Hugging Face has recently introduced ‘Doodly’, a demo tool that enables users to turn their doodles and raw drawings into generated art with the help of language guidance.
Users can simply draw figures and objects then make them into a fully formed artwork just by typing in a text prompt describing what they would like to see.
In a recent tweet, Sayak Paul, a machine learning engineer, showcased a video demonstrating how fast and efficient the tool works.
Based on the preview example, the user typed in a text prompt elaborating the details and specification he would like to see. It quickly turned an unfinished bird figure into a realistic-looking owl in 4K.
Sayak Paul continued to share more examples in a twitter thread.
It is amazing to see how simple line drawings could be generated into colorful, detailed and exquisite artworks.
Reportedly, the demo uses two technologies, namely, the T2I-Adapter and SDXL:
- T2I Adapter – a machine learning model that enhances text-to-image synthesis.
- SDXL – an AI model that can generate realistic images.
Now, users can try out Doodly and turn their creative ideas into visual art.
In previous reports, Hugging Face released IDEFICS, a state-of-the-art visual language model that has the ability to process images and turn them into coherent text outputs.
Similar to Doodly, it has the AI capabilities to turn doodles into a full blown artwork in the form of an image.
Definitely, this demo exhibits how Hugging Face is continuously progressing in the AI industry as they consistently build products that are not only helpful and easy to use, but advanced.