HappyPolygon

Two days after the Utility Splines Collection Launch Party the new Season started. Chris got carried away with the new capsules and forgot to mention when the season would start.

AMD has released FSR 2.2. Fidelity FX Super Resolution 2 (FSR 2.2), is the latest version of its open-source image upscaling technology, widely used in games and now also supported in AMD’s workstation GPU drivers. The changes are mainly targeted at games rather than DCC tools: the biggest one is to reduce ‘high velocity ghosting’, seen in racing games where a third-person vehicle moves at speed over a flat surface. AMD has also added a new Debug API Checker feature to pass debug messages from the FSR 2 runtime back to help troubleshoot new FSR integrations. However, there are also changes to improve temporal luminance stability, reducing flickering on rendered objects, and to make FSR compatible with a “wider set of high dynamic range inputs”. Licensing and release dates FSR 2.2 is compatible with Windows 10+, and supports DirectX 12 and Vulkan. The source code is available under a MIT licence. https://gpuopen.com/fidelityfx-superresolution-2/ Dimension 5 Techs has released D5 Render 2.4. (The firm is based in China, and is governed by Chinese law) It looks to be a significant update, introducing a new real-time Global Illumination system: a modified implementation of the Nvidia-developed ReSTIR algorithm that Dimension 5 calls ReSTIR Surfel GI. Unlike the previous GI solution, it takes account of the skylight during caching, resulting in more natural lighting in interior scenes; and calculates indirect as well as direct illumination from emissive materials. According to Dimension 5, its output is “very close” to path tracing, providing a better balance between render speed and visual quality. Other key changes include a new subsurface scattering material for semi-translucent materials like marble and skin; and a new Z-depth channel, shown in use in the video above to adjust environment fog. The update also adds controls for camera focal length and render culling distance for vegetation; new corner radius controls for the Path tool; and support for AV1 and QSV video encoding. The asset library gets over 580 new assets, including sets of Asian characters and dynamic water particles. D5 Render is available for Windows 10. It requires a GPU capable of hardware-accelerated ray tracing: Dimension 5 recommends a Nvidia GeForce GTX 1060 or higher, or an AMD Radeon RX 6700 XT or higher. Integration plugins are available for 3ds Max 2014-2016 and 2018+, ArchiCAD 21+, Blender 2.93+, Cinema 4D R20+, Revit 2018+, Rhino 6.1+ and SketchUp 2017+. The Community edition is free; subscriptions to the Pro edition, which includes the Widget, access to the online asset library and product support, cost $38/month or $360/year. https://forum.d5render.com/t/release-notes-of-d5-render/5791 Foundry ships Modo 16.1 The release – the second of two in the Modo 16 Series, and originally due to ship last year – adds interesting new geometry-based workflows for applying decals to the surface of a model. Other changes include support for triplanar texture projection, a new Particle Clone operator for object scattering, and updates to Modo’s modelling tools, Advanced Viewport and mPath renderer. https://learn.foundry.com/modo/content/help/pages/welcome_modo/whats_new.html Other Foundry News In separate news, Foundry has announced major changes to the development of Modo, including an expanded dev team, a broader market focus, and a switch to a single major release of the software per year. Foundry has stopped development of Colorway, its software for generating and visualising variants of product designs, as part of a wider move away from design-specific tools. At the time of writing, Colorway is still listed on Foundry’s website, but the firm has confirmed that its old development team is now working on Modo, Foundry’s 3D modelling and rendering software. The changes suggest a broader target audience for Modo, which in recent releases had seemed to be moving away from its origins in game development and VFX and becoming increasingly focused on product design – specifically shoe design, since its major users include Nike and New Balance. Brown described the current strategy as ‘Modo for everyone’, commenting that “focusing on a specific industry isn’t the direction that 3D is going”. “All of the people we worked with in design are customers we care about [but to best cater to the needs of users] you just need to to make an application that is high-performance, stable [and] reliable”. Clavicula - Free SDF modelling tool The software has been in development for a couple of years, but has just gained interesting a new Boolean modelling toolset based on Signed Distance Fields, also used in Adobe’s Substance. Although it can be used to rough out organic characters, most people will probably use it for hard-surface models: from product designs and games props to base forms for use in concept art. Like Substance 3D Modeler, you can even work in virtual reality, although in Clavicula, VR support is currently experimental, and only available in the Windows edition. https://clavicula.link/ Chaos has released V-Ray 6 for 3ds Max Update 1 A wide-ranging update, adding new features to the VFB, V-Ray Decal, the V-Ray physical camera, and V-Ray lights. The release also implements Intel’s Open Path Guiding library when rendering on CPU, a new compressed textures mode to reduce memory usage when rendering on GPU, and AI upscaling in V-Ray Denoiser. Visual effects artists get initial USD export, making it possible to export V-Ray materials, lights and cameras to other host applications in USD format. Outside the core application, the new workflow introduced in V-Ray 6 for 3ds Max for sending renders to Chaos Cloud directly from the VFB to solicit feedback from clients or collaborators has been extended. Users can now add visual annotations, including lines and arrows, to both still images and image sequences; and Chaos Cloud now supports versions and A/B comparisons. https://docs.chaos.com/display/VMAX/V-Ray+6 Hextile for Houdini and Unreal