Let's start with very helpful tools that will become the standard in the future:
AI Rescale:
Advanced tool that enlarges the image by analyzing each pixel.This technique is based on the fact that artificial intelligence uses an algorithm to fill missing pixels.Traditional up-scale or Rescale methods use interpolation for example Bi-Cubic with AI is trained in recognizing a given image and completes the missing pixels.
Example:
By Adrian Kulawik
Left Side:Octane Render 4 native Rendering to 4K.
Right Side:AI Rescale from FullHD to 4K.
In the future, this technology will also be implemented for animation.
AI Denoiser:
By Adrian Kulawik
AI gathers enough information about a scene to determine where the edges are.
She make very good prediction of the final image.
AI Lights:
Works like a portal panel for Photons.
Tracking emissive points and their direction.
AI DLSS (Deep Learning Super-Sampling)
Deep Learning Supersampling is a post-processing technique that improves performance against traditional edge smoothing methods in two ways:
1.)Takes fewer samples per pixel than current anti-aliasing solutions, which translates into lower GPU load.
2.)DLSS is entirely implemented by TENSOR CORES cores instead of standard CUDA.
DLSS By NVIDIA.
AI 2D Images to 3D World:
Generative Query Network (GQN) is a machine-trained artificial intelligence algorithm that specializes in image recognition.
Based on 2D photos AI predict the position of elements in three-dimensional space.
(for now it is able to make very simple geometric solids)
By DeepMind.
AI Animation AutoRig:
DensePose-RCNN recognizes the silhouettes of people moving in real time.
By DensePose
Conclusions:
The future, a large branch of 3d infrastructure will be associated with artificial intelligence but also in:
Games
VR(Virtual Reality)
Automotive
Robotics
2D
Homes
in every area of our lives it will have a lesser or greater impact.
Next 5 Years:
It has always been the dream of all 3d artists to be able to render images in real time.
The next 5 years We will measure render time by FPS(Frames per Second) not hours or minutes for animation.It is expected that the processing power of GPUs due to architecture and artificial intelligence will increase in 2025 to 1000x.
The 3D industries in which AI will be most visible:
-Motion Capture.
-Retopology 3D Models.
-Clean up, simplify, or validate meshes.
it can be said that these directions in the industry are the most endangered.
Directions in the industry that are the future:
-Arch Viz.
-Freelancers.
-3D Modelers.
-Concept Art.
-VR and Games.
AI Rescale:
Advanced tool that enlarges the image by analyzing each pixel.This technique is based on the fact that artificial intelligence uses an algorithm to fill missing pixels.Traditional up-scale or Rescale methods use interpolation for example Bi-Cubic with AI is trained in recognizing a given image and completes the missing pixels.
Example:
By Adrian Kulawik
Left Side:Octane Render 4 native Rendering to 4K.
Right Side:AI Rescale from FullHD to 4K.
In the future, this technology will also be implemented for animation.
AI Denoiser:
By Adrian Kulawik
AI gathers enough information about a scene to determine where the edges are.
She make very good prediction of the final image.
AI Lights:
Works like a portal panel for Photons.
Tracking emissive points and their direction.
AI DLSS (Deep Learning Super-Sampling)
Deep Learning Supersampling is a post-processing technique that improves performance against traditional edge smoothing methods in two ways:
1.)Takes fewer samples per pixel than current anti-aliasing solutions, which translates into lower GPU load.
2.)DLSS is entirely implemented by TENSOR CORES cores instead of standard CUDA.
DLSS By NVIDIA.
AI 2D Images to 3D World:
Generative Query Network (GQN) is a machine-trained artificial intelligence algorithm that specializes in image recognition.
Based on 2D photos AI predict the position of elements in three-dimensional space.
(for now it is able to make very simple geometric solids)
By DeepMind.
AI Animation AutoRig:
DensePose-RCNN recognizes the silhouettes of people moving in real time.
By DensePose
Conclusions:
The future, a large branch of 3d infrastructure will be associated with artificial intelligence but also in:
Games
VR(Virtual Reality)
Automotive
Robotics
2D
Homes
in every area of our lives it will have a lesser or greater impact.
Next 5 Years:
It has always been the dream of all 3d artists to be able to render images in real time.
The next 5 years We will measure render time by FPS(Frames per Second) not hours or minutes for animation.It is expected that the processing power of GPUs due to architecture and artificial intelligence will increase in 2025 to 1000x.
The 3D industries in which AI will be most visible:
-Motion Capture.
-Retopology 3D Models.
-Clean up, simplify, or validate meshes.
it can be said that these directions in the industry are the most endangered.
Directions in the industry that are the future:
-Arch Viz.
-Freelancers.
-3D Modelers.
-Concept Art.
-VR and Games.