Cinema 4D Release 19: Departure on several fronts

For normal users, the core of a software programme processes the function calls in a rather hidden manner. These can be modelling functions, rendering algorithms or import and export functionalities. You might think that once these functions have been programmed, they can stay that way for all eternity. Unfortunately, this is not the case. Certainly, some users of other 3D packages still painfully remember how hesitant the transition between 32-bit and 64-bit was in some cases, or how frustrating it can be when functions are not able to utilise all processor cores. However, the core not only has a major influence on the performance of software, but also on its growth potential. A flexible and modern programmed core makes it easier to implement new functions, port to different hardware platforms and generally maintain the software. Updating the core is therefore an investment in the future of the software.

However, due to the complexity of software such as Cinema 4D, which has grown over the years, such a task cannot be accomplished overnight or with a single update. Therefore, parts of the core have already been replaced in previous updates. Release 19 continues this in the area of the modelling core and especially in the parts of the program responsible for file handling. As a result, changes that affect the user are now visible for the first time.

Feature guide

As will become clear later on, the numerous new features are scattered throughout the software. If you are initially looking for the new features in order to familiarise yourself with them as quickly as possible, you can now rely on a highlight feature, a coloured highlighting of icons, menu entries and parameter names, wherever new features are hidden. If a new function has been called up at least five times, the highlighting disappears. However, this fading out can be prevented via the programme settings and highlighting can also be completely omitted. You can also choose between the new features in Release 18 or even options for loading your own highlighting. This could also be useful for plug-in developers, for example, who want to inform their customers about new features after an update.

Higher, faster, further

The bitmap file handling has been completely revised and, in addition to faster loading of images, image sequences or videos, also offers some new formats or extended options for the already known file formats. For example, .tiff images can now also be saved with lossy or lossless compression. When loading, alpha channels integrated into videos are recognised correctly and CMYK and YUV profiles are now interpreted correctly. Non-square pixel aspect ratios can be saved with bitmaps and reloaded from there. This also applies to saving renderings from the image manager. Textures that have already been loaded can also be subsequently assigned an individual colour profile.
Some file formats, such as .tiff, OpenEXR or .dds, now support multi-pages, a procedure that dates back to the fax age, in order to be able to store several pages or images within a single file. External help tools such as IrfanView (for Windows, www.irfanview.com) or GraphicConverter (for Mac, www.lemkesoft.de) enable the creation of such files, e.g. from multi-page .pdf documents. New formats available are ICO (Windows Icon), CUR (Windows Cursor) and MPO. The previously read-only .dds format can now also be written.
There are also small but significant additions to 3D file formats that have been available for some time. For example, object selections can be created before exporting to .fbx format in order to save only certain parts of a project. Global coordinates can optionally be used instead of local values. If objects with the same name are contained in an .obj scene, these can be automatically welded into a geometry during import.

However, the most interesting new features are certainly the Alembic format. If a file only contains packets for each frame, a new option enables the interpolation of intermediate images in order to be able to display motion blur correctly during rendering. Information such as speeds, colours or other vectors that could not previously be interpreted directly by Cinema 4D from the Alembic file can now be imported as vertex map tags so that they are not lost when transferred to third-party programs.
Of particular interest, however, is a new option for morphing animated objects, i.e. Alembic animations in which the number of points and point sequence of an object remain constant, into normal polygon objects without losing the connection to the Alembic file. The connection is maintained by a new Alembic tag. This results in exciting new options. For example, such an animation can be transferred very easily to any polygon objects with the same number of points. In addition, familiar polygon selections or vertex maps can be created to facilitate the assignment of materials, simulations or deformations, for example.

Hot shortcuts

So-called hotkeys make daily work easier, as they generally save mouse movements or mouse clicks. Cinema 4D has always made special use of Shift, Alt and Ctrl when creating selections, switching views or grouping objects in the Object Manager. Now further combinations have been added which are evaluated when a new object is called up. For example, holding Shift Alt causes a copy of the selected object to be assigned to all selected objects – very handy when assigning deformers to several objects at the same time. With Alt Ctrl, the called object automatically becomes the top parent object of a group of the selected objects. For example, several splines can then be grouped directly under a new extrude or loft object. Ctrl or Shift Ctrl directly creates a number of copies of the newly called object adapted to the current object selection and arranges them above or below the selected objects.

