!dream prompt
to create images
/dream prompt
as above
Heres an example request with most options used manually
!dream a redneck Vitalik+ man wearing a tank top, sunglasses, drinking beer, doing a thumbs up gesture, dancing, laughing at his house being on fire, high quality, detailed, a beautiful retrowave artwork, trending on artstation withLora(vitalik,0.8) withLora(add_detail,1.2) [<neg-sketch-2>] --width 697 --height 929 --steps 30 --seed 951987030 --scale 7 --sampler ddim --model degenerate526urpm --hires_fix --upscale_level 2 --upscale_strength 0.5 --gfpgan_strength 0.5
Simply using !dream
on its own will generate a random image with the default settings
You can also just @ mention the bot in place of the !dream
command.
!background
and attach image to remove background
!crop
and attach image to automatically remove tranparent pixels
!expand
and attach image to expand the image with transparency (expandsides/expanddown/expandup/expandleft/expandright also available)
!fade
and attach image to apply gradient transparency to edges
(fadeleft/faderight/fadeup/fadedown also available)
!text your text here
and attach image to add text to an image (alt: !textTopLeft/Top/TopRight/Left/Centre/Right/BottomLeft/Bottom/BottomRight)
!metadata
extract exif metadata from an attached image
!avatar @username
display full size discord avatars of all mentioned usernames
!recharge
show credit recharge information
!help
show arty github and intro post with usage information
!lexica prompt or image url
Displays some results from lexica.art for your query, can analyse image urls and show prompts with a similar style
!split columns rows
and attach image, to split the image into sections (broken, WIP)
!models
show all the currently installed diffusion models
!imgdiff
and attach 2 images to show the visual difference
!embeds
show all the installed Textual Inversions and LORA's with usage instructions
!randomisers
shows all of the available text files in the \txt\ directory
!meme animate
attach images to animate them together
!meme blink
same as above
!meme blur
attach image to blur 10%
!meme greyscale
attach image to make it greyscale
!meme invert
attach image to invert colors
!meme flip
attach image to flip vertically
!meme mirror
attach image to flip horizontally
!meme rotate [degrees]
attach image to rotate, 90 degree default if no degrees are added
!meme animateseed 123
(animate images with a specific seed, add prompt text to image, bugged on certain prompts)
..
2 dots and a space will allow you to modify an existing render, follow it with additions to the prompt or !dream parameters . All other settings from the render you reply to will be kept, except for hires_fix and upscaling options.
inpaint prompt
will use the replied image as a template, use the inpainting model and inpaint your prompt. If the image has transparencies, those will be filled. If it does not, use the --mask face
to automatically select areas matching your search (face) and replace them with your prompt. If width and height are not set, they will automatically match the input image.
template prompt
will take the image you reply to, set it as a template and set the matching width/height . You can follow your prompt with any of the usual settings from !dream . If width and height are not set, they will automatically match the input image.
seed
replies with an easy to copy seed number
info
replies with the full !dream command to recreate the image you reply to
samplers ddim plms
will rerun the render using whatever samplers you choose, 1 by 1
models model1 model2
will rerun the render with whatever model names you give it
embeds search1 search 2
will search all the textual inversions and LORA's for your search terms, and test them all 1 by 1.
If anyone adds positive emojis (😂👍⭐❤️) to a render result, it will be copied over to the servers Star gallery channel (if configured in dbGalleryChannels.json)
If someone adds an envelope ✉️ emoji it will be DM'd to them
If the creator or bot admin adds a 🙈 / :see_no_evil:
it will move the render to the servers NSFW channel (if configured in dbNFSWChannels.json)
If the creator or bot admin adds a negative emoji (👎
!wipequeue
the nuclear option, delete the full job queue (old and new) and start fresh
!queue
shows queue status
!cancel
sends cancel command to invokeai backend (WIP, buggy)
!pause
sets bot status to DND, if queue is empty new jobs will not trigger till unpaused
!unpause
!checkpayments
triggers an immediate check for new payments to hive account
!restart
quits the bot process, trigger an automatic restart
!credit 123 @username @user2
manually give credits to mentioned accounts
!creditdisabled
temporarily disables the credit system for free renders
!creditenabled
enables credits
!say channelid message here
broadcasts a message as arty to channelid
!guilds
dumps the current guild/server information to console log
!leaveguild guildid
exit a specific guild id
!getmessages channelid
collects a backlog of messages in a specific channel to log (WIP)
!updateslashcommands
triggers a manual update of the application slash commands
!deleteslashcommands
manually deletes slash commands (to be used with above)
!text your text here
to add text to an image (alt: !textTopLeft/Top/TopRight/Left/Centre/Right/BottomLeft/Bottom/BottomRight)