Table Of Contents
Click on a category below to jump to that section, or copy these links to share with someone else.
- Prompter → A free tool to assist and inspire prompt building.
- MidJourney → A text-to-image artificial intelligence, that is awesome.
- Beginner Tips → Useful info everyone starting out should know.
└ Signing Up (Discord and Subscriptions)
└ Pro Tips (what to know when you start)
- Workflow → Some methods I use to create with AI.
└ Additive, Poetic, Images
- Discord Commands → Prompting, settings, and more.
└ Beginner Commands, GPU Hours, Prompting, Privacy
- Discord Interface → How to navigate the main interface.
└ Initial Options (re-rolls, variations, and upscales)
└ Upscaling (options and quality)
└ External Upscale (upscale using other software)
└ Remix (edit prompts or switch models)
└ Remaster (convert V1/2/3 into Test Creative)
└ Emojis (delete images, get seed numbers, and star images)
- Prompt Writing → Things to think about when designing prompts.
└ Main Idea
└ Descriptor, Camera, Lighting, Color
└ Artist, Movie, Game
└ Weights (::) Separators (,)
- Image Prompts→ Use your own images as prompts.
└ Uploading, Image Weight, Image Mixing
- Parameters → Fine tune your prompt with quality, stylize, seeds, and more.
└ MJ Algorithms (V1, V2, V3, V4), MJ+SD (test, testp)
└ Output Parameters
└ Quality, Stylize, Style, Chaos, Stop, Tile, HD, No
└ Seeds, Upscaling, Video
└ Frame Size (detailed examples in the Aspect Ratios section)
- Aspect Ratios → Change up your frame size.
└ Cinema, Phones, Photography
└ Super Wide, Super Tall
Prompter is my gift to the MidJourney AI creative community. What began as a personal project to better understand AI imagery since grew into a all-inclusive prompt builder and a multi-page visual notebook. With thousands of visits and a buzz on social media, I think we are onto something, so I plan to keep at it!
I created Prompter to assist, inspire, and ease prompt writing. A simple one-page user interface has everything you could possibly need. Select from common aspect ratios, various mediums, and even parameters like stylize, quality, chaos, and so much more… This project is completely free, though I appreciate all the donations and support many of you have given me.
MidJourney is a powerful artificial intelligence (AI) powered tool that turns text into imagery. You simply enter a “prompt” (a descriptive string of words) and the AI analyzes the language then outputs an image via diffusion methods. Though the datasets were trained on billions of images, what you get back in return is one that has never existed before.
Among other popular AI options (such as Stable Diffusion, Dall-E 2, Craiyon, NightCafe) what makes MidJourney stand out above the rest is its amazing amount of flexibility with multiple algorithms, quality settings, stylize, and chaos, and the fact that the company is so community driven with some of the most helpful, fun, and inspiring creators. You can also make literally thousands of images per month on the Standard Plan, something hardly possible on other AI platforms (unless you are burning that 3090 GPU).
The project is now in open beta! Sign up for a free Discord account if you don’t have one, then head over to the MidJourney website to join in the fun… To summarize, if you have never heard of or seen text-to-image AI - it will blow your mind. No lie. If starting out seems intimidating, don’t stress, just read through my helpful MidJourney Notes and become a master!
(click the prompt to copy it)
/imagine prompt: a pixar character design of a friendly mushroom creature, large eyes, happy, silly, friendly
An Alien's Skull
(click the prompt to copy it)
/imagine prompt: a dark dripping skull of an alien on display in a glass cube
Steampunk City View
(click the prompt to copy it)
/imagine prompt: an old vase with a plant growing out of it that is half steampunk machine with gears and wires, in the background is a window with raindrops and a scene of city buildings on a rainy day with lights in the windows, photorealistic, incredibly detailed, shallow depth of field
Rainy Day Corgi
(click the prompt to copy it)
/imagine prompt: an adorable little black tricolor corgi in a yellow raincoat holding a yellow umbrella in the rain
Grandfather Clock Cabin
(click the prompt to copy it)
/imagine prompt: in the foreground is a colorful autumn forest scene, a small gothic house grandfather clock, insanely detailed
(click the prompt to copy it)
/imagine prompt: photograph of a hurricane storm with monster fangs and claws, insanely detailed from above
This is not an official guide, just a passion project of mine. You can also browse the official MidJourney Documentation gitbook. That said, I like to organize things in different ways, so you might like my notes as well.
