Category: software
BVH (Bounding Volume Hierarchy) in Ray Tracing
Bounding volume hierarchies are used to support several operations on sets of geometric objects efficiently, such as in collision detection and ray tracing. A bounding volume hierarchy (BVH) is a tree structure on a set of geometric objects. All geometric objects, which form the leaf nodes of the tree, are wrapped in bounding volumes.
BVHs are often used in ray tracing to eliminate potential intersection candidates within a scene by omitting geometric objects located in bounding volumes which are not intersected by the current ray. BVH is a crucial component in ray tracing rendering engines like Arnold, as it helps accelerate ray intersection tests and reduce resource costs.
Users do not have control over RAM consumption of the BVH. Here are some tips to optimize Arnold renders when BVH is the bottleneck:
Optimize Your Scene Geometry. Simplify or optimize your 3D models and scene geometry. Complex geometry can lead to larger BVH structures and longer BVH build times. Consider using LODs (Level of Detail) or proxy objects for distant geometry to reduce the BVH complexity.
Use Arnold Stand-ins and Proxies. Arnold Stand-ins and proxies allow you to load complex geometry only when needed, reducing the BVH complexity during the initial BVH build. This can be particularly useful for scenes with a lot of high-poly assets.
Denoising. Applying denoising to your final render can help reduce the number of rays required and consequently, the BVH intersection tests.
Render in Layers. If your scene has many elements, consider rendering it in layers. This allows you to optimize each layer individually, potentially reducing BVH build times.
Distribute Rendering. If you have access to a render farm or multiple machines, distribute the rendering workload. This can significantly reduce rendering time as each machine can handle a portion of the BVH calculations.
CamP: Camera Preconditioning for NERF (Neural Radiance Fields)
https://arxiv.org/pdf/2308.10902.pdf
“…any errors in the camera position solve can dramatically reduce the quality of a reconstruction. “
“Introducing CamP — a method to precondition camera optimization for NeRFs to significantly improve quality. With CamP we’re able to create high quality reconstructions even when input poses are bad.
CamP is a general purpose trick that you can apply in many places. Apply CamP to prior SOTA NeRF models like Zip-NeRF, and you improve quality and establish a new SOTA. Applying CamP to existing camera optimization models, like SCNeRF, to improve their performance.” – Keunhong Park
World Creator 2023
https://www.world-creator.com/
World Creator lets you create terrains of any size. There are absolutely no limits. You can create terrains with a few meters and terrains with thousands of kilometers. On top of that, our new terrain system allows you to create terrains of any detail. From meter precision down to centimeter precision, such high details can only be achieved with World Creator.
SecretWeapons MixBox – a practical library for paint-like digital color mixing
Internally, Mixbox treats colors as real-life pigments using the Kubelka & Munk theory to predict realistic color behavior.
https://scrtwpns.com/mixbox/painter/
https://scrtwpns.com/mixbox.pdf
https://github.com/scrtwpns/mixbox
https://scrtwpns.com/mixbox/docs/
Executing Python Scripts With a Shebang
https://realpython.com/python-shebang
In this tutorial, you’ll:
- Learn what a shebang is
- Decide when to include the shebang in Python scripts
- Define the shebang in a portable way across systems
- Pass arguments to the command defined in a shebang
- Know the shebang’s limitations and some of its alternatives
- Execute scripts through a custom interpreter written in Python
In short, a shebang is a special kind of comment that you may include in your source code to tell the operating system’s shell where to find the interpreter for the rest of the file:
#!/usr/bin/python3
print("Hello, World!")
Introduction to Autodesk ShotGrid
https://customersuccess.autodesk.com/learning/course/introduction-to-shotgrid
Learn about ShotGrid’s basic capabilities and functionality in this introductory course. Set up your account, gain an understanding of the structure of data within ShotGrid, learn to navigate ShotGrid, determine your role, including what you can and cannot do, and customize the view of on-screen data.
Unpremult and Premult in compositing cycles
Steve Wright
https://www.linkedin.com/pulse/why-oh-premultiply-steve-wright/
James Pratt
https://jamesprattvfx.wordpress.com/2018/11/08/premult-unpremult/
The simple definition of premult is to multiply the alpha and the RGB of the input together.
Un-Premult suggests that this does the opposite operation to the premult node. Therefore instead of multiplying the RGB values by the alpha, it divides instead.
Alan Martinez
“Unpremult” and “premult” are terms used in digital compositing that are relevant for both those working with computer-generated graphics (CG) and those working with live-action plates.
“Unpremult” is short for “unpremultiply” and refers to the action of undoing the multiplication of a pixel by its alpha value. It is commonly used to avoid halos or unwanted edges when combining images. (This by making sure that edits to a layer are added independently from edges’ opacity levels.)
“Premult” is short for “premultiply” and is the opposite process of “unpremult.” In this case, each pixel in an image is multiplied by its alpha value.
In simple terms, premult crops the RGB by its alpha, while unpremult does the opposite.