Engine Ubisoft Singapore senior technical artist Mohsen Tabasi has released Hextile, a free add-on for Houdini and Unreal Engine that minimises visible repeats when tiling textures. Hextile tiles textures using an underlying hexagonal grid, rather than a conventional rectangular grid, minimising visible repetitions when the source texture is repeated many times. It works particularly well for textures with random patterns, like stones or gravel: while it also works with regular patterns, like brickwork, it creates a kind of ‘crazy paving’ effect that may not suit every project. Hextile is available as a HLSL shader for Unreal Engine, and as both a .hip file and a HDA for Houdini: the HDA should also work in other DCC applications via Houdini Engine. It’s a free download. https://www.sidefx.com/community-main-menu/contests-jams/tech-art-challenge-2022/ https://github.com/proceduralit/PI_Sandbox/blob/main/LICENSE.md Howler 2023.5 The developers of hard-to-classify digital painting, animation and video processing tool Howler have released Howler 2023.5, the latest version of the software. The update adds a new filter for generating sky backgrounds complete with volumetric clouds, one-to-three-point perspective guides and a new Undo browser, and updates several of Howler’s existing filters. The update adds the option to import 3D models and generate renders of them for use as custom brushes, as shown in the video above, opening up the possibility of interesting hybrid 2D/3D workflows. Other changes include a new colour temperature filter, and a rotation parameter for brushes. http://www.pdhowler.com/WhatsNew.htm SideFX releases Houdini for Apple Silicon The Apple Silicon build is now considered ready for use in production, and will receive regular daily builds and production builds, alongside the builds of Houdini for Windows, Linux and Intel Macs. However, there are still a few known limitations, including lack of GPU acceleration in Karma XPU, the CPU/GPU version of Houdini’s new renderer – still officially in beta, and only supported on Nvidia GPUs. In addition, of SideFX’s Houdini Engine plugins for DCC applications and game engines, only the Unity plugin is supported, since neither Maya or Unreal Engine yet have native Apple Silicon support. Reallusion ships iClone 8.2 + iClone Unreal Live Link 1.2 Both iClone and iClone Unreal Live Link now support characters with facial wrinkles generated by Reallusion’s long-awaited dynamic wrinkles system, introduced in Character Creator 4.2. It automatically generates wrinkles on a character’s face in response to changes in expression, layering finer, texture-based wrinkles over the larger folds of flesh generated by Character Creator’s expressions system. The wrinkles can be triggered by keyframe animation, or in real time by Face Puppet, iClone’s facial puppeteering system, or by motion-capture data streamed in via the add-on Motion Live system. https://www.reallusion.com/iclone/update.html https://www.reallusion.com/iclone/live-link/unreal-engine/update.html Reallusion releases Character Creator 4.2 Reallusion has released Character Creator 4.2, the latest version of its 3D character creation tool. The update adds the much-anticipated dynamic wrinkles system, which automatically generates wrinkles on a character’s face in response to changes in expression, adding realistic detail to facial animations. The textures themselves are derived from scans of real actors, optimised for the topology of Character Creator’s CC3 Base+ (CC3+) characters. The system uses five source texture maps (normal, crease, ambient occlusion, skin redness and a flow map) from which Character Creator calculates three derived 4K maps: diffuse, normal and roughness. Character Creator 4.2 ships with four base wrinkle patterns: two generic, and two specific to a young male character and an older female character. A further 10 wrinkle patterns – five realistic, and five for cartoon characters – and 48 facial morphs intended for use with the wrinkle system, are available via a paid add-on, Wrinkle Essentials. Other new features: better support for MetaHumans, updates to AccuRig and GoZ https://www.reallusion.com/character-creator/update.html Modif - free AI-based mocap app for iOS and Android The app has already been available in Korea for a few months, but Plask is now rolling it out in more territories, including the US and Japan. Already available as a browser-based service, Plask’s machine-learning-trained technology extracts the motion of an actor from video footage and retargets it to a 3D character. Modif packages that tech into a mobile app, with users able to crop videos captured on their phone or tablet, then have Modif extract the motion of the actor’s body and retarget it to a stock character. Modif is available for iOS 13.0+ and Android 7.0+. It’s free to use to process up to 60 minutes of video, after which you will need to register a new account with Plask. You can process up to 15 seconds of video per clip. https://www.modif.app/ Photoshop 24.2 The update adds the option to password-protect links to documents shared for review, improves performance of the Object Selection tool, and adds support for app window snapping on Windows. The iPad edition of the software has also been updated, with Photoshop on the iPad 4.4 adding support for copying and pasting text and images to and from the iOS clipboard. Performance of the AI-trained Object Selection tool has been improved, although the release notes don’t put a figure on the improvement. The Windows edition now supports Windows’ Snap feature, making it possible to snap the app window to those of other apps when running multiple applications side by side. https://helpx.adobe.com/photoshop/using/whats-new/2023-1.html Forest Pack 8.08 The release extends the new ForestSet plugin, making it possible to use ForestSets as Object Exclude areas in order to remove items from a scatter, as shown in the video above. There is also a new Disable Set Content option to turn off linked assets without having to remove the from the ForestSet object https://docs.itoosoft.com/forestpack/forest-release-notes Source: CGchannel