Discord → MidJourney runs through Discord, a popular chat application that (for meeeee) feels like you are in an awful 90’s chat room. That said, gamers love it, and a lot of people use it, so I’m probably finally just becoming an old grumpy man… If you do not yet have an account, hop over and sign up for free. It might be intimidating at first, but you’ll get the hang!
MidJoureny → Make sure you are logged into Discord then accept the MidJourney Invite to join. After your trial (which I am not going into detail in these notes, yet), I highly suggest you sign up for the $30 per month Standard Plan (or enter /subscribe in Discord and follow the link).
Subscription → Don’t do the $10 plan because you can’t use /relax mode (making images “free”) and you’ll use up all your GPU hours in a day or two. Seriously though. The $30 plan might seem like a high cost at first, but think about it - you can literally get thousands of renders in relaxed mode (fast mode is still limited). I made about 3,500 in my first month (mostly testing and playing) which is like $0.008 or so per image. That’s a steal, hence I think it’s silly when people complain about monthly payments.
Bot → If you don’t want all the clutter of a public channel while you work, once you subscribe you can chat with the MidJourney Bot and things won’t be so cluttered. It’ll just be you and all your work. Your images and prompts will still be public on the MidJoureny website, however, but you can pay more for Private mode, where you can publish or unpublish images (more about this later).
Servers → You can also create your own Discord Server, invite the MidJourney Bot, and do all your work in separate channels. I have channels for #projects, #testing, #help, and #images to keep things organized (yep, I upload all my photos for image prompts in one channel so I can find old photos quickly!). You can also invite your friends and make work together, but of course, they need to also be subscribed as well (otherwise one member could give free access to thousands, lol).
Relax → Immediately switch to relaxed mode by entering /relax in Discord. I suggest doing all your work in relax mode, then switch to /fast only for max upscales (which you can’t do in relaxed mode) or if you are really in a rush. Relax images are free, so you get infinite renders!
Shhhh → Don’t make stupid posts on Facebook and stuff about how you don’t think AI is art or some crap. It’s an unoriginal boring platitude and just makes you look bitter and dull. The second you realize literally everything is art you’ll be a much happier person… Move on, and make new things. ✌
I have two main workflows, and I really enjoy both ways of working. Basically, in one I spend a lot of time developing the prompt and testing it piece by piece until I get what I want, and in the other I just let the AI take the wheel.
Get a basic idea in your head, and don’t be dead set on what you imagine, but let AI guide you. How about a city being taken over by aliens?! So type in your main idea to set the scene then run it. See what it looks like. Now add to it. Run it. Add more. Run it. Keep testing to see where the idea is going.
/imagine prompt: a city street
/imagine prompt: a city street full of tall buildings
/imagine prompt: a polaroid photograph of a city street of tall buildings
Is something not right? Remove terms. Want more? Change terms. Once you get a really awesome scene going, you can try adding things to it. Are ideas going away? Add weights (learn more about that HERE). Really want a certain style? Use a photo URL in the prompt and try to get it to come through (learn how HERE). Maybe you start to care less about the alien theme because you got a new idea from the evolving scene. Go with it! So now we might have something like this in the end, then Remix to V4 (learn about remix HERE).
/imagine prompt: polaroid photograph of a city street:: tall buildings::5 night::3 laser lights::8 in the style of the video Blade Runner --quality 5 --stylize 1500 --chaos 25 --ar 16:9
As an artist, many times I look at AI as a form of television. It’s entertainment for me, but I am in charge, and the controller has a few billion buttons! I enjoy typing random things, but many times the results are amazing and might inspire bigger ideas. Try simple juxtapositions, lyrics from a Grateful Dead song, or pull a quote from the Bhagavad Gita and run a few variations. Let’s see what these look like in Version 4.
/imagine prompt: half man half clock
/imagine prompt: in and out the window like a moth before a flame
/imagine prompt: I am become death, destroyer of worlds
Late one night (possibly after a nice glass of Scotch, or something) I typed in a bunch of complete gibberish into MidJoureny, not even real words, and got some of my favorite images to date. When you let the AI go crazy, you get some really interesting images. Also, a fun tip is to add a lot of stylize (like --stylize 50000) or chaos (--chaos 100) just to see where things go. Have fun!
Prompting with text is fun, but images in Version 4 can take you to a whole new level. You can upload a photo of yourself to try to get a selfie standing on the moon. Or think functionally. One idea I had was to upload a single letter of the alphabet at a time and create some complex fonts with each. Lastly, feed a MidJoureny image back into itself (simply copy its URL link into the prompt), combine a few, or add some new words to evolve it. Read all about images in the Image Prompts section.
