Mastering Midjourney

From prompts to parameters, settings and weights, multi functional place

The most recent Midjourney update shows that the project keeps coming up with exciting latest feature implementations. This not only repeatedly expands our toolbox for AI images, but additionally sometimes makes it a bit difficult to maintain track of all of the functions and parameters

That’s why I created the “Midjourney Cheat Sheet”, a handy overview you may print out and put next to your workspace.

It covers all the essential settings and prompt parameters, and even points out some advanced concepts. Should you think something is missing, please let me know within the comments. 🙏

As all the time, here’s a text version in case you would like to work on it with ChatGPT:

Basic Prompt Anatomy: 
/imagine prompt: [PREFIX] [SCENE] [SUFFIX] [Parameters]
PREFIX defines image medium & style
SCENE defines content
SUFFIX modulates Prefix & Scene
(in actual practice these categories overlap)

Permutation Prompt Example:
/imagine prompt: cinematic shot of astronaut on {horse, turtle} --c {20,80}
...gets translated into 4 prompts:
->   cinematic shot of astronaut on a horse --c 20
cinematic shot of astronaut on a turtle --c 20
cinematic shot of astronaut on a horse --c 80
cinematic shot of astronaut on a turtle --c 80  
Basic Parameters:
--ar [WIDTH:HEIGHT]   (=aspect ratio)
--c [0-100]           (=chaos, unusual results)
--q [.25|.5|1]        (=quality/time spent generating the image, 1=default)
--seed [0-4294967295] (=start line for initial grid)
--stop [10-100]       (=stop at earlier percentage)
--s [0-1000]          (=stylize, artistic interpretation)
--tile                (=seamless patterns)
--iw [W]              (=sets image weight to W)
--no [X]              (=gives X a negative weight of -0.5)
--repeat [N]          (=repeat prompt N-times)
--video               (=create movie of image generation)Version Parameters:
--v [1,2,3,4,5,5.1,5.2] (=model version)
--style raw             (artistic fine-tuning for V5.1 & V5.2)
--style [4a, 4b, 4c]    (artistic fine-tuning for V4)
--test, --testp, --creative (=test models, legacy)Niji Parameters:
--niji / --niji 5     (anime trained model)Niji Style Options:
--style cute
--style expressive
--style sceenic
--style originalWeights:
/imagine prompt:       hot dog
hot:: dog
hot::2 dogOption Commands:
/prefer option set [NAME OF OPTION] [VALUE]
... e.g:
->     /prefer option set mycoolpreset dadaism --c 80
...now
/imagine prompt: an astronaut, --mycoolpreset
...becomes
/imagine prompt: an astronaut, dadaism --c 80/prefer option set mycoolpreset
...deletes mycoolpreset!
/prefer option list
... shows presets
/prefer option remix
... toggles remix mode
/prefer suffix 
... suffix so as to add to the top of each prompt
/prefer auto_dm
... mechanically send DM when jobs complete
/prefer variability
... toggle variability modeMore Commands:
/describe
/mix
/shorten/stealth & /public    (=toggle stealth mode)
/fast & /chill out        (=toggle fast mode)
/info                 (=show account info & queued jobs)
/settings             (=change bot settings)
/subscribe            (=change subscription plan)

Should you feel like diving into more advanced topics, listed below are some places to start out from: