Complete Beginners Guide to Midjourney V6!

Summary

この動画は、Midjourney Version 6の使い方を詳しく解説しています。Midjourney を使った AI 画像生成の基本から、アニメ風の描画やアップスケーリングなどの高度な機能までを網羅的にガイドしています。プロンプト指定の方法や、様々なパラメーターの使い方を学ぶことができ、リミックスモードによる柔軟なイメージ生成も体験できます。Midjourney Webサイトの機能についても紹介されており、Discordとの連携方法なども説明されています。この動画を見れば、Midjourney の豊富な機能を十分に活用できるようになるでしょう。

Takeaways

• 😀 Midjourney は Discord に基づく AI 画像生成ツールで、プロンプトを入力するだけで AI によって画像が生成される。
• 📈 Midjourney の基本的な操作は Discord サーバーに入り、Midjourney ボットとのプライベートチャンネルでプロンプトを入力して画像を生成することから始まる。
• 🎨 プロンプトには様々な詳細を含めることができ、メディアタイプ、スタイル、カラー、対象物などをプロンプトに追加するとより本物らしい画像が生成される。
• ⚙️ 各パラメータを調整することで、スタイル化、アスペクト比、カオスなど画像の表現を細かくコントロールできる。
• 🖼️ 複数のバリエーション画像を生成したり、画像の一部を拡大・縮小したり、「very region」機能で特定の領域を変更することで、より理想的な画像に仕上げることができる。
• 🔄 リミックス機能を使えば、元のプロンプトを変更しながら画像をさらに改善できる。
• 🏞️ アップスケールやカスタムズームなどの機能で画像の解像度を高めることができる。
• 🔍 参照画像を使うと、その画像のスタイル、テクスチャ、人物などを生成画像に反映させることが可能。
• 🎻 コンシステントスタイル機能を使えば、参照画像のスタイルを完全に引き継いでより本格的なイラストを生成できる。
• 🌐 Midjourney のウェブサイトでは過去の作品を管理・フィルタリング・検索でき、他のユーザーの作品も参照できる。

Q & A

Mid Journeyの機能や使い方を学ぶ前に何が必要ですか?

Discord アカウントを作成し、メールアドレスを確認する必要があります。その後、Mid Journey のウェブサイトにアクセスして Discord アカウントとリンクさせます。

Mid Journeyの基本的な使い方を教えてください。

Discord内の Mid Journey ボットに /imagine コマンドを送信すると、その後プロンプトを入力して画像を生成できます。生成した画像は Discord のダイレクトメッセージ内に表示されます。

Mid Journeyの主な機能はどのようなものがありますか?

解像度のアップスケール、アスペクト比の変更、画像へのペイントや編集、プロンプトの微調整、スタイルの調整、リファレンス画像の利用など、さまざまな機能があります。

プロンプトの記述方法について教えてください。

基本的には画像の説明を簡潔に記述します。色、スタイル、アングル、細部の指定などを含めると良い結果が得られます。パラメータを使うと更に細かくコントロールできます。

スタイル化の程度を変更する方法を教えてください。

プロンプトの後に --s (ハイフン2つとsの小文字) とスタイル化の値 (0~1000) を付加することで、Mid Journey がどの程度スタイルを注入するかを指定できます。

Mid Journeyで作成した画像をアスペクト比を変更する方法はありますか?

プロンプトの後に --ar (ハイフン2つと ar の小文字) とアスペクト比 (例: 16:9) を付加することで、アスペクト比を調整できます。

画像の一部分だけを編集する方法を教えてください。

/vary コマンドを使うと、画像の特定の領域を選択し、その領域に新しいプロンプトを適用できます。領域選択ツールを使って範囲を指定し、プロンプトを入力することで目的の編集ができます。

リファレンス画像を利用する方法を教えてください。

画像をアップロードしてコピーリンクを取得し、プロンプトの最後に --iw (ハイフン2つとiwの小文字) とリファレンスの重みつけの値 (0~2) を付加することで、リファレンス画像をもとに画像生成できます。

一貫したスタイルで生成する方法はありますか?

プロンプトの最後に --sre (ハイフン2つとsreの小文字) とリファレンス画像のURLを付加すると、その画像のスタイルを適用した画像を生成できます。

Mid Journeyのウェブサイトはどのような機能がありますか?

ウェブサイトではユーザー自身の生成画像の管理や、他のユーザーの作品を探索することができます。将来的にDiscordと同等の機能が追加される予定です。

Outlines

00:00

🔍 mid Journey バージョン 6 の紹介

この段落は、mid Journey バージョン 6 の基本的な機能と特徴について包括的に説明しています。ユーザーがアカウントと Discord サーバーを設定し、プライベートの mid Journey ボットにアクセスする方法を説明しています。基本的なプロンプトの使用方法と、mid Journey が画像を生成する仕組みについても触れています。

05:02

🖼️ プロンプト、アスペクト比、スタイリゼーションなどのパラメータ

この段落では、mid Journey で使えるさまざまなパラメータや機能を詳しく説明しています。プロンプト、アスペクト比、スタイリゼーション、カオス、テキストの使用方法など、画像生成に直接影響するパラメータがカバーされています。パーミューテーション機能の使い方や、オプションセットの保存方法についても言及されています。

10:06

🙎🏼‍♀️ アップスケール、バリエーション、リロール

この段落では、画像のリサンプリングやリロールなど mid Journey の重要な機能を取り上げています。アップスケーラーを使って画像の解像度を上げる方法、バリエーションボタンで画像のバリエーションを生成する方法、リロールしてまったく新しい画像を生成する方法が説明されています。

15:08

🖌️ 画像の編集: 以内ペインティング、パン、ズーム

この段落は、mid Journey が提供する画像編集機能について詳しく説明しています。以内ペインティングを使って特定の領域を編集する方法、パンとズーミングを使って画像を拡大・縮小する方法が解説されています。また、ブラックボーダーの解消方法やアスペクト比の変更など、よりよい画像を生成するための実用的なヒントも紹介されています。

20:09

🔁 リミックスモード: プロンプトの編集とペイントツールの強化

リミックスモードが mid Journey に追加されたことで、プロンプトの編集や画像の編集がより自由にできるようになりました。この段落では、リミックスモードを有効にした場合のプロンプトボックスの使い方や、パンやバリエーションボタンとの連携方法が解説されています。リミックスモードの恩恵を最大限に活かすための使い方が丁寧に説明されています。

25:10

🌄 参照画像の使用と一貫したスタイル

この段落では、参照画像をアップロードし、その画像の特徴を新しい画像へと反映する方法を説明しています。人物の顔を一貫させる手順や、参照画像のテクスチャをオブジェクトにマッピングする方法が取り上げられています。さらに、最近追加された一貫したスタイル機能を使って、参照画像のスタイルを他の生成画像に適用する方法も解説されています。

30:11

⚙️ その他の設定と mid Journey ウェブサイトの活用

最後の段落では、mid Journey の他の設定オプションや、mid Journey ウェブサイトの活用方法について説明しています。プロンプト履歴の検索、生成画像のフィルタリング、プロンプトのキーワード検索など、ウェブサイトの機能が詳しく解説されています。また、今後のウェブサイトの発展についても言及されています。

Keywords

💡Mid Journey

Mid Journeyは人工知能を使って画像を生成するプログラムのことです。このビデオは、Mid Journeyのバージョン6の使い方をステップバイステップで解説しており、Mid Journeyの中心的な役割を果たしています。プロンプトの使い方、パラメーターの設定、画像のカスタマイズ方法など、Mid Journeyの特徴的な機能がビデオの中で詳しく説明されています。

💡プロンプト

プロンプトとは、人工知能に画像の生成を指示する短い文章のことです。このビデオでは、プロンプトの書き方やパラメーターを使って微調整する方法が解説されています。例えば「グラデーションのパープルと桃のバックグラウンドに対して、幻想的な大規模で流線型のホログラフィックドラゴンの3Dモデル --ar 3:2」というプロンプトが使われています。的確なプロンプトを書くことが、美しい画像を生成する上で重要です。

💡パラメーター

パラメーターとは、プロンプトに付随して画像の生成を制御するための設定のことです。アスペクト比、スタイリゼーション、カオス値など、さまざまなパラメーターが紹介されています。例えば「--s700」のように、パラメーターをプロンプトに追加することで、生成される画像の性質を調整できます。パラメーターを適切に設定することで、理想的な画像を得られる可能性が高まります。

💡リミックスモード

リミックスモードでは、生成された画像をベースに、プロンプトを微調整しながら画像を更新していくことができます。パン、ズームイン、アウトペインティングといった機能を使って、基本となる画像の一部を変更し、新たにプロンプトを追加してリアルタイムに画像を修正することができます。このようにプロンプトと設定を細かく変更しながら、理想的な画像を作り上げていく手法がリミックスモードです。