/imagine prompt: IMAGEURL a man in an astronaut suit standing on the moon, dark sky and stars in the background
/imagine prompt: IMAGEURL 3D rendering of the letter S with textured metal surface:: top down photography of a motherboard, wires, transistors, backlit glowing led colored lights, computer fans, electronics, against a black background::3
/imagine prompt: IMAGEURL1 IMAGEURL2
Below are some of the main commands that you enter into Discord. There are others that can be found in the main MidJourney guide, but I think these are the only ones you really need.
/help → Have no clue what the hell you are doing? Help might help! Entering this command displays a bunch of helpful information to get you started: Links to the GitBook, tips, and more.
/subscribe → Trial over and you still have the taste for more? This command sends you a link to bring you to the MidJoureny subscriptions page. Click select a membership option. As mentioned in my Beginner Tips, I highly suggest the $30/mo Standard Membership and then using /relax mode to get unlimited images!
/daily_theme → If you do not want to participate in the Daily Themes but get annoyed by those Discord notifications (I sometimes have issues with uncleared notifications), you can use this to turn those on or off. This does not work in a private chat or server, so just click one of the Newbies channels in the MidJourney Discord and type /daily_theme then No and it’ll turn off notifications. You can do the same with Yes to rejoin again… I think you might need to do this over again when the team makes big changes. I think.
GPU Hour Commands
/info → Shows user info, images made (aka you have an addiction problem), if relax or fast mode is active, GPU hours remaining, in progress items, and such.
/fast → Fast mode, quick results but they count against your hours, so be careful!
/relax → Slow mode, slower processing but doesn’t count against hours, basically infinite.
NOTE: With a Standard plan, stay in relax mode all the time, unless you need to make a max upscale or you are incredibly impatient, that way you get thousands of images “free”.
/imagine → Main code to submit a prompt, then you click “prompt:” and type in your image ideas after that. You are going to use this a few thousand times (unless you use Prompter, which automatically adds this so all you have to do is copy and paste!).
/settings → Adjust default settings like version, style, quality, relax mode. Careful because when you make a change it “sticks” until you go back into settings and change it. I prefer to use parameters in prompts like --version or --quality, because then they only apply for that single prompt, not all as default, but settings is still a super useful tool.
/show → This is super useful if you want to bring up an old image to run upscales or variations, or to bring an image into a new channel you are working in, you just add the job ID to this code. There are two ways to get a job ID:
(1) Discord - You can hit an image with an envelope ✉️ emoji, and the MidJoureny bot will send you a message with the seed and job ID.
(2) MidJoureny - You can click the three dots to open options, then click Copy Job ID.
Now back in Discord you type /show then paste the job ID (like this: /show dfd1d656-d5c5-4e3e-a1b6-420c1b52ad81) and hit enter. It will bring up the old image so you can keep working!
/private → If you pay for a private account, entering this command will make all of your work on the MidJoureny website private, only visible to you. Keep in mind, if you are in a public channel of Discord, even with /private mode turned on, people can see your work, so if you are a true paranoid - message with the MidJourney Bot or work on your own server. Since I test things a lot, my feed is full of boring clutter, so I got private mode and decided to “hide” most of it and just showcase a few images to the public.
/public → If you are in /private mode you can turn it off with this command, making your work visible to everyone. With /public turned on (or if you don’t pay extra for private) others can see your work, even if you are in a private server or chatting with the MidJourney Bot, because it will all show publicly on your MidJoureny profile.
Once you enter a prompt, completed images are sent back as a message in a grid, and read like a book, number 1 and 2 on top, and 3 and 4 on bottom. Then you can use various buttons to make variations or upscale.
↻ → Re-rolls the same prompt, new seed is used so all images will be different.
V1 → Make variations for the first image (V3 for third and so on), will make 4 more images based on the chosen one, so they will look similar but different.
U1 → Upscale the first image (U3 for third and so on), will upscale and might change some of the appearance, for better or worse, but that is where the next upscale options come into play.
Emoji Buttons → Rank Options appear after you upscale an image once. You can simply click an emoji face reaction (😖😞😀😍) to rate your image, and it will show on your MidJoureny gallery. This is useful if you have a few images you really like, or dislike.
/imagine prompt: a long hallway inside a futuristic space station --ar 3:2 --v 4
Upscale to Max → A final upscale that is only allowed in /fast mode. Makes image larger, adds details, but makes things a little weird for some images (you’ll likely see what I mean), in which case you can use the below upscale options:
Light Upscale Redo → If you aren’t happy with the upscale, this will redo it with less detail changes and look more similar to the previous image, however the image is smaller in resolution.
