by Edan Meyer
This project provides a wraper over DeepDaze for generating artwork. It focuses on simplifying the process even further and adding more automation. It adds 3 things in addition to the base ability to generate images from text:
- Predefined preset for varying levels of quality
- Generation of videos and gifs from the images
- Automatic generation of text prompts for end-to-end image generation
$ pip install -r requirements.txt
Generate a random piece of art
$ python gen_cli.py
Generate many, low-quality random pieces of art
$ python gen_cli.py -p low -n 10
Generate a specific piece of art
$ python gen_cli.py -t 'A doge in a coin going to the moon'
DESCRIPTION
Generates art varying based on provided parameters.
FLAGS
--text / -t
Default: None
Prompt used to generate the image.
--text_file / -f
Default: None
Path to file with a list of prompts delimited by new lines,
and image for each prompt will be generated sequentially.
--preset / -p
Default: 'med'
Preset for the level of quality of the image generated, valid
presets are: ['test', 'vlow', 'low', 'med', 'high'].
--genre / -g
Default: None
A string giving the genre of art you want to be generated.
Leaving it as None will result in no specific genre.
--count / -n
Default: 1
The number of pieces to produce. If --text is passed in, n
variations of the same piece are produced, if --text_file is
passed in, n variations of each prompt are produced.
If neither is specified, n random pieces are produced.