💡アウトペインティング

アウトペインティングとは、既存の画像の外側に画像を拡張する機能のことです。中心に描かれた主題を変えずに、背景を追加したり画像の範囲を広げたりすることができます。矢印のボタンを押すと、その方向に画像が広がっていきます。例えば、初めは女性の上半身しか写っていない画像に対し、アウトペインティングを使うことで、全身像にしたり、背景に湖や木々を追加することができるでしょう。

💡ベリリージョン

ベリリージョンは、画像の特定の部分を選択し、細かな変更や修正をする機能です。ラッソツールで領域を選択し、その領域に対してのみ新しいプロンプトを適用します。画像の特定の部分だけを書き換えたり、足りないものを追加したりする際に役立ちます。例えば、イルカの頭にかぶせるはずだった王冠がモデルの頭につけられてしまった場合、ベリリージョンを使ってイルカの頭の部分だけを「王冠をつけた」プロンプトで上書きすることができます。

💡参照画像

参照画像とは、生成画像に何らかの影響を与えるために使われる既存の画像のことです。参照画像のテクスチャや色合い、スタイルなどを、生成された画像に反映することができます。例えば、ピンクと青の絵の具の参照画像を使った場合、生成された花瓶の質感にそのマーブル模様が適用されるでしょう。参照画像は、一貫性のある特徴を持たせたい場合なども有効な手段となります。参照画像は「-SRE」パラメーターを使って、その画像のスタイルを適用できます。

💡一貫性のあるスタイル

一貫性のあるスタイルとは、参照画像のスタイルを生成画像に適用する新しい機能のことです。水彩画風のスタイルの参照画像を使えば、生成される画像に同じようなスタイルが一貫して反映されるでしょう。風景画と動物、擬人化されたキャラクターなど、さまざまなジャンルの画像に、統一されたスタイルを適用できます。このようにスタイルの一貫性を保つことで、画像間の統一感を出すことができます。

💡ミッドジャーニーウェブサイト

Mid Journeyには、画像を管理したり、フィルターをかけたりするための公式ウェブサイトがあります。ここでは生成した画像にアクセスでき、様々な条件でソートやフィルタリングができます。例えば、アスペクト比別の画像を見るなどして、探しやすくなっています。作品をシェアしたり、参照画像をアップロードしたりといった機能の追加が予定されており、今後ますますユーザーフレンドリーになっていくでしょう。

💡パン

パンとは、画像の枠を広げて画像の一部を拡張表示する機能のことです。画像の端に矢印のボタンがあり、それをクリックすると画像が矢印の方向に広がります。例えば、半身だけを描いた画像の下半身を追加して全身像にしたり、人物の背後の風景を追加してスケールを広げたりできます。パンは複数回使用可能なので、好きなだけ画像を広げられます。リミックスモードと組み合わせると更に柔軟に使えます。

Highlights

Introduction to Mid Journey version 6, a comprehensive guide on AI image generation, covering basic photo generation, canvas expansion, inpainting, remixing, and new features.

Step-by-step guide on signing up for Mid Journey through Discord, including selecting a subscription plan.

Overview of generating AI images with Mid Journey bot on Discord, including basic prompting techniques.

Exploration of different image mediums and styles, such as watercolor, impressionist oil painting, digital art, and cartoons.

Introduction of the special model for anime style images, including how to access and use it for generating specific art styles.

Detailed explanation of the stylize setting in Mid Journey, demonstrating how it affects the creativity and style of generated images.

Introduction to generating text within images using Mid Journey, highlighting the effectiveness for common phrases and the challenges with longer sentences.

Comprehensive guide on using custom parameters to tailor image generation to specific needs, including aspect ratio and stylization adjustments.

Explanation of the use of negative prompts and the chaos parameter to control image variation and unexpected results.

Tutorial on using the permutations feature to easily iterate through multiple variations of prompts.

Insight into advanced features like image upscaling, variations, rerolling, and very region adjustments for detailed image customization.

Guide on using the pan and zoom features to extend and adjust the borders of an image, including tips for removing black borders.

Introduction to remix mode for further customizing prompts and images, enhancing the creative process.

Tutorial on using reference images to generate new images that resemble the original, including adjusting image weight.

Guide on the consistent style feature, showing how to map styles from reference images onto AI-generated images for uniformity.

Overview of the Mid Journey website features, including exploring generated images, filtering options, and future development plans.