Beta Upscale Redo → New algorithm for upscaling, smooth like the Light Upscale, but with more details, and also much larger resolution.
Detailed Upscale Redo → This is basically the standard upscale, but the button shows up if you did a Light or Beta Upscale, basically meaning to go back to the original upscaler, adding details and such.
External Upscaling → You can also download an image and use external software to upscale them yourself. The benefit to this is that it will not change the style, just the size, whereas some MidJourney upscaling alters the look slightly.
I use Visions Of Chaos (which is an amazing program if you have a killer computer, constantly updated, and worth donating to the developer), but HERE is a great link for other options of AI upscaling apps, and I will add more in time.
AI Upscale with Visions Of Chaos
(MidJourney on the left / Visions Of Chaos upscale on the right)
Remix → An exciting new feature that allows you to alter your prompt while making variations. To turn this feature on or off just enter /prefer remix or /settings then select Remix Mode. After you get your initial grid of images, the variation buttons (V1, V2) will allow you to edit your prompt. A box will pop up with your prompt, and here you can add or remove things, change parameters, or even move to the --testp algorithm, for example. Have a portrait of a man but with it was a woman? Now you can change it!
1. You can either alter a prompt or delete it all and add new terms. For example, change “a magical forest” to “a magical forest at night” or remove the whole thing and just write “night” (which is usually a more drastic change). If you don’t see a nighttime scene show up, make a few re-rolls and see if it comes in (or use weights). Your last prompt will still have an effect on the new images, so if you say “fall” and then change to “winter” you might still get some fall foliage colors in the snow.
2. Changing the aspect ratio tends to “stretch” the image as it recreates it, especially something like a landscape or drastic shifts in size. That said, I have done some cool stuff going from really tall to really wide, and it will change where people are and add to the background.
3. Stuff like quality or chaos are only set in the initial roll, so for example, you can’t start without quality and then add --quality 5, nothing will change. Also, image prompts don’t work in TEST as usual, but you can run a V3 image prompt and then Remix to TESTP. You can’t add an image prompt via Remix though.
4. Since the codebase was changed, jobs before October will not allow you to run variations or remixes, so the simple fix is to hit the image with an envelope emoji, then the Bot will send you info. Copy the Job ID then type /show and paste the Job ID. This will bring the job back and variations and remixing will work!
5. As things change with this new feature I’ll update these notes.
Below is a visual example of how you can use the Remix feature to change your prompt while working through an idea, and sometimes end up somewhere completely different. This is a little random, but I just wanted to show you how a scene feels the same while changing characters, mediums, and versions. It’s actually pretty amazing.
1. a photograph of a man walking down a street --v 4
2. a photograph of a woman walking down a street --v 4
3. a painting of a woman walking down a street --v 3
4. a 3d render of a woman walking down a futuristic neon street --v 3
5. a 3d render of a woman walking down a futuristic neon street --v 4
6. a tintype photograph of a woman walking down a 1930s street --v 4
Remaster → Allows you to re-roll a V1/V2/V3 image into the new TEST algorithm, and adds --test --creative to your prompt. Since TEST has many limitations, like aspect ratio (only 1:1, 2:3, or 3:2), things may not be perfect, but it is a pretty awesome feature to play with. I love going through really old prompts and trying them out with a Remaster.
❌ Emoji → React to the image with the red :x: emoji to cancel and delete an image. This is a great to declutter your public MidJourney gallery while you work, but be aware that any image you delete is gone for good, on Discord and MidJourney. (I rarely use it because I am a disorganized slob)
✉️ Emoji → React to the image with the :envelope: emoji and the MidJourney Bot will sent you a DM with the seed, job ID, and video link (if you had --video in the prompt). This is a useful feature for three reasons, and I’ll talk about these in more depth later, but basically:
- If you like an image you can use -seed ### or --sameseed ### with the number given by the Bot and get (maybe) a similar output composition.
- If you use --video in your prompt you can download a generative video showing your image go from complete noise to final form.
- If you want to work on an old image, get the job ID from the Bot and use /show then job_id and copy and paste the job ID. The old image will enter and you can then Upscale, Remix, and so on.
⭐️ Emoji → React to the image with the :star: emoji to mark as favorite. Your image will be posted in the #favotires channel on Midjournery’s Discord. Then you can be a star.
With the onset of artificial intelligence, visual artists and creatives are now magicians of words, and we craft our imagery with letters, syllables, and descriptors. Okay, that kinda romanticized things a bit, but it’s it kinda true?