It's set by default to search for words in current page. You need to select "All content" from Content Type For me a big disadvantage is the absence of searching words in posts made from a certain user.

It's not always to make a checker-board selection because the way the polygons get selected is structural and not visual. This means that the sequence in which polygons are constructed is not always from left-to-right and top-to-bottom. And it will certainly not work when more than 4 polygons share a common point like bellow. One solution I suggested here some time ago was to use a Shader Field with a checker-board shader but it will still not give the best results.

I wasn't able to track-down that particular developer but found out that Nitro4D is on it so I expect a low-cost plugin. And someone else wrote a script that auto-sends a rendered image to Google Colab to use the Stable Diffusion Model. It's very cheap but subscription based and the "bad" thing with Colab is that it's also subscription-based if you want a fast and high-res output.

Haha I forgot about that, you are right ... Did you write this or was it Deepstory / Dramatron ? 😜

Sorry but I don't use Octane and have no idea how to use it. It is 100% an Octane matter. It probably has to do with the projection methods it uses. As I can see the original gradient is used but not the stepped method or the tilling I described above. You have to recreate the Material in Octane as an Octane Material and probably use the Octane projection mechanisms to achieve the same result. For the gradient see here. I've found an important note on C4D shaders for Octane here that links to the C4D Gradient and the differences between that and the Octane gradient For projection tilling se here or here As I said I have no idea about Octane so my links might not be helpful at all or even my approach won't even work with Octane's logic, for example UV Transform may only work with bitmap textures not shaders and you'll have to recreate the gradient in Photoshop...

Oh... there is nothing exactly like that but there has been a video about an AI-driven liquid simulation.

yeap

sauce ?

Already there

Ubisoft Singapore senior technical artist Mohsen Tabasi has released StableDiffusion_Houdini 2.0, the latest version of his free tool that links Stable Diffusion into Blender. The plugin makes it possible to use the open-source AI image-generation model inside Houdini, in order to use 3D scenes to guide the image that Stable Diffusion generates. With it, users can convert renders of simple blocking geometry into detailed illustrations and concept art. Version 1.0 required an online Stability AI DreamStudio account to generate images, but version 2.0 makes it possible to set up a server based on open-source tool AUTOMATIC1111 and run entirely on a local machine. https://www.cgchannel.com/2023/02/free-tool-mohsen-tabasis-stablediffusion_houdini-2-0/?fbclid=IwAR3CI2Bjm-NxRx5mzFam2ISGKwECCj8Qij-rRpoNvPU6VwmwtEzqIfOb4Hk

Oh that's very clever. I thought I was in the Scene Nodes pit. It took me quite some time to figure out how the hell you managed to delete polygons using fields.