Transcripts

if you're getting started with mid

Journey version 6 and you're looking for

a complete stepbystep guide on all the

amazing features and tools or if you

just need a refresher on AI image

generation you're watching The Right

video will cover everything from

generating basic photos expanding your

canvas the Imp painting and remixing

features to get the perfect images and

all the other quote new features that

have been added to version 6 so far like

using a consistent Style by the end of

this video you'll be comfortable with

all the tools available in mid Journey

this is going to be a long tutorial so

make sure to check below for time stamps

the first thing we have to do is sign up

for a mid Journey subscription to do so

we actually need a Discord account sign

up for an account and make sure you

verify your email address once you're

done with that log into Discord and head

over to Mid journey.com use a signin

button on the bottom right which should

automatically ask you to link your

Discord

account once you've logged in you'll see

the explore page where it'll show a

bunch of other images from other people

we'll come back to the website later for

now we need to sign up for subscription

go down to the account settings in the

bottom left which is the person icon

from there scroll to the manage

subscription option now I already have a

plan set up but you'll be able to pick

one here if you're just starting I use a

standard plan personally but the basic

plan is perfect for beginners and is

more than enough to get you

started once you've signed up open

Discord and you should be able to see

the mid Journey server inside Discord on

the left that looks like a ship icon if

you can't see it I'll post an invite

Link in the description of the video to

the Discord Channel server there's a lot

of info in the mid Journey server about

how to get started the latest

announcements prompt tips things like

that but right now we need to start

prompting

to access the private mid Journey bot

where we'll be generating our images

click on this show member list button in

the top right which looks like an icon

with two people in it then scroll to the

mid Journey bot on the list right click

on it and in the popup menu select

message and a private message with mid

Journey will open up for you we're all

set up now so let's start generating AI

images inside the direct message with a

mid Journey bot at the bottom will enter

a description of what we want the AI to

generate this is called prompting type

SL imagine which is the command to enter

a prompt and press space inside the

prompt box let's go with photo of a

woman in CYO mid Journey automatically

generates a grid of four different

images based on your prompt usually this

goes by pretty quickly unless than 30

seconds or so these images will be

randomly generated each time once it's

done we can zoom in on the image grid by

clicking on it if you're happy with them

you can open up the mid Journey website

and head over to the my images

folder from there just select the image

you like and use a download button to

save it to your

computer we can alternate between

different image mediums and styles I

prompted for a photo here but you can

try asking for a watercolor painting

impressionist oil painting digital art

or TR cartoons for your promp start

basic and slowly adding more and more

details we'll give her a motorcycle so

she doesn't have to walk now let's add

some more fine grin details like colors

make the motorcycle a black Harley bike

she'll need the clothing to match so

let's dress her with a brown leather

jacket and blue jeans I forgot to give

her a helmet since we're responsible

let's throw that in there too if you

want to take a deeper dive into the

specifics of prompting I'll link my mid

Journey version 6 prompt guide in the

description of this video and also at

the end for this particular image though

let's also try generating her an anime

art style drawn with Copic markers and

see what it looks like these images are

pretty good and do capture the essence

of anime but mid Journey actually has

released a special model specifically

designed for anime style images you can

access them by using the settings menu

so go down to the direct message type SL

settings inside and press enter the

settings menu will pop up inside the

models drop down we can see all the new

unnown models together currently I have

the newest version six selected which is

the one I suggest you use for the most

part but we can also try the nii model

version 6 if we specifically want anime

or cartoon inspired images with that

selected let's prompt for the woman in

Cairo again this time specifically

prompting for anime art styles a few

basic ones to try might be the Shonen

art style for anime targeted towards

teams or the cuter chibi Style just like

the normal mid Journey models nii images

does sometimes struggle with hands a bit

different themes and styles are

effectively generated though like a

detective robot or this pencil drawing

of a boy devil it's not just limited to

anime art style either the Nija models

also work great for Western comic art

and cartoon illustrations in general I

really enjoy Landscapes generated using

the nii

style let's switch back to the normal

version 6 model and walk through some of

these other settings the stylize setting

which goes from low to medium high and

very high controls how much mid Journey

injects its own style and creativity

into your images that low styli setting

gives a more direct image based on your

prompt and a higher style setting will

add a lot of extra style and flare to

the image we enter a prompt asking for a

robot in the style of the Victorian

English era it should be serving lunch

and have pastel colors on it and the

exterior of the robot has a texture of

polished concrete let's compare what

happens when we increase the stylage