When creating Prompter, I broke down prompts into categories of descriptive terms, like medium, color, movie, or lighting. Below are visual examples of each category to show the basics of how you can alter and add to a prompt (and soon I will be adding links to visual examples of hundreds of terms, and Version 4 as well).
AI is so powerful that it is up to us to imagine interesting scenarios. You got this! Be creative. Be funny. Be deep. Be ourself. You can borrow descriptive parts of prompts from other people, but the main idea should really be yours. Have you been feeling a certain way and want to portray it in the language of imagery? Did you have a dream last night of something so surreal only art could show it? Perhaps you just want to make someone laugh by combining silly things. Maybe it was something from childhood, or something from the future! Anyway, the main idea is the special sauce of your prompt, and everything else just helps form and shape it.
/imagine prompt: a friendly fox
/imagine prompt: a treehouse
/imagine prompt: an evil monster
When I created Prompter, I broke down the language of a prompt into various descriptive themes, and then organized these themes into a database of terms. Below are the eight categories and what they do. Even if you don’t use my Prompter tool, this is a great way to think about building out the details of a good prompt.
Descriptor → This section is for other descriptive terms that don’t really fit into the other categories, just additional ways to alter your prompt. I include two descriptor sections in Prompter, but you can multi-select, so pick what is useful to you. You can make things more complex (fractal / insanely detailed), speak about time (1950s style / Low Poly), add elements (ice / fire), and so much more. Obviously, the list can be endless, but I provided enough in my database to get you inspired.
/imagine prompt: a friendly fox, fractal
/imagine prompt: a treehouse, low poly
/imagine prompt: an evil monster, ice
Medium → The medium defines the main style of the output and is usually the most important descriptor. I find that placing it before the main idea of your prompt usually works best. This term can be as vague as an art movement (relief carving / psychedelic art style), a specific type of flat output (ballpoint pen drawing / blueprint), dimensional output (cross-stitch / relief carving), or even types of photography if you are a camera geek like me (calotype photograph / expired polaroid).
/imagine prompt: relief carving of a friendly fox
/imagine prompt: blueprint of a treehouse
/imagine prompt: expired polaroid of an evil monster
Camera → The camera sets your point of view, your lens, and how the scene is spatially presented. You can go from one extreme (microscopy / macro lens) to another extreme (tilt-shift lens / satellite imagery), define focal traits (bokeh / motion blur), or even attempt to bring out the views in photography (50mm lens / 80mm lens).
/imagine prompt: a friendly fox, microscopy
/imagine prompt: a treehouse, satellite imagery
/imagine prompt: an evil monster, motion blur
Lighting → The lighting descriptor defines if there is a special aspect to light and shadow. You can use terms to describe really dramatic forms of light (contre-jour lighting / film noir lighting), a time of day (golden hour sunlight / sunset lighting), or certain forms of light (infrared light / laser light).
/imagine prompt: a friendly fox, contre-jour lighting
/imagine prompt: a treehouse, sunset lighting
/imagine prompt: an evil monster, laser light
Color → Sometimes when your prompt doesn’t contain a particular mood, the colors default to a blend of cool and warm, cyan and reds. If you really want a specific color style, mention it. You can be blunt (green / dark purple), describe types of color (neon color / vibrant color), computer displays (1-bit grayscale / RGB), or even really strange mixes (diffraction pattern color / tie-dye color).
/imagine prompt: a friendly fox, dark purple
/imagine prompt: a treehouse, neon color
/imagine prompt: an evil monster, 1-bit grayscale
Artist → As an artist myself, I find it really fun to see what different scenes look like by other artists. it is entertaining to mimic, or combine. There is so much art out there, everything from Art Nouveau (Gustav Klimt / Antoni Gaudí), over to Post-Impressionism (Vincent van Gogh / Henri Rousseau), or even some photographers (Edward Burtynsky / William Henry Fox Talbot).
/imagine prompt: a friendly fox by Jackson Pollock
/imagine prompt: a treehouse by Joan Miró
/imagine prompt: an evil monster by William Henry Fox Talbot
Movie → Adding a movie or TV show (I thought of calling it Film, but let’s just go with Movie), if the AI is trained on enough of the style, can really alter colors and elements. For example, some have a lot of drama (Dune / Spirited Away / Stranger Things) but others don’t work so well. I don’t use these often, but sometimes it’s fun. The new Version 4 works really well, and I’ll have updated images soon.
