This is a distributed graphics file processing manager for a local network. It interacts with workstations running a unified version of ComfyUI.
The manager allows for the simultaneous use of ComfyUI tools on multiple workstations for batch processing of multiple images,
resulting in a significant increase in performance.
The manager assigns individual tasks to workstations based on their performance and current workload, collects the results,
and presents them to the user.
The system is designed so that any preconfigured ComfyUI tools can be connected to the manager, with minimal Python coding.
Some examples of the most commonly used tools (LEFT - original image, RIGHT - after processing):
Categories of tools shown at the bottom of the screenshot above (expandable set):
In this project, a set of photorealistic jewelry images was created and then animated using neural network tool services.
This project involved creating and animating several cinematic backgrounds using neural network tool services.
A lip sync was also performed across several shots using a custom-built neural network tool.
Opening credits sequence for the feature film. This project involved creating some static concepts of a future city using neural network services. These concepts formed the basis for our team's production of photorealistic title sequence.
Visualization add-on for Swart Solution platform. This project was built from scratch using Unreal Engine 4.
Its purpose was to visualize the microcontroller's software structure to assist developers.
Other team members helped develop concepts and create 3D models.
Opening credits sequence for the feature film. In this project, I was involved in previsualization and camera preparation in Unreal Engine.
For this project, I built a fighting arena and lighting rig construction, as well as a model of a UFC belt.
