Creative coding process
Chittu 13.14 is a purpose-built procedural generative art technology platform developed in-house by Chitrapata.
Preview of Chittu 13.14 closed beta release is available on an “invite-only” basis from March 2023. Visit the homepage and subscribe to our newsletter for regular updates.
Chittu 13.14 is primarily built using the Oracle Open Java Standard Edition Development Kit, with additional classes and aliased mathematical functions & operations for simplification. It is also enhanced with machine learning and artificial intelligence capabilities, including the ability to visualise convolutional networks, reconstruct images, synthesise textures, and transfer styles using neural networks. It will offer features such as makeover, super resolution, merging, patterns, texture, moods, feelings, aesthetics, restoration, proximity, style transfer, hairstyle, looks, and perspective.
We commit to open source our core codebase in the near future through public repositories. The details of the open source technology stack used are listed in the table below:
| Programming platform | Oracle Open Java Standard Edition Development Kit 19.0.1 | GNU General Public License |
| Integrated development environment | Editor Macros (Emacs) | GNU General Public License v3 |
| GPU API access library | Metal 3 | Apple Public Source License (APSL) |
| Defect tracking system | trac 1.4 | Berkeley Source Distribution (BSD) License |
| Statistical computing and graphics | R 4.1.3 | GNU Affero General Public License 3 |
| Mathematical libraries | NumPy 1.23.5, SciPy 1.6.0 | Berkeley Source Distribution (BSD) License |
| Visualisation and plotting library | Matplotlib 3.6.2 | Berkeley Source Distribution (BSD) License |
| Graphical library | Processing 4.0.1 | GNU General Public License |
| Code repository* | GitHub 3.7.0 | Creative Commons License |
| OS level virtualisation platform* | Docker Desktop 4.14.1 | Apache License 2.0 |
| Container orchestration system* | Kubernetes 1.25 | Apache License 2.0 |
| Data storage and processing framework* | lettuce - advanced Java Redis client | Apache Derby 10.16.1.1 | Hadoop 3.3.4 | Apache License 2.0 |
| Machine learning models | Deep learning, decision trees, random forest, linear regression, boosting, neural network, clustering and dimensionality reduction |
Recently, there has been a rise in use of large language and vision models in the art industry, which allow machines to generate art with credible and sometimes exceptional results. We have examined some of these third-party software from the perspective of a visual artist to understand their capabilities and limitations: DALL·E2, released by OpenAI in January 2021, is an AI system that can generate original and realistic images and art from a short natural language description. Midjourney, released in July 2022, is a research lab that explores new mediums of thought and expands the creative abilities of humans. Stable Diffusion, released in August 2022, is a text-to-image model developed by the CompVis group at LMU Munich using deep learning and latent diffusion, a type of deep generative neural network. Visit the “Text-to-image” page for further details.
We have accumulated substantial technical debt due to our internal inefficiencies and chaotic approach to learning. To address this issue, we plan to refactor our code and improve its design, structure, and implementation, while still maintaining its original functionality. This refactoring process will allow us to implement a Serverless Stateless Microservices architecture, which will enable domain-driven design, continuous delivery, platform and infrastructure automation, scalable systems, polyglot programming, and persistence.