You are right. And it's a quite old bug. R25 old

Damn politics ...

I don't use RS. If ZEROSIXTWOSIX's approach doesn't work with RS try brute forcing it... How many knots does your gradient have ? Make that many copies of the object each with a specific Material color from the gradient. You could also use (if there exists an equivalent in RS) the MoGraph Multi Shader...

I'm not sure if the "technical demo" was a forced (marketing) move due to the release of Blender Geometry Nodes but it definitively was the best move since that gave MAXON the biggest possible feedback. Up until then every new feature was tested exclusively indoors or with close NDA collaborators. Introducing such a big feature (that would definitely be called a module back in 2010) needed a bigger audience, not just people who worked in the field for decades. And the next iteration of Scene Nodes in R25 was a huge improvement. You got an iteration node or conversion node for free when connecting specific nodes together saving you from aimless debugging or logical dead-ends. There's a Scene Root default node. Better search algorithm and of course capsules. Well, I'm not 100% sure if this is really an open question. In order to define the system as Turing-complete you have to define it in mathematical terms and prove or disprove the original hypothesis. From my point of view if there are no equivalent structures between nodes and coding like variables, if, while-do, case etc. then it's probably not Turing-complete (and to be clear, commands like if, while-do, case have nothing to do with the Turing-completeness theorem). My first struggle was with variables. You can define a constant and feed that anywhere but you can't store a value back to it. There's only an output. Maybe I'm wrong and everything does have an equivalent node structure and I'm the one that fails to recognize because I'm not used to alternative paradigms especially plethoric in expression (more than one node/statement). I think nodes are on the right track. They are getting constant feedback and introducing new Capsules like the latest Spline chamfer. Bit by bit all the old handy plugins will be replaced by native tools. I believe that even Ricochet can be reconstructed using scene nodes. The big question is whether Scene Nodes will compete with Geometry Nodes face-to-face providing what C4D is best known for (intuitiveness) combined with the quality and power of Houdini. We expect a lot of improvements in all aspects of C4D but due to NDAs all will be reveled at the right time. And of course God willing concerning war and natural disasters...

The Volume Builder/Mesher is probably the only complete feature in C4D. For everything else there is room for improvement.