Directional

The first changes to the modelling core are already visible on the user interface of Release 19, particularly in the commands for aligning or flipping the surface normals. Up to now, these commands have struggled somewhat when dealing with N-Gons. For example, the polygons used within the N-gons could be recalculated or the normal on curved N-gons was not calculated correctly. These problems are now a thing of the past. The functions have also been extended so that the information stored in normal tags can also be updated. Normal tags are regularly created when importing data from CAD applications, for example.
With the plane section tool, a uniform arrangement of sections along an object, world or camera axis can now also be realised.

Less is often more

Photogrammetry is on everyone’s lips and is also finding its way into Cinema 4D through the back door of the Motion Tracker, as we will see in a moment. As fascinating as the resulting geometries can be, their meshes are also complex and often unnecessarily dense. Whenever a retopology is not worthwhile, e.g. because no deformation is required, a moderate reduction of the polygon density is sufficient. A deformer was previously available for this purpose. This has been replaced by a generator in Release 19, which has several advantages. For example, several objects can now be reduced simultaneously as a group or merged into one geometry. In addition, the algorithm has been modified in such a way that in many cases only a pre-calculation needs to be waited for. Once this has been completed, the intensity of the reduction can be controlled in real time via a strength slider or via specific specifications for the number of points or polygons. We can therefore see the result as soon as we change these values, which naturally makes the process much more pleasant.

Another problem with polygon reduced scans arises when viewing the UV coordinates. These are usually divided into many rather small islands, depending on the number of photos used for the scan and the complexity of the shape. Although UV coordinates can also be transferred to the reduced number of points, this often leads to smeared textures at the edges of the original UV polygon islands. A new option also eliminates this problem. The original dots at the edges of the UV islands are retained even after the reduction, and only the areas within the UV islands are divided more coarsely.
Another option ensures that open edges at openings or boundary edges of the geometry retain their shape. A square plane then retains this outer shape even after reduction.

Rendering? I don’t need it!

We all know how annoying waiting for images and especially animations during rendering can be. Any acceleration of this process is therefore most welcome. However, we must not ignore the question of whether ray tracing is really always necessary. Surface shadows, caustics, translucencies, transparencies with refractions and global illumination are certainly nice to look at, but they are not required for every project. Especially when it comes to more illustrative representations or mainly simple metal and plastic surfaces, the OpenGL representation definitely offers potential. An envious look at current game engines shows how far the display quality can now be optimised there. The difference to rendered results is hardly noticeable, especially when you consider that we are dealing with real-time images. Of course, there is a corresponding optimisation and pre-calculation of shaders and baked textures behind this, but the trend towards more and more VR, AR and interactivity leaves no other choice but to deal with this topic anyway.

Since the last version, Cinema 4D has devoted itself more intensively to the advantages of a high-quality OpenGL display in the viewport and is once again making significant improvements in Release 19: The textures loaded for the preview of environmental reflections can now be rotated globally or per material as required in order to optimise the position of the reflections, especially in still images.
However, another new reflection feature is particularly interesting. Reflections between objects can now also be displayed directly. However, some restrictions must be expected here. The effect is based on the screen space, as we already know it from the ambient occlusion preview in the viewport. Therefore, only the surfaces that can be seen directly by the camera can appear in reflections. For this reason, the effect works best with a flat viewing angle of the reflective surfaces. However, this is not a problem, at least for product shots that are to be presented on a reflective floor. The fact that reflections between objects only work with perfect surfaces is somewhat more restrictive. Roughness in the reflectivity is therefore not supported. However, the use of a slight bump can help to soften the perfection of the reflective surfaces somewhat. In addition, a distance value can be specified over which the reflections can be smoothly faded out.

A completely new feature in the viewports is the display of depth of field (DOF). This is also thanks to the extended OpenGL functions. The calculation takes into account the target distance of the camera as well as the aperture setting and the sensor size. Even different aperture shapes can be used for the visualisation of bokeh. Although this effect is not physically precise, it looks very natural and immediately creates a fascinating depth, especially in macro shots and more effective perspectives.
In order for all this to actually become an alternative to normal ray tracing, the results must of course also be able to be saved. This has always worked with the hardware OpenGL renderer. This now also supports all new effects and improvements from the viewports. It has also been given an additional quality setting that can, for example, smooth out transparencies and reflections to a high standard. It complements the antialiasing, which is still available and smoothes the edges of objects. Finally, the OpenGL representations of reflections and tessellation can now also be switched on or off per object using a representation tag.

“If I don’t see it, I don’t need it.”

