Midjourney is a paid monthly subscription model where you get a certain amount of "fast" renders per month for your subscription, which after used, you swap to a "slow" render for your image outputs. Currently Midjourney is only available on the official Midjourney Discord server , or via your own server using the official Midjourney Bot if your server has less than 30K members. Here's the official Midjourney website . (Note: You have to login and authorize/connect your Discord to sign into the Midjourney website.)

All additional parameters must be added at the end of your text prompt to ensure that the render or job is usable.

--version `version number` or --v `version number` Will force Midjourney to use a specific released/final version for the prompt. For example, using --v 2 will use Midjourney Version 2 on the prompt that you run. Acceptable version numbers are 1, 2, 3.

--beta Will force Midjourney to use the new experiemental algorithm. Will take longer to run and only produces 2 outputs instead of 4 (or 1 for non-square resolutions/aspect ratios). Does not work with --hd, --stylize, or --q `value`.

--aspect `ratio` or --ar `ratio` Makes it so the generation will have an aspect ratio as close as possible to the desired one. It can be a reduced ratio, such as 3:2, or it can be reducible, such as 1920:1280. If not specified, images will have a square aspect ratio, equivalent to --ar 1:1.

--w `integer` Will set the width for the resulting rendered image. Typically used in conjunction with --h. For best results, use --ar.

--h `integer` Will set the height for the resulting rendered image. Typically used in conjunction with --w. For best results, use --ar.

--hd Alters the Midjourney algorithm for a high definition render.

--no `item to remove` Attempts to exlude or remove the items following the command. This is similiar to giving it a negative weight. Example: vivid snowy mountain --no clouds will attempt to render the vivid snowy mountain without any clouds in the render.

--stop `value 10-100` Will stop at an earlier percentage through the generation process. Must be a value between 10 and 100. Example: vivid snowy mountain --stop 60 will stop the render 60% through the generation.

--uplight Will draw with more light, less shadows and dark spots.

--seed `integer` Can reference a previously generated seed in an attempt to keep things more consistent and in-line with the provided seed. The seed value must be an integer.

--sameseed `integer` Sets the same seed across all images of the rendered grid in an attempt to minimize any changes between the grid images. The sameseed value must be any positive integer.

--stylize `value` or --s `value` Sets the same seed across all images of the rendered grid in an attempt to minimize any changes between the grid images. The sameseed value must be any positive integer. Default is 2500.

--quality `value` or --q `value` Sets how much rendering quality to use/spend when generating the rendered image. Higher values take more time and cost more and lower values than the default cost less. Can use decimaled values. Default is 1. Example: vivid snowy mountain --q .5 will render the image at half the default quality.

--video Saves a progress video that can be DM'd to you via the Midjourney imagine bot after you react to the resulting/rendered output image with the basic envelope emoji.

--chaos `value 0-100` Sets the variability, randomness, and "difference" the rendered result will be. Must be between 0 and 100. The higher the value, the more "interesting" and unusual the rendered result, but possibly less reliably usable.

--fast Causes the image generation to be rendered faster, with less consistency, and less quality. Use --q `value less than 1` for similar results.

--iw `value` Uses the image prompt value as the weight against the text prompt weight. Use a decimal value. Default is .25. Example: vivid snowy mountain::2.5 puffy clouds::.5 will "rank" the vivid snow mountain portion of the prompt much higher than the puffy clouds portion of the prompt.

::`value` Will set that portion of the text prompt to the designated weight versus the rest of the prompt. Example: vivid snowy mountain --iw 1 will "rank" the image prompt portion of the prompt higher than the text prompt portion.

--wallpaper Will attempt to render your text prompt into a "wallpaper-worthy" image. Note that this will not always work as intended and generally will deviate from your text prompt and go wild with creativity.

The words/text you wish to see in the image must be the first element in your prompt and also repeated in the rest of the prompt context.

The text must be separate from the rest of the prompt with two colons ::

A physical format must be specified/declared in the prompt where you wish the text to appear on. For example, a painting on the wall, book, movie theater sign, etc.

If the aspect ratio is not big enough, the text may not be generated in the image.

Example: "Wand Sale!":: a shop sign showing "Wand Sale!" hanging on a glass door to a shop in Daigon Alley --aspect 16:9

You can use a single hyphen - to denote that a multiple word value is one value. Example: Killer-Instinct will help the generator evaluate this part of the prompt as "Killer Instinct" instead of "Killer" "Instinct".

The order of your styles of your prompt will have an effect on the output. Example:
a mouse wearing a baseball hat, intricate, synthwave
will produce a different image than
a mouse wearing a baseball hat, synthwave, intricate
The former putting more emphasis on an 'intricate' render where the latter will put more emphasis into the 'synthwave' factor of the output.

Punctuation can have an effect on the output. Example:
a mouse wearing a baseball hat, intricate, synthwave
will produce a different image than
a mouse wearing a baseball hat synthwave intricate
It's estimated that this is because the generator will interpret the latter as 'one idea or scene rather than multiple ideas/scenes.

Stable Diffusion is a paid or free model. If using the DreamStudio Web App or the Stable Diffision Discord bot it will use your credits. Credits are purchased and used per render. The amount of credits used depends on some of the settings and variables you use for your render. E.g., a higher pixel widgth and height will cost more credits for that image. If you want to run your own Stable Diffusion generator you can do so for free either locally on your machine or on a Google Collab Notebook (Google Collab has free usage or it's own subscription if you wish to take advantage of some of those perks.) Stable is available on the Stable Diffusion Discord server , the DreamStudio Beta Web App , locally , or on their Google Collab Notebook . Additionally you can run the it on your own Google Collab Notebook.

Specifies the width of the output render. More width will cost you more credits per render. Default is 512px, range is 512px-1024px.

SD Discord bot command: width:`value`

Specifies the height of the output render. More width will cost you more credits per render. Default is 512px, range is 512px-1024px.

SD Discord bot command: height:`value`

The cfg scale is the variable which controls how close the renderer will try to match your prompt. The higher the value the closer the renderer will try to match your text prompt for the output. Higher values should not increase the per render credit cost. Default is 7.

SD Discord bot command: cfg_scale:`value`

The steps value is effectively how many iterations the renderer will go through to sharpen and form the image. The higher the step count the more credits per render. Default is 50, range is 10-150.

SD Discord bot command: steps:`value`

Number of images that will be generated off of your prompt. The higher image count will cost you more credits per render. Default is 1, range is 1-9.

SD Discord bot command: number:`value`

Specify a seed number to "ground" different rendered images to try and get a similar result render output. Using a seed will not increase render cost.

SD Discord bot command: seed:`value`

SD allows you to use parenthesis (()) and brackets [[]] to increase or decrease the weights of parts of your prompt.

Example: a single ((apple tree)) in the middle of a meadow, (((sunset))) background, white fence, [[cows grazing]].
apple tree will have more weight than text not in parenthesis. sunset will have more weight than apple tree. cows grazing will have a decreased weight.

Using [[]] is not the same as a negative weight, it's just a decreased weight comparatively.

The order of your styles of your prompt will have an effect on the output. Example:
a mouse wearing a baseball hat, intricate, synthwave
will produce a different image than
a mouse wearing a baseball hat, synthwave, intricate
The former putting more emphasis on an 'intricate' render where the latter will put more emphasis into the 'synthwave' factor of the output.

Punctuation can have an effect on the output. Example:
a mouse wearing a baseball hat, intricate, synthwave
will produce a different image than
a mouse wearing a baseball hat synthwave intricate
It's estimated that this is because the generator will interpret the latter as 'one idea or scene rather than multiple ideas/scenes.