It’s important to perform color corrections on CG renders in a sort of sandwich approach. First, divide the image to extend the edges fully of the RGB channels. Then, apply the necessary color corrections. Finally, pre-multiply the image again to avoid artifacts on the edges.
Typically, most 3D rendered images are premultiplied. As a rule of thumb, if the background is black or even just very dark, the image may be premultiplied. Additionally, most of the time, the 3D render has antialiasing in the edges.
Aaron Strasbourg
https://www.aaronstrasbourgvfx.com/post/2017/06/23/002-unpremult-and-premult
KeenTools 2023.2: Introducing GeoTracker for Blender!
https://keentools.io/products/geotracker-for-blender
Changes in add-ons Blender
- GeoTracker open beta
- Up to 3 times faster texture generation
- Minor bug fixes and improvements in FaceBuilder
Changes in GeoTracker for After Effects
- Video analysis up to 2 times faster
- Improved tracking performance up to 20%
- Accelerated surface masking
- Fixed primitive scaling issue
- Minor bug fixes and improvements
Changes in Nuke package
- Video analysis up to 2 times faster
- Improved tracking performance up to 20%
- TextureBuilder up to 3 times faster
- Accelerated surface masking
- New build for Nuke 14 running newer linux systems (RHEL9)
- Minor bug fixes and improvements
Photography Basics : Spectral Sensitivity Estimation Without a Camera
https://color-lab-eilat.github.io/Spectral-sensitivity-estimation-web/
A number of problems in computer vision and related fields would be mitigated if camera spectral sensitivities were known. As consumer cameras are not designed for high-precision visual tasks, manufacturers do not disclose spectral sensitivities. Their estimation requires a costly optical setup, which triggered researchers to come up with numerous indirect methods that aim to lower cost and complexity by using color targets. However, the use of color targets gives rise to new complications that make the estimation more difficult, and consequently, there currently exists no simple, low-cost, robust go-to method for spectral sensitivity estimation that non-specialized research labs can adopt. Furthermore, even if not limited by hardware or cost, researchers frequently work with imagery from multiple cameras that they do not have in their possession.
To provide a practical solution to this problem, we propose a framework for spectral sensitivity estimation that not only does not require any hardware (including a color target), but also does not require physical access to the camera itself. Similar to other work, we formulate an optimization problem that minimizes a two-term objective function: a camera-specific term from a system of equations, and a universal term that bounds the solution space.
Different than other work, we utilize publicly available high-quality calibration data to construct both terms. We use the colorimetric mapping matrices provided by the Adobe DNG Converter to formulate the camera-specific system of equations, and constrain the solutions using an autoencoder trained on a database of ground-truth curves. On average, we achieve reconstruction errors as low as those that can arise due to manufacturing imperfections between two copies of the same camera. We provide predicted sensitivities for more than 1,000 cameras that the Adobe DNG Converter currently supports, and discuss which tasks can become trivial when camera responses are available.
Denoisers available in Arnold
https://help.autodesk.com/view/ARNOL/ENU/?guid=arnold_user_guide_ac_denoising_html
AOV denoising: While all denoisers work on arbitrary AOVs, not all denoisers guarantee that the denoised AOVs composite together to match the denoised beauty. The AOV column indicates whether a denoiser has robust AOV denoising and can produce a result where denoised_AOV_1 + denoised_AOV_2 + … + denoised_AOV_N = denoised_Beauty.
OptiX™ Denoiser imager
This imager is available as a post-processing effect. The imager also exposes additional controls for clamping and blending the result. It is based on Nvidia AI technology and is integrated into Arnold for use with IPR and look dev. The OptiX™ denoiser is meant to be used during IPR (so that you get a very quickly denoised image as you’re moving the camera and making other adjustments).
OIDN Denoiser imager
The OIDN denoiser (based on Intel’s Open Image Denoise technology) is available as a post-processing effect. It is integrated into Arnold for use with IPR as an imager (so that you get a very quickly denoised image as you’re moving the camera and making other adjustments).
Arnold Denoiser (Noice)
The Arnold Denoiser (Noice) can be run from a dedicated UI, exposed in the Denoiser, or as an imager, you will need to render images out first via the Arnold EXR driver with variance AOVs enabled. It is also available as a stand-alone program (noice.exe).
This imager is available as a post-processing effect. You can automatically denoise images every time you render a scene, edit the denoising settings and see the resulting image directly in the render view. It favors quality over speed and is, therefore, more suitable for high-quality final frame denoising and animation sequences.
Note:
imager_denoiser_noice does not support temporal denoising (required for denoising an animation).
Stack OverflowAI
https://stackoverflow.blog/2023/07/27/announcing-overflowai/
EDIT
OpenAI and Stack Overflow’s recent partnership announcement has raised concerns for some users that their data is being used without permission. Some users who wanted to delete their answers have had their accounts suspended. The site does not generally allow the deletion of posts if the questions have any answers. Stack Overflow moderators say that it doesn’t allow users to remove posts as they hurt the company as a whole. While the GDPR gives users in the EU the right to be forgotten, websites have a right not to delete data if it doesn’t contain identifiable information to prevent the flow of a topic or question from being disrupted.