/imagine prompt: a friendly fox, in the style of the film Dune
/imagine prompt: a treehouse, in the style of the film Spirited Away
/imagine prompt: an evil monster, in the style of the film Stranger Things
Game → I’m like half-gamer(ish). I grew up with Atari, NES, GameBoy, but a few (Quake / Doom / Zelda) were my favorites, many years ago. I would design levels and host deathmatches in Quake 1. In college I got everyone hooked up for LAN battles on Quake 3. It was epic at the time (a time when phones could hardly take pictures and streaming a movie was unheard of). Anyway, I enjoy some VR from time to time now, but gaming will always come in waves for me. Combining video games with AI is really fun (and even better V4 images).
/imagine prompt: a friendly fox, in the style of the video game Quake
/imagine prompt: a treehouse, in the style of the video game Tetris
/imagine prompt: an evil monster, in the style of the video game Zelda
Double Colon (::) → A double colon is a hard break to split up parts, also known as a multi-prompt, or also used to add weight (see below). This is actually a super-useful tool to break up one thing from another in your scene.
Weight (::2) → You can add weight, more focus or importance, to ideas using the double-colon right after them (blue::2), the higher the weight, the more the AI will focus on that idea, and negative weights (::-0.5) will get less or no focus (similar to the --no command, however, negative weight works better in V4 so use ::-0.5 for those prompts). Weights seem to be relative to each other and the whole. See examples below:
/imagine prompt: clock::5 clouds::2 green::-0.5
Clock gets most focus, less on clouds, and likely create images that are not green.
/imagine prompt: clock, clouds::2 green::-0.5
Commas do not separate an ideas, so in this case clock and cloud are weighted 2.
/imagine prompt: clock:: clouds::2 green::-0.5
Here I added a :: to split up clock and clouds, now clock is 1 and clouds is 2.
/imagine prompt: URL1.jpg ::2 URL1.jpg ::2
Lastly, you can use :: to weight multiple images differently, but remember to add a space before the :: separator.
This is a strange one. A lot of people format their prompts in various ways, using commas to separate ideas, periods, plus signs, and so on. I have done some testing and found that it does indeed change the image, but not really in a meaningful way. Basically, things are similar but different, but not in some logical way where you can say “this is the best format”.
Comma (,) → A comma is a soft break in an idea, however, I have tested things with and without commas and saw little logical difference, so maybe this is just a good way to appease our brains so things look more organized.
Single Dash (-) → A single dash is said to be a good way to connect two or more terms into one idea. So while we all know “grandfather clock” is a big fancy clock, AI might think an old man, and a clock (maybe not, but just as an example). So you could type “grandfather-clock” to tie the words into one idea. That said, some test
Others → Sometimes I see others using periods, brackets, parenthesis, and other odd punctuations in their prompts. Again, I don’t see any logical reason to add more punctuation when it doesn’t really steer the image. I assume the AI pretty much ignores them anyway. Maybe skip that unless you like to hit your keyboard more than you need to. I usually run tests when I see someone doing (((this))) to see if it does anything, and usually, it doesn’t. Maybe they are just getting their AIs mixed up… Either way, below are a bunch of tests using the same seed so you can compare grids for yourself:
In Version 3 and below, MidJourney uses image input for style, color, and so on - so don’t expect to use a photograph of something and get a similar image in return. That said, the new Version 4 algorithm is amazingly powerful for image prompts! Though sometimes hit or miss, you can do a lot with them. I sometimes have luck with faces (like my self-portraits below) however I really find image prompts to be useful to define the look of a landscape or altering the design of the output.
/imagine prompt: IMAGEURL1.jpg IMAGEURL2.jpg --iw 2
/imagine prompt: IMAGEURL1.jpg ::2 IMAGEURL2.jpg ::8
Since image prompts use URL links, the easiest way to use an image is to send a JPG to the Bot, then click the image, click Open Original, then copy that URL into your prompt. If you really want parts of the image to be carried over use weights, and maybe describe the specific details in your prompt after the URL.
You can also feed a MidJourney image you make back into an image prompt. Do the same thing, click on the MidJourney image, Open Original, then copy that URL.
--iw → Add this parameter to the end of your prompt to add image weight. Use a decimal value (1 for full strength, 0.25 is default, 0.5 for 50%, or even high numbers to really focus on the image over other parts of the prompt that might be higher weighted. As mentioned above, if you want to weight multiple images differently, you can use the standard :: feature, but add a space after the URL.
Be aware, Version 4 is not yet compatible with weights, but it will be soon. Also, if using just a single image with V4, you’ll need to add a text prompt as well.
Another really powerful feature is the ability to add multiple images to a prompt without any text, therefore mixing them into a new image. This can be fun, but also functional. For example, you can mix two photographs to combine elements (like the example below with mountains and boats) or you can attempt to alter the material or style of an image by combining it with a specific look (like changing the white rabbit below to metal or wood).
