Ever wanted to get an endless supply of creative generations but running out of ideas, looking for new prompt inspiration, or just want to let a program handle the prompt wording?
This is a javascript project that generates prompts. The original purpose was to generate completely random prompts with completely random keywords because it was discovered that they can often turn out pretty good and creative however it has since evolved into giving you a wide range of control over prompt generation.
Click to watch video on Youtube
When it's properly installed and setup, you just run
webui.bat
You can now one-click export any animation to a seperate folder that can be used to dump into programs like FrameFlow on Windows or DAIN Google Collab amongst others to smooth out FPS from something like 5fps to 30fps, 50fps, or 60fps
Prompt: landscape, Starflower, Plum tree, vegetation, Marouflage, Vintage Photography, Ghost imaging, Focus Stacking, [desaturated look|Lens Flares], Background light, god ray, light shaft, volumetric lighting, wada kazu, yomu sgt epper, wata do chinkuru
Prompt: room, interrior, [Dining Room], grunge, rundown, broken floor, mold, shattered glass, ceiling hole, clutter, messy, furniture, accesories, Cabin building style, evening, myst, sunny, Industrial painting, Bracketing, Color Blast, Tiltshift
Prompt: portrait, princess, royalty, woman, Quiff Haircut, up-close, sceptor, crown, robes, castle, roadview, moat, courtyard, castle keep, castle wall, castle drawbridge, Brigham City, lake, Cottage building style, dusk, vegetation, Bamboo, ((drizzle)), [Al-Qatt Al-Asiri|Glue-size], popqn
Prompt: grimoire, heiwajima shizuo, nihonga, pokemon frlg, single elbow glove, saigado
Prompt: boxcar, defenseless, [indubitable::2], [portage|Sweet], Robert Childress, Ejnar Nielsen, Hubert Robert
Prompt: room, interrior, Basement, dirty, rundown, broken, torn wallpaper, shattered glass, holes, ceiling hole, clutter, messy, furniture, accesories, window, Golden Ratio, [Multiple exposure:Kinetic Photography:3], Miniature Faking, god ray, light shaft, volumetric lighting, Brian Mashburn
Auto-Image Generation
This program can connect to a Stable Diffusion WebUI running with the --api
command line argument and send generated prompts there to have them rendered
into images. It will then save the images into the outputs folder.
Auto-Image Upscaling This program can also automatically ask Stable Diffusion to upscale and crop the images if you prefer.
Dynamic prompts are a big part of this program, they are pieces of javascript code that dynamically generate a prompt. You can see them first-hand in the WebUI. (The images above were all generated with dynamic prompts)
Prompt: landscape, Gaillardia, Catharanthus, Clematis, Prickly pear cactus, overcast, Winslow Homer
Prompt: landscape, Ixora, Peach tree, evening, i.u.y
Prompts phrases frequently used can be added into an expansion file which can be expanded out nto a prompt. Expansion happen before any other dynamic prompt modules meaning any special dynamic words in an expansion will be further expanded.
Lists are the heart and soul of this prgram. A list is a text file with many lines. When refenced, one of those lines will be randomly selected. By default, referencing a list is guarenteed to pull a unique entry unless all entires have been exaughsted.
There's a wide range of default lists available from vocabulary words to artists to danbooru tags. You can ad your own custom lists in, some have already been made.
There are many settings, all of the mcan be overrideen by command-line arguments, but sometimes you want to have setting presets to save you time and effort. These are simple script files that can add-in settings and even potential randomization.
These are all accessible in the WebUI
If this is not enabled, images above 512x512 will have high-res fix applied wrong and may fail to generate
This is how this program talks to Stable Diffusion, if the --api flag is not set this program cannot talk to Stable Diffusion and you will not be able to auto generate images.
- Make sure git is installed
- Make sure NodeJS is installed, version doesn't matter. Link Here
- Make sure to run WebUI with the command line parameters
--api, this will be in yourwebui-user.batfile if your on Windows Very Important - Open a terminal in a folder of choice and run these commands
git clone https://github.com/junebug12851/random-ai-prompt.git
cd random-ai-prompt
npm install
node server
- Download files
- Extract to folder of choice
- Make sure to run WebUI with the command line parameters
--api, this will be in yourwebui-user.batfile if your on Windows Very Important - Open terminal inside folder and run these commands
npm install
node server
- If using git, i reccomend to use the
update.batif your on Windows, otherwise make sure you do annpm installafter agit pullin case new packages have been added. - If not on git, then download to a new folder and copy your
outputfolder over, i do not reccomend simply copying new files over old files, you may run into issues.
- You can double-click
update.batto auto-update - You can double-click
webui.batto auto-run Web interface
Cause I know javascript best, your welcome to make a python version, if you do, I'd appreciate a mention somewhere. I can even list it on here.
This would also be a good idea but noto nly do I not know python, I also really don't want to put in the work in re-writing this project and formatting it for Automattic1111 WebUI. Like above, your welcome to do this if you want and I'd love a mention somewhere.