parameter from low all the way up to

very high at a low stylage value the

image is fairly straightforward it's not

as decorative and even looks kind of

bland but it does Follow The Prompt

correctly as the stylus value increases

we can see mid Journey's influence more

it starts adding in accessories and

decorations the background goes from a

small living room to a grand hallway in

a

palace the raw mode option offers the

most direct interpretation of mid

Journey's prompt without adding an extra

flare it's similar to using a low

stylized value in that it won't add mid

Journey style to the image in mid J your

version six The Prompt understanding is

already really good though so the raw

mode is is not as important as it used

to be in previous models a really nice

new quality of life update in midor

version 6 is the ability to generate

text it's easy to add text into your

prompts just put the words you want

printed in quotation marks like I did

here I asked for a floral design with

red roses and green Vines against the

white background then the words happy

birthday in quotation marks written in

cursive in Gold you'll notice that not

all of the images generated the text

correctly but for common phrases like

happy birthday it usually does a pretty

good job with it let's try designing a

product we'll use a tall aluminum can of

juice with the white text Fresh

vertically printed I'll go with sansera

font this time the can should also have

a blue label I like how clean the text

looks in these images and it prints the

word fresh very clearly you can also try

putting the text in different mediums

like carved on a gravestone or painted

in with watercolor you if you try longer

sentences or more complicated phrases it

doesn't work as well though here I

prompted for the tong twister she sells

seashells by the seashore it gets some

part of the sentence correct but it's

not quite there

yet to generate the best possible images

you'll need to use custom parameters

parameters tailor mid your needs

specifically to fit your needs remember

these styliz buttons you don't have to

up the settings menu every time you want

to access these to take a shortcut just

type d-s after your prompts followed by

a number between 0 to a th000 the two

dashes tell mid Journey you're using a

parameter and S is a parameter code for

styliz after that is a parameter value

in this case that's a number between 0

to a th000 where zero is no stylization

and 1,000 is a maximum amount of

stylization let's compare images

generated with different amounts of

stylization the lower values have more

plain images without too many special

effects as we increase the stylization

value though the images get more

artistic with lights and

sparkles the most common parameter

you'll use is the aspect ratio which is

the proportional relationship between

the width and height of the image the

default aspect ratio is one1 but we can

change that to other popular ones like

the vertical 9 to6 or horizontal 16 to 9

here's our initial prompt of a dolphin

made out of transparent glass floating

in the sky to change the aspect ratio to

a wide full screen image type-- AR

followed by 16 to9 we can also combine

multiple different parameters I want to

highly stylized image so I'll add-- s700

after the aspect ratio now when we run

the AI we'll get a watch 16 to9 image

but it'll also be pretty

stylized

[Music]

there's a ton of other parameters that

we can use the no parameter is a

negative prompt and tells me Journey

what you do not want to be in the image

for example let's prom for a video of

California poppies in this field

Outdoors these are pretty photos but

what if I don't want the sky to be in

the background and I only want to see

the flowers and grass in that case I

would Das Das no with the word Sky

afterwards to tell mid journey to not

generate the sky in my

images and we'll get photos with only

flowers and grass in them the chaos

parameter tells mid Journey how much

variation you want in your image

Generations it's a value between 0 to

100 Zer is a default value and 100

generates the most unusual and

unexpected results let's try generating

this girl in a field of orange flowers

with teal tattoos and necklaces starting

with a chaos value of zero the images

all look pretty similar to each other

and follow the prompt I use pretty

closely bump the chaos up to 50 and

there's a lot more diversity and

variation in the images that we generate

from the different camera angles to

phasal markings and their ethnicities

but the images still follow our prom

pretty well using the maximum chaos of

100 though and we get a huge amount of

variation the images no longer follow

out prompt accurately and the resulting

images are definitely unexpected

if you want a lot of prompt control I'd

stick with a default chaos value of zero

but if you want to explore what's in mid

Journey's visual Library you can try

using the higher value the weird

parameter does something similar to

chaos setting it to zero will generate

standard images but increasing the value

all the way up to 100 and you'll get

some quirky and unique

results

when you're prompting you might want to

make Minor Adjustments to specific words

or items in your prompt it can be

annoying to type the prompt multiple

times just to change a few words for

example let's say we have this prompt a

cute anime style inspired watercolor

painting of a baby blue dragon happy and

smiling what if we wanted to test out

different colors like red green and

yellow though the permutations feature

lets you easily iterate through multiple

different prompts just put the colors

inside curly brackets like I've done

here and then when you go to enter your

prompt in mid journey and check yes so

mid Journey ask you to confirm

processing for multiple different jobs

the four different colors we ask for

will be generated

[Music]