This is how the level or detail function known from computer games could be described. The graphics card must be relieved of the constant flood of polygons and textures wherever possible. Objects that only take up a small screen area are particularly suitable for this. As a rule, these are objects that are further away from the viewer or, of course, objects that are no longer seen by the camera. In fact, the new LOD object offers precisely this function, whereby it is not just a matter of simply switching visibility. The LOD object can manage entire groups of objects and offers numerous options as to which criteria should be used to control which properties of the objects. For example, the number of polygons of an object can be automatically reduced with increasing distance from the camera or the quality level of the display in the editor can be reduced. The most common use will certainly be to create different versions of an object and categorise them under the LOD object. At the top is the most detailed version, followed by any number of simplified versions. A criterion must then be selected that is to be used for switching. This can be done manually, e.g. via keyframes or XPresso, or automatically by comparing the object size with the screen area or the distance of the object to the camera.
Another alternative is to add objects to a model step by step. In this way, new components can gradually appear as you approach them without having to switch completely between different models.
Within the Cinema 4D environment, LOD objects can be used as mograph clones, for example, or whenever performance in the viewport is an issue. This effect can also be used for rendering, but may not be quite as helpful there, as the additional objects in the scene increase the memory requirements without directly improving the display quality. However, the LOD objects and settings can be transferred directly to game engines or other 3D packages such as Maya and used there.

All-round view

A small but significant enhancement to the camera now also enables the rendering of full spheres, domes or partial spheres. Full-sphere images of an environment can be used for baking an environment texture or an HDRI. However, they will certainly be used even more frequently in VR applications, where both still images, animations and stereoscopic renderings can be created. An interactive player for such images or videos is not available in Cinema 4D, but video platforms such as YouTube or social networks such as Facebook have long offered free players that can at least be used to control the rendered material.

Character care

Character animation can be a nerve-wracking process, especially if the deformed geometry in the joints does not deform naturally enough. Hips and shoulders in particular often cause headaches when weighting. A new PSD mode of the pose morph tag provides a remedy here.
After assigning the tag, bring the rig into a typical, critical position in which, for example, unsightly deformations appear on the shoulder. Correct these afterwards using the brush tool, for example. The pose morph tag not only remembers the change to the point positions, but also the position of the joints that have an influence on these points. After switching the tag to animation mode, the model will now show the corrected shape whenever the joints move to the memorised position – with correspondingly smooth transitions to the original shape, of course.
There are also new features for the familiar weighting tools. The separation between the weighting manager on the one hand and the weighting tag and weighting tool on the other has been made even stricter. The listing of weighted joints has therefore been removed from the tag, which may seem unusual at first. However, by right-clicking directly on the mesh with the weighting tool active, a submenu now opens which can be used to select the joints to be weighted.
The mirroring tool has been extended with functions from the weighting manager so that the often time-consuming work on weighting can be limited to the bare essentials. This has significantly improved the recognition of mirrored join hierarchies and symmetrical points for the assignment of weightings.

Short and sweet

Certainly a highlight of Release 19 is the extension of the Voronoi break function, which can be used to split any object into fragments. New modes allow the gaps between fragments to be filled with geometry and the actual fragments to be hidden. This results in a mesh structure that can either run through the object in three dimensions or just cover its surface. The Voronoi structure of the surface can also be thickened by extrusion.
Completely new fragment shapes are made possible by the fact that the scaling of the fragment cells can be scaled along any axis direction. This allows, for example, the typical splinters of bursting wood to be realised.
In addition, the previously smooth sides of the fragments can be deformed using various noise functions so that they resemble natural materials.
This deformation can also be extended to the course of the fracture edges and even to the surface of the bursting object. Using a curve, the intensity of this deformation between the surface and the core of the object can, for example, be designed to taper off smoothly.
Numerous new options have been added to further influence the distribution and size of the fragments. For example, areas can be marked with polygon objects or falloff objects in which fragments are to be fused. Alternatively, a fixed number of fragments can be specified, or the fragments that fall below a certain proximity to each other can be glued together. In either case, this results in new fragment shapes that can deviate significantly from the typical Voronoi fragment shape. As is known, all fragments are given an internal numbering, which can be analysed using a step effector, for example. This sorting of the fragments can now be changed even more easily, e.g. to run along an axis or even to be dependent on the distance to an object or spline.
As Mograph clones and therefore also fragments are often used in conjunction with Dynamics, there are also new features in the Voronoi fragment object. The fragments can be automatically linked to connectors, springs or even motors. The combination with fixed connectors is particularly appealing, as individual breaking forces can be specified above which the fragments lose their connection. This makes the shattering of an object when it falls onto a collision object even more realistic, as fragments and cracks primarily appear where the greatest forces occur during the collision.