Since R23 there was a discussion about how Scene Nodes were supposed to actually help things out. The discussion was mainly philosophical because from the existential point of view (why are Scene Nodes even needed in the first place) there wasn't much backing up of the idea. It seemed that the main reason Scene Nodes were developed was the constant expressed opinion of the public wanting a node-based modeling system like Houdini, just because Houdini is king in CG production (Blender is also known for it's own system but Geometry Nodes were released some months after Scene Nodes so I don't think Blender was a factor). The new (Neutron) project had pitfalls for the following reasons, making the result look like a failure : The public demanding a node-based system had a false perception of what that would look like and how it would work. Mainly because most of them didn't even know how Houdini works, they just knew it was "better". Also many of them (users) weren't familiar with XPresso or Material nodes. So when the system was finally delivered the learning curve seemed overwhelming. Considering the poor documentation made things worse. Many users already knew that a node-based modeling/animation system's biggest advantages are the procedural and parametric characteristics. What they failed to acknowledge was that C4D already had the capability of procedural and parametric setups due to the powerful Object Manager, something that other DCCs are trying to replicate. Considering the above MAXON added the Capsule system. A bridge between the Object Manager and Scene Nodes. Now we know why Scene Nodes are needed and these are the following: Not all C4D operations are procedural or parametric. These operations are positioned mainly in Tools, Mesh and Spline menus. These are operations that are not stackable in the Object Manager. Best examples are the Inset, Extrude, Bevel and Bridge operations. Scene Nodes provide a better space (2D) than the Object Manager (linear) to accommodate for complex groups of operations AND procedural/parametric operations that are not available in the OM. Scene Nodes offer capabilities that XPresso couldn't like UV/color management during modeling and some attribute manipulation mostly with instances that was impossible before. A better graph structure. XPresso had a less restrictive brunching structure which made it prone to logic conflicts when same attributes were manipulated simultaneously by many operations. Scene nodes are less prone to these kind of conflicts due to their "one-way" structure. Personally I have seen very few occurrences of cyclic structures (infinite loop) in Scene Nodes and when that happens it immediately gets detected. The ability to construct new parametric primitives. The ability to construct new tools without diving deep in the C4D Python API. Procedural modelling. Points that are yet to be resolved: Visual aid on selections. Boolean Operations. A better category system to provide better housing for the expanding number of nodes. (sub-categories for example or a tier-based metadata tag filter to better serve users with different experience or type of work (newbies want to see only capsules, techies want the low-level structure nodes) Access to specific Capsules from any custom pallet (I think it has been solved but not sure) More high-level nodes for complex operations (the "bad" thing is that MAXON tries to construct high-level nodes from low-level nodes and not just hard-code new nodes, this is time consuming because it leads to technical difficulties) Nodes are harder to update. I'm talking about high-level nodes like Dual-mesh. It's not the same thing as adding new features to an existing coded feature. Although nodes are scalable, adding new features/mechanics in an existing structure can be much trickier compared to coding. Faster loading times when the node pool is summoned. Make scene nodes a high-level workspace for procedural and generative modeling by adding mechanics to accommodate specific needs. For example Adaptive Grid Space for asset placement - used to create procedural cities, buildings etc. Terrain modeling tools - diffusion, erosion, terrace and many more modifiers are easier to work with in a 2D node space rather in OM. Agent system - for crowd animation, traffic animation or simulating flocks of birds, schools of fish etc. Some philosophical issues still remain open: If Capsules are used like deformers in the OM, why still use them as such and not hard-code them later as regular Deformers adding more functionality to them like having Dual Mesh affect certain regions of geometry using Fields ? Should all features be mirrored as nodes in the node pool ? For example instead of dropping a Cloner in the Scene Node space from the Object Manager, just drop it from the node pool, because you don't want to have the cloner or any other hierarchy chain of native C4D elements in the OM, you just want the result to manifest from the Scene Nodes. Should Selection Strings be adopted to the rest of the app or evolve into a better system with all the benefits from Selection Strings and the usual Selection form ? Should custom icons be made for most nodes in the Geometry Modifiers category ? (what's the point of having the same icon if the icon is not used for identification) Should there be a ready-made Palette with deformer-like capsules and primitive-like capsules for users to select from without having to open Scene-Nodes ? (helps new users to know all available geometry manipulation options making features look system-agnostic (I call systems all node systems and the python interfaces)) Is Scene Nodes Turing-complete ? This levels Neutron with programming. If the system is Turing-complete that means that anything that can be programmed can also be translated to nodes. This is very important because it can save you from chasing your own tail. If I know that something is possible with Python and that Scene Nodes is Turing-complete then I can make the attempt to implement it using Scene-Nodes. This tread has forked more than any other thread 🤣

But you've already did that back in October... is this faster or did you update the project into a capsule ?

I found 2 hours late the new VFX and Chill broadcast today. They tried replicating the exact same scene as you... I we were watching it we could provide some useful insight cause they seem quite lost and didn't really bother making their homework about how exactly they were originally made before trying to tackle the scenes they picked...

Yes, and all modeling tools like extrude, weld, bridge, cut ...

Please fill the appropriate info in your profile. Your case looks like a very good opportunity for the Volume Builder/Mesher to shine. Especially the latest C4D version has the very useful Radius blend. After that you can polish the topology with the ZRemesher from the Remesh generator. It can definitely be modeled using the traditional tools with a Subdivision Surface finish but you need some prior experience. I would not recommend any Boolean operations for this kind of modeling. Do you need a step-by-step guide ? If yes, @Cerbera is our modeling professor. P.S. You forgot to upload the file you mention, unless you're referring to the image.

Before the new update comes available you can try out this HDRI 360 panorama creator. The output is at 1024x512 but you can upscale it using this handy AI resizer https://bigjpg.com/ Currently the output is not seamless but this is expected to be patched in the next version.

What I had in mind was an animated shader. But, yes, it's not meant for interactions between objects unless they can figure out a way to bake the displacement and have it run as a simulation from a point on.