separately permutations can also be

applied to parameters like the aspect

ratio or anything else to get more

advanced we can apply permutations

multiple times in our prompt for example

this prompt of impressionist oil

painting of a red or blue cat or dog

which will generate four separate

paintings mixing and matching between

red blue and cat and

[Music]

dog if there's a set of parameters you

really like you can save them using the

prefer option set feature I've got a set

of parameters here-- n 6-- AR 9 to6 D-

styliz 500 nii tells me journey I want

to use a specific anime nii model I

don't want to have to enter these

parameters in the mid Journey every time

for every prompt so instead I'll type SL

prefer option

set press space and then inside option

enter the name for my parameter set

which will be vertical nii

select the plus one more button and

select value from the popup menu inside

I'll put in my parameters -- n6 D AR 96-

S500 now when I enter this prompt mid

Journey saves those settings to the

keyword vertical nii let's try

generating a new prompt a yellow fox

bear that's made out of yellow dandelion

there are many flower petals on the

fox's skin then add Das Das vertical

need instead of the set of

parameters notice that after we enter

this prompt mid Journey automatically

adds our saved parameters into the

prompt line and it will generate a

vertical image of a very stylized Fox

spirit in anime

style I've covered the main parameters

you want to know if you want to check

out the full parameters list I'll put a

link in the description so far we've

learned a lot about prompting and

parameters which are the fundamental a m

Journey let's get into some of the other

cool features we can use underneath the

image grid is a set of buttons labeled u

and v these correspond to each of the

four images in the grid if we click on

the U button it opens up the

corresponding image and gives you a

variety of different options for

customization starting with the upscaler

which doubles the resolution of an image

currently there's two upscale options

called subtle and creative subtle

increases the resolution while

preserving the general structure in

detail details and creative will upscale

the image but adding new details here's

an image of two women jogging let's

compare using the subtle upscale and

creative upscale for the subtle upscaled

image on the right the resolution is

increased and you see some additional

sharpness to the image but the overall

structure stays the same by the way when

generating faces with full body shot

mid-journey struggles a bit if the face

is just a very small part of the entire

image now if we switch over to the CR

creative upscale image the facial

structure has changed the person's

expression is different now but it's

quite a bit sharper and looks slightly

more realistic than before also to save

this image we can just right click on it

and scroll down to save

image next up are these VAR buttons for

if you like the general structure of the

image but want to see what it looks like

if you add a little bit more Randomness

and variation to it very subtle produces

a new image grid consisting of

variations of the original image the

variations are automatically added in by

mid journey and are usually fairly minor

the colors and structure of the original

image get preserved and only small

details are changed like the shape of

the eyes of our dog

painting very strong will produce more

diverse variations of our original image

while the colors in subject matter are

largely preserved the direction the head

is pointing and the way the ears flop

around its head look different I do use

these variations pretty frequently it's

a good idea to use them if the image is

already pretty close to what you want

and instead of generating a new image

completely from scratch you can just do

these variations to see what else you

can get you can also access these

variation options from the initial image

grid with these numbered V buttons just

click on one of them and mid Journey

directly creates variations of the

corresponding image for you you can

control whether they're small or strong

variations so open up the settings menu

and then check whether you want a high

variation or low variation mode also

there's the reroll button which looks

like the two arrows chasing each other

it runs mid Journey again on the same

prompt to generate a new set of images

if your initial image grid doesn't match

exactly what you want but is pretty

close I'd recommend rerolling a couple

of times before changing up the prompt

since mid Journey does have some

Randomness in its

generation the very region button is a

special feature that lets you do imaging

and painting to make adjustments to very

very specific parts of the image let's

try demo take this image of a floating

dolphin I'll use a square region

selection tool and select a small part

of the image to adjust I want adding a

priest watching the dolphin in the area

where I've selected the square so inside

the promp box I'll add in an ancient

Egyptian priest looking upwards shot

from behind and then if we enter this

prompt we'll get an updated version of

the image with our Egyptian priest now

this tool is a especially powerful for

when mid Journey misinterprets are

prompt in this image I tried to generate

a young woman swimming next to a dolphin

specifically I asked for a gold crown to

be on the Dolphin's head but as you can

see mid Journey gets confused and

generates the crown on the woman's head

instead I really like the layout of this

image though and don't want to change it

I just need to move the crown from the

girl onto the dolphin let's fix that

using the very region feature we'll open

up the very region tool to and this time

I'll use the lasso selection tool this

gives you a little more flexibility in

the region that you're going to be

selecting I'll drag it around the crown

on the girl's head and then inside the

prompt box I'll change it to a young

Indian woman to remove the crown by the

way if you mess up with the region you