Everything dances to my tune

The sound effector has been completely revised and offers new graphical user interfaces for processing and visualising frequencies and volumes. Any sections of the frequency band can be marked with frames and read out. Clones can be coloured, moved or their visibility controlled via frequencies. A completely new feature is the noise falloff function, which is now available to all effectors. The effect is similar to using a random effector, except that the falloff can now be used to randomise the area of influence of an effector. Here too, a new GUI element directly in the view windows helps with the selection and configuration of the noise falloff.

At every turn

The motion tracker and object tracker have been with us for several versions. Release 19 also brings improvements in the evaluation of footage in this area, as from now on not only the brightness contrast, but also any colour component can be used for manual tracks. The shape of the areas responsible for pattern recognition can also be created in a circle – practical when
spheres are used as tracking markers, for example.
However, things get really exciting once the camera movement has been reconstructed. An optional scene reconstruction can be connected here. The previously placed trackers and camera positions are used to project additional points into the 3D space, which fill in the areas between the trackers. The distribution and density of this point cloud can be finely controlled and lays the foundation for the final calculation step, in which connecting polygons are created between the points. Small, detached groups of points can be automatically ignored here in order to obtain a model that is as noise-free and coherent as possible. As the generated point cloud takes over the colouring of the footage via a vertex colour tag, a vertex shader can then be used to colour the reconstructed polygon object. Alternatively, the original footage can of course also be projected onto the geometry.
The quality of the reconstructed geometry depends heavily on the previously calculated camera movement and generally on the illumination of the depicted objects as well as the resolution of the footage. Fine structures are generally less easy to capture. However, larger structures such as a rock face work surprisingly well. The quality does not yet come close to that of external photogrammetry solutions, but should be perfectly adequate for simple reconstruction tasks or to provide a meaningful template for post-modelling.

New territory and outlook

Even before the release of Release 19, an innovation was publicised: the new ProRender render engine is also able to use the GPU for rendering. A novelty for Cinema 4D users, apart from the possibility of using external GPU renderers such as Octane or Cycles.
ProRender is under continuous development by Radeon, so it is not a tool programmed by Maxon itself and is therefore also available for other 3D applications. In addition, this renderer is OpenCL-based, so it does not necessarily require a CUDA-capable graphics card. It is clear that ProRender is still a very young product whose range of functions is not yet on a par with renderers that have been available for longer. It must therefore be pointed out that this renderer is not yet ready for production and that the current version is by no means intended to replace the native renderers of Cinema 4D. Nevertheless, ProRender is already a valuable addition in the current version, as we can use it for quick test renderings in the viewport, for example, or of course for final renderings. However, there is currently still quite a long list of limitations to consider. For example, hardly any native Cinema 4D shaders are supported. These must therefore be baked beforehand, which means that effects such as SSS or thin film, which depend on the viewing angle of the surface, are ruled out from the outset. There are also restrictions on the use of point and spot light sources as well as various post effects. Multi-passes cannot be used either. Only one alpha channel and an ambient occlusion solution can be rendered.
If we ignore these limitations for the time being, the very clear quality settings, which are limited to the definition of the beam depth for the reflectivity and GI calculation, the quality of the anti-aliasing and the upper limit of the iterations, are particularly surprising. This is typical for physically based unbiased renderers, which do not use any simplifications or tricks to achieve the shortest possible render times. Instead, each pixel is always processed with the same precision (brute force). As this process can be computationally parallelised very well, such a renderer is particularly suitable for processing on graphics cards that can process hundreds or even thousands of computing steps in parallel. Their performance improves much faster than that of processors. In addition, it is often much easier and cheaper to improve the performance of the computer by adding or replacing graphics cards. However, it is important to note that the entire scene must fit into the memory of the graphics card for rendering. This can still be a problem with complex architectural projects, for example.
In order to fully utilise the potential of ProRender, care should be taken to ensure that the scale of the objects and the scene is realistic and that the materials and lights have realistic properties. Release 19 therefore comes with a new physical material and a physical light source that can be used directly as a basis. Although both elements are only suitably predefined standard materials or surface light sources, this eliminates typical sources of error when texturing and illuminating scenes.

Conclusion

Release 19 brings helpful enhancements, particularly in the areas of OpenGL, Mograph, Motion Tracker and the optimisation of projects through polygon reduction or level of detail. The new GPU renderer invites you to experiment and will make physically based rendering even easier and, with the appropriate hardware, faster in the future. Simpler projects can certainly already be realised with it. However, ProRender is currently mainly used as an interactive renderer for quick previews in the viewport.
However, a large part of the new features still remain hidden. However, work on the new core of the software will soon bear fruit and provide an optimistic view of the future of Cinema 4D.