Parameters are little codes you can add to a prompt to change various things from the size to the algorithm version used. It is best practice to put all parameters (the terms below) at the end of a prompt. If you are using my Prompter tool, you can simply check boxes and enter numbers, so no need to really know all of them by heart.
--version 1 → Uses the first OG algorithm, good for macro and textures. (can also use --v 1)
Limitations: Not compatible with Stylize or HD.
--version 2 → Uses old algorithm from version 2, more abstract. (can also use --v 2)
Limitations: Not compatible with Stylize. HD is not compatible with Quality values above 2.
--version 3 → Artistic and creative. (can also use --v 3)
Limitations: HD is not compatible with Quality values above 2.
--version 4 → New model, powerfully smart and complex, and incredibly with image prompts. (can also use --v 4)
Limitations: (Version 4 is in alpha testing, so many things do not work now, but they will be added) Not compatible with HD, Tile, Video, SameSeed. Quality must be between 0.25 and 2. Style can be set to 4a or 4b (--style 4a). Stylize goes from 0 to 1000 (default 100). Not compatible with separate image weights. Single image prompts are not allowed (need 2 or one with text also). Aspect ratio 1:1 or 2:3 or 3:2 only.
/imagine prompt: a waterfall --v ___
/imagine prompt: a portrait of an explorer --v ___
(waterfall / explorer)
(waterfall / explorer)
(waterfall / explorer)
(waterfall / explorer)
Test And Testp (a mix of MidJoureny and Stable Diffusion)
--test → (temporary beta testing) New algorithm, general artistic mode, more coherence. (there are many limitations, but since this will change by the hour I am not adding details here)
--testp → (temporary beta testing) New algorithm, photo-realism mode, more coherence. (there are many limitations, but since this will change by the hour I am not adding details here)
--creative → (temporary beta testing) Use with --test or --testp to make the image more creative and chaotic.
Limitations: (TESTand TESTP have many limitations but I am not going to list everything here. Personally I feel like TEST might be gone eventually, and Version 4 is where it’s at. Anyway, aspect ratio is limited to 1:1, 2:3, and 3:2. You wont get a warning but it’ll limit to those sizes. Also --stylize can only go from 1250 and 5000.)
/imagine prompt: a waterfall --test
/imagine prompt: a portrait of an explorer --test
/imagine prompt: a waterfall --test --creative
/imagine prompt: a portrait of an explorer --test --creative
(waterfall / explorer)
(waterfall / explorer)
(waterfall / explorer)
(waterfall / explorer)
--quality → Spends more time on details (--quality 2) (0.25 faster, 1 default, 5 super high)
/imagine prompt: a photograph of a circus tent at night with street lights --quality ___ --sameseed 420 --v ___
--stylize → How artistically stylized images are (--stylize 800) (650 off, 2500 default, 60000 crazy)
/imagine prompt: a stone pathway winding through the woods --stylize ___ --v 3
/imagine prompt: a field of tall flowers with storm clouds above --stylize ___ --v 4
--style → When using the Version 4 model, you can switch between the old 4a style and the newer 4b style settings. (--style 4a) (4a or 4b currently available)
/imagine prompt: a watercolor painting of a man in a hat standing in a rainforest at sunset --style ___ --v 4
/imagine prompt: art nouveau style of a jellyfish floating in an industrial tank --style ___ --v 4
--chaos → How random and abstract the output is (--chaos 85) (0 to 100, higher is chaotic)
/imagine prompt: a skyscraper with snow-capped mountains in the background --chaos ___ --v 4
--stop → Used to stop image generation at a percent (--stop 50 to end 50% through). This is great if you want to quickly test some ideas, and maybe see the colors or compositions. Be aware that when you upscale an image that is all blurry, it will no longer be “stopped” in the process.
/imagine prompt: an old turntable next to a lamp with vintage wallpaper in the background --stop ___ --v 4
--tile → Creates a seamless texture that you can tile as a pattern (works best with TEST).
/imagine prompt: old colonial style buildings, city streets, palmetto trees, rivers, and beach drone view ultra detail cartographic map, fantasy game, final fantasy style, 2d, icons and symbols, anime studio trigger style, cartoon style, illustration --tile --testp
/imagine prompt: a city --tile --testp
--hd → Algorithm “add-on” (compatible with V2 or V3) that’s better for large images, with a less consistent composition, but good for abstracts.