selected and want to restart just right

click on the selected region to start

over the images have finished generating

and we can see that the Gold Crown has

been removed from the girl open the the

new image with the very region editor

I'm going to drag the lasso selection

tool around the Dolphin's head where I

want a gold crown to be placed then

inside the prompt I'll putting a gold

crown all right so the image finished

generating and it didn't work as I

expected depending on the context of the

image mid Journey can struggle to

generate specific objects and items in

them we'll have to experiment with the

prom some more this time I'll put in the

dolphin has a gold crown on its head to

get give Mery more clear directions with

what I

want looking at the results we were able

to get a few images that generated our

prompt correctly sometimes it'll take

you a few tries and changes to the

prompt to get exactly what you want from

mid Journey so don't get discouraged if

it doesn't work the first time work on

your prompts a bit to make the

instructions as clear and obvious for

Mid Journey as possible so we've seen

how to adjust interior of our images but

what about the outside region we can add

more to the borders by using the image

out painting feat feature the pan

features which are these arrows

underneath the image let you extend the

image in the direction the arrow is

pointing towards so if we use a arrow

pointing to the right the imigy extends

the image towards the right side and we

can see more of the woman's

torso and if we use a downwards pointing

Arrow imary extends the image downwards

and we'll get some interesting results

it looks like she's wearing a dress in

the water you can continue to use a pan

feature to extend your image in

different directions as many times as

You' like it's a really cool tool that

lets you add extra size to your images

whenever you need next up we can use the

zoom buttons to zoom out in all

different directions the two time zoom

button and 1.5 zoom button pull the

camera back and expands the outside

borders of the image and fills them in

using AI one trending topic was the

infinite Zoom videos that were made

using mid Journey where people added

together images generated from the zoom

feature to get this infinitely zooming

effect I'll put a link in the

description for a tutorial on how to

make these types of

videos here's a pro tip sometimes Mia

produces images like this one where

there's black borders around the image

it can get pretty annoying when you get

an otherwise Perfect Image but it has

these black bars outside one way to get

rid of them is to use a custom Zoom

feature if we open it up you'll see the

prompt with an option to change the zoom

Factor just change that to something

small like

1.1 then when you generate the zoomed

out images most of the time the black

bars will get removed and you end up

with a nice clean image also the custom

Zoom feature is a really convenient way

to change the aspect ratio of the image

while preserving all of the

information if we look inside the prompt

box we can change the zoom Factor but

also the aspect ratio let's try making

this a wide 3 to2 image and now change

the zoom factor to

1.1 this is my favorite way to change

the aspect ratio of an image I already

like without using the pan features the

make Square feature automatically fills

a border of any rectangular image to

turn it into a square I don't use this

feature that much personally but it's

there if you need it so we've covered a

lot of features for variations in paint

pain in and out painting however we can

take our mid Journey skills to the next

level by using remix mode go to the

settings menu by typing slash settings

to turn on remix mode just select the

remix mode button remember the reroll

button which runs our prompt again now

that we have remix mode turned on when

we go to regenerate our images a new

prompt box will pop up this allows us to

make any adjustments to our prompt

without having to type it in again but

as your prompts get longer and you add

more and more parameters this helps a

lot for this case let's try changing the

camera angle from a low angle shot to a

high angle shot from

above and now we can see our dolphin

from a different perspective with remix

mode turn on we can do much more with

mid jour's features when you go to pan

in your image a custom prompt box will

open up and you can change up the prompt

like adding in a Blue

Lake then when the pan feature goes to

extend the image it'll actually take

into account the the new prompt that

you've

entered this also works for variations

buttons I really like the image of a

float IND dolphin but I'd like to adding

some green trees and water below to make

it feel more lively if we use very

subtle inside the new remix prom box

that pops up we can try adding in some

Greenery with vegetation and water

flowing this should preserve the overall

structure of the image with the dolphin

and City below but sprinkles in some new

details and there we go we've managed to

preserve the structure of the original

image while adding in trees and water

the very strong button also allows for

this but it does make more dramatic

changes to the original image if we try

to do the exact same thing and adding

palm trees and water to our new prompt

our Dolphins will be a bit more diverse

this

time also remember that the custom zoom

button also allows you to customize it

prompt so in this case let's try try

adding in some aquatic plants and water

lies around this girl and the

dolphin the custom zoom button can do

this even without remix mode turned on

so make sure to take advantage of that

for me I always have remix mode turned

on because of the extra control it gives

me over the images that I generate also

if you're remixing and panning your

images it can be kind of hard to keep

track of where the original images in

the entire Discord feed so instead of

scrolling around just click on this

headline above the newly generated

images which will bring you back to the

original

[Music]

