
Now Inverse Kinematics honours the new Link Poses and its convergence code has been reworked
so that it always shows the current position of the End Effector without requiring the user to click multiple times
to refine the IK solution found. This is a little bit slower but should be much more intuitive.
BACK TO SUMMARY
Improved depth of field
Two new Depth of Field modes have been added in addition to the original method found
in previous releases of T3D. The new Depth of Field Panel inside Render Panel allows to choose between the Jittered,
original, method and the Fast and Accurate Focus that depend on the new camera-model.
The Focus modes blur the resulting image in accordance to the distance from the camera
to the point onto the Image, and produce visually superior results, but are unable to jitter objects behind transparent
surfaces. They are preferable, however, since the visual result is pleasing even in Low quality while the Jittered
method requires high Quality settings, and also animation is much more intuitive. The accurate Focus method is
many, many times slower than the Fast one, and is not recommended to use it. The only real difference is that objects
over a black, infinite, background will be defocused incorrectly with the Fast Focus setting. Since most DOF scens
have a definite background, like the walls of an interior room or the long-distance mountains of an exterior scene,
this is a limitation that will rarely if ever be encountered in real-world productions.
Animating the camera parameters automagically animates the Depth of Field effect.
BACK TO SUMMARY
Spherical volumetric lights
Tornado3D does real volumetric lightning computations, with full density and integration.
In previous releases volumetrics were restricted to SpotLights because those were the most important lights, but
the code is general enough to allow for any shape. Now Point Lights can become volumetric as well, and can compute
textures (not brushmaps) inside their volumetric space: this is called Hypertexturing. No shadow-casting support
is allowed yet, since Raytraced shadows are too slow for volumetric computations in point-lights.
BACK TO SUMMARY
Realtime memory-meter
Now Tornado3D shows in realtime the memory usage, updating the status line in the lower
right of the screen so that the user can be warned when memory falls below a safety limit. The memory readout inside
info panel has been removed and info panel is now only used for credits and copyright.
BACK TO SUMMARY
Wave 3D procedural texture
A new 3D procedural texture has been added, Wave. Wave is expecially useful for displacement
mapping, and produces concentrical waves along the three axes. It does also do all the standard application methods
and Bump Mapping.
Wave parameters are animation time and Distub, that adds a noisy displacement to the waves.
A Disturb of 0 produces perfect sinusoidal waves, while a Disturb of 1 leads to rippling, chaotic patterns. Used
along with a projector lightsource, Wave produces interesting caustics-like effects.
BACK TO SUMMARY
Saver for RAW images
A new save format is allowed, raw, mostly useful for game-developers. It saves the image
as a dump of ARGB data, in top to bottom, left to right order, without any other information
BACK TO SUMMARY
Plus many more enanchements...
Apart from plain bug-fixes that have made the software much more stable, a lot of things
have been improved and extended in v1.5.
- Merge, join and replications are now much more precise in computation of points;
- Extrusion is now undoable;
- Many more controls have been added to prevent the user to do something he or she would
regret later (like adding points to an object with animated Morph poses);
- Spline computation always happens with powerful Arlenght parametrization, and spline
extrusion now uses incremental techniques to avoid unwanted twist;
- Non-spline extrusion lenght is now expressed as a percentage of the Z axis, making its
use more intuitive;
- The Find Object panel shows, and allows to select Hierarchies;
- There is a warning before closing Projects, and opening a project does not automatically
replace the existing one anymore;
- The Make Surface panel now doubles as Assign surface, allowing to set selected faces
to existing surfaces or to create any number of new surfs;
- Many Panels and operations now perform automatic redraws;
- The % key allows to mirror an object along the active axes;
- Pasting objects increments their names;
- Pick Concave picks non convex polygons to be triangulated;
- Pick culled picks backfacing polys as seen in preview window for easy flipping of their
normals;
- DXF loader reads LAYER entities as surfaces;
- Almost every message, gadget and string is translated in supported languages;
- Adaptive grid uses an even size that makes it more useful to people used to fixed grids;
BACK TO SUMMARY
|