/imagine prompt: a photograph of a ship in the ocean --v 3 --hd
/imagine prompt: a painting of the surface of the moon --v 3 --hd
--no → Do not include something, same as using a weight of -0.5 (--no trees would avoid trees or --no blue would avoid blue, but you can also use weights like blue::-0.5). My research has found, so far, that in V4 the --no parameter doesn’t work as well as a negative weight, so that might be the way to go.
/imagine prompt: a forest --no green
/imagine prompt: a forest:: green::-0.5
--seed → Variations use the same style, number from 0 to 4294967295 (--seed 1234). This feature spreads the same noise patterns across all 4 images in the grid. This means that if you run a prompt with a specified seed a few times, image 1 should match image 1 on each grid (same with image 2 and 2, 3 and 3, and so on).
--sameseed → All images will use the same seed, to look similar (--sameseed 1234). This feature spreads the same noise pattern across all 4 images in the grid separately, meaning they will all have similar traits. This can be useful if you like a certain composition and want to have many variations of it.
NOTE: Keep in mind that in V3 and below, seed images are similar but still different, so you can never get an exact match. In V4, however, a seed is precise, meaning you can run a prompt with a seed many times and the images will be 100% identical (SameSeed is not compatible with V4).
/imagine prompt: a rainbow over a waterfall in the rainforest --seed 420 --v 3
/imagine prompt: a rainbow over a waterfall in the rainforest --seed 420 --v 3
/imagine prompt: a glass sphere with a desert inside, a tall saguaro cactus --seed 420 --v 4
/imagine prompt: a colorful glowing mushroom in the forest --sameseed 420 --v 3
If you know you want a specific type of upscaler, you can skip a step (saving GPU hours) by specifying the upscaler in your prompt. So if you add --uplight to a prompt and click U4, for example, the upscale will be UpLight instead of the default Detailed Upscale.
--uplight → Upscale light, results are closer to the original, smaller file though.
--upbeta → New algorithm for upscaling, smooth, more details, and larger final resolution.
The video command only works for the first 4-up. Once the image is finished give it a ✉️ emoji and the Bot will DM you the images and video to download. If you do a regular quality image, the video is short and fast (7 seconds), however, if you include --quality 5 or something, you get a really long video (like 34 seconds). Just make sure to stay in /relax mode otherwise higher-quality settings will eat up a lot of GPU hours.
--video → Saves a progress video in DM, need to react with a ✉️ emoji too!
Quality 1 with Video
(version 3 grid)
Quality 5 with Video
(version 3 grid)
--aspect or --ar → Used to add an aspect ratio that isn’t a square (--ar 16:9 for a wide output)
--w and/or --h → Similar to aspect, you can use these to set width and height. (--w 1000)
/imagine prompt: octane 3d render of an alien landscape at sunset, volumetric lighting, green, red, purple --v 4 --ar 1:1
/imagine prompt: octane 3d render of an alien landscape at sunset, volumetric lighting, green, red, purple --v 4 --ar 3:2
/imagine prompt: octane 3d render of an alien landscape at sunset, volumetric lighting, green, red, purple --v 4 --ar 2:3
Images in MidJoureny are square by default, however, you can use parameter commands (--ar, --w, --h) to alter the aspect ratio of your images. You can match screen sizes, use 9:16 for a story on social media, or get REALLY wide or tall with them. In Prompter, I broke down the ratios into categories for Cinema, Phones, Photography, and more. For visual examples of each of these, check out the chapter on Aspect Ratios.
4:3 → Close to early television and film (--ar 4:3)
16:9 → Modern HD standard for televisions, computer monitors, phones (--ar 16:9)
2.39:1 → Close to anamorphic widescreen, widest cinematic, landscapes (--ar 239:100)
1:1 → Square, Instagram, default in MidJourney (--ar 1:1)
9:16 → Vertical HD, standard for most phones (--ar 9:16)
9:19.5 → Close to vertical for new iPhones (--ar 90:195)
3:2 → Widest view - prints at 4×6, 8×12, 16×24, 20×30, 24×36, 30×40 (--ar 3:2)
2:3 → (same as above, but vertical) (--ar 2:3)
5:4 → Less wide - prints at 8×10, 16×20, 24×30 (--ar 5:4)
4:5 → (same as above, but vertical) (--ar 4:5)
5:8 → Tall (--ar 5:8)
8:5 → Wide (same as above, but horizontal) (--ar 8:5)
2:1 → Taller (--ar 2:1)
1:2 → Wider (same as above, but horizontal) (--ar 1:2)
9:20 → Taller-er (--ar 9:20)
20:9 → Wider-er (same as above, but horizontal) (--ar 20:9)