image to upload reference images you

just need to drag and drop them in a mid

Journey here I've got a photo of a woman

I'll drag her into mid journey and press

enter to upload it now I can right click

on it to bring up a menu and inside go

to copy image link this will allow us to

use image as a reference to generate new

images then when I go to enter prompt

like photo of a woman on a camping trip

in Alaska with winter clothing I can

paste in the reference image Link at the

end then when I generate the image it

should reference the person in my

original image and the photos that are

generated will look very similar to the

original person you can change how

important the reference image is by

using the image weight parameter just

add-- IW at the end of your prompt

followed by a number between 0 to two

where higher number means you want the

image to be weighted more heavily so

back to a prompt of a woman camping in

Alaska this time if we add-- IW at the

end of The Prompt followed by the number

two if we look at the new images

generated they look much more similar to

the original image than before in terms

of the clothing that she's wearing which

all looks like a white collared shirt

and also the direction that she's

facing consistent characters are really

tricky to do they they're rarely ever

perfect if you want my step-by-step

video guide for how to get consistent

characters with different facial

expressions I'll also add it in the

description there's a lot of different

ways to use reference images here I've

uploaded an image of pink and blue paint

mixed together let's see if we can apply

this texture onto an object like a vase

so we'll prompt for a porcelain vase

with the texture of pink and blue paint

against the plain white background then

just copy the image link into the prompt

and it should apply the pink and blue

texture onto the base and there we go it

looks like the texture from the

reference image is mapped down to the

vase recently mid Journey released a

consistent style feature which lets you

take a style from a reference image and

map it onto an AI generated

image I've uploaded a couple of images I

really liked here so let's take this

ground level shot of an asphal path in

Arizona

and see if we can generate a similar

style of image in a different location

I'll right click and copy the image link

then to use this image style onto

another image I just need to enter a

prompt then add D- SRE at end of it

followed by the image URL which tells me

journey to use the style of my reference

image I'll ask for a ground LEL shot of

a dirt path on a mountain trail with

some Ferns and small animals on it

here's what it would look like in the

Discord prompt box at the end of The

Prompt adding D- SRE with the copied

image

URL I tried applying this to a couple of

different images and it works pretty

well here's another example of a

reference image I had for watercolor

style children's book drawing I applied

the style to generate some images of

seahorses you can definitely feel the

influence from the original image from

the shape of the seahorse to the

saturation of the colors in the image

you can turn up how much the weigh the

reference Style by using the D- s SW

parameter this holds a value between 0

to 1,000 with 100 being the default if

we turn the style weight all the way up

to 1,000 when generating our seahorses

you'll notice that the tail of the

seahorse follows the shape of the or

image a bit

[Music]

closer there's a couple more options in

the settings menu to control the

generation speed turbo will generate

images the fastest but takes up the most

GPU time which is a limited resource

depending on your subscription for me I

usually just leave it on fast mode which

is a default it's plenty fast enough for

me then you have relax mode which is

available to standard plan users and

higher it's a lot slower but it doesn't

cost you any GPU time so you

theoretically can generate unlimited

images in the settings menu there's also

a sticky style button which is meant to

be used with a style tuner feature I'm

not going to cover that feature since

the consistent style feature has been

released it does something very similar

and it also costs way less GPU time and

is much easier to use in my opinion

let's take a closer look at the mid

Journey website and what you can do with

it the features are still being

developed right now but it's really

useful for sorting through your images

on the explore page you can check out

what other people are generating and

there's also a liked tab where you can

keep track of images you like the most

useful part though is being able to keep

track of your work and filter your

images the my images tab holds every

image that you've ever generated ated on

the right you can actually filter

through your images so for example I

could ask for images generated using the

only version six or images that only

have wide landscape aspect ratios inside

the search prompts tab you can look up

specific keywords in prompts so if I

type in horse it'll bring up every image

I've generated with a keyword horse then

if I select one of them I can choose

from a list of possible options like

copying the image link or downloading it

or opening it up in Discord there's

future plans right now to have this

website do everything that Discord can

do but right now that hasn't been

released to all the users yet I'll make

an updated guide when that's available

for everybody that's it that's

everything you need to know to get

started with mouri go and check out this

Advanced prompt guide I made that has a

lot of different prompt tips like colors

lighting Style and a lot more and make

sure to subscribe if you want to see

more from

[Music]

me