Creative Artificial Intelligence. Index.
Audio, Visual Media, Music, Text generation. With Examples and Friend-Links.
Do you want it? A place where all potential creative tools are collected? With direct links and short explanations — for you to decide where to start your art? Well, here it is — an ongoing index of all creative uses of AI. At least those being covered by Merzazine. So without boring introductions…
Image and Video Generation
Google Deep Dream
What it does: Pattern recognition, object detecting, and re-interpreting of visual data. Link to my article: Deep Dream comes true Direct link: Google Colab Notebook
Example:
BigGAN
What it does: Trained on visual data, it generates new unique images and translations between images Links to my articles:
Direct link: Google Colab Notebook
Examples:
Style Transfer
What it does: Transfer of specific visual features between images Link to my article: Style Transfer (and you can do it as well) Direct link: Google Colab Notebook
Example:
Comixify
What it does: Summarizing impactful frames from a video into a visual storyboard. Video => Graphic Novel Link to my article: Comixify Direct link: https://comixify.ai/
Example:
Transforming of Matrix Trailer to a Visual Storyboard:
CycleGAN
What it does: Transfer of visual features between styles, in the absence of paired images. Link to my article: Visual Anarchist Realism of CycleGAN Direct link: https://github.com/junyanz/CycleGAN/
Example:
StyleGAN, StyleGAN2
What it does: Framework, based on iterative Generator/Discriminator process, able to be trained on various imagery datasets and generate HiRes images and videos. Including Projection as DeepFake detection. Link to my articles:
Direct link: Google Colab Notebook
Examples:
GauGAN: Creating with Segmentation
What it does: Using sketch tools you draw a picture. Every color is assigned to a specific texture. After conversion you will get a photorealistic (or/and abstract) image. Chose your style using various presets. Link to my article: 14 Deep and Machine Learning Uses that made 2019 a new AI Age. Direct link: NVidia AI Playground — GauGAN
Art Re-Generation: PainterByNumbers
What it does: Trained on 24k world art images, the system generates new art pieces. Link to my article: How to Train Your Artist Direct link: Google Colab Notebook
Examples:
WikiArt StyleGAN2
What it does: StyleGAN2 based model trained on WikiArt generates new imagery after categories: artists, genre, and style Link to my article: The Non-Treachery of Dataset Direct link: Google Colab Notebook
Examples:
ArtBreeder
What it does: BigGAN and StyleGAN2 based suite of models as an intuitive web application. Generates images and videos, upload functionality, social co-creation. Link to my articles:
Direct link: https://artbreeder.com/
Example:
DeOldify
What it does: Colorizing black&white images Link to my articles:
Direct link: Google Colab Notebook for Images and Videos
Example:
First Order Motion Model
What it does: Transfer of facial movements from video footage to a single image. Link to my article: First Order Motion Model Direct link: Google Colab Notebook
Example:
3D Ken Burns Effect
What it does: Spatial 3D-visualization from a single image Link to my articles:
Direct link: Google Colab Notebook
Examples:
3D Photography Inpainting
What it does: Spatial exploration of a single image Link to my article: 3D Photography Inpainting: Exploring Art with AI. Direct link: Google Colab Notebook
Example:
Wav2Lip
What it does: Speech recognition and facial animating of a single image/video footage. Link to my article: Wav2Lip: Between Art and Deepfake Direct link: Demo-Version
Example:
Music generation
JukeBox
What it does: Trained on 1+ Mio songs, JukeBox creates unique audio-experiences Link to my articles:
Direct link: Google Colab Notebook / JukeBox.OpenAI.com
Examples:
AI-generated Playlist:
Text generation
GPT-2
What it does: Text generation with NLP model by OpenAI Link to my articles:
- Lakribuchi Trilogy,
- FridAI series,
- Kings and Queens. Tramedy in 5 Acts. Stories written by #AI,
- Donald Trump is an Android.
Direct link: Google Colab Notebook
Example:
GPT-3
What it does: Advanced multilingual pre-trained NLP-system, capable of manifold use cases. Trained on 570GB texts from the global human cultural heritage. Link to my articles:
- GPT-3: Creative Potential of NLP
- AI about “Human circle”, reincarnation, favorite food (and secrets),
- Love Letters, written by a Toaster
- 20 Creative Things to try out with GPT-3
- The Art to Start: Designing Prompts for GPT-3 (series)
Direct link: https://beta.openai.com/
Examples:
Tutorials
AI as a Movie Maker
What it does: Using AI-generated screenplay, visuals, music, and voices for short movies. Link to my article: AI as a Movie Maker Example:
Creating a Podcast with AI
What it does: Creating a podcast with GPT-2 / GPT-3, AI-generated voices, and AI-created music Link to my article:
Example:
Applications and suits
RunwayML
What it is: a suit from various ML/DL models and networks for audio, textual, and video generation My Article: coming soon
TokkingHeads
What it is: Combining Wav2Lips, StyleGAN2, GPT-3, Text-To-Speech, and First Order Motion model, this web application creates realistic facial animations. My article: Coming soon
Example:
