This Midjourney Update is WILD + Pika's New AI Video Feature

Summary

スクリプトの魅力的な要約で、短く正確な概要を提供し、ユーザーの興味を引き付けます。

Takeaways

• 🎬 AIが映画制作に大きな影響を与えている—特にMid Journeyの新機能と11 Labsの音声エフェクトツールが注目されています。
• 🌟 Mid Journeyがキャラクターの整合性機能をリリース—同じキャラクターを簡単に再作成できるようになりました。
• 👔 キャラクターの服を簡単に変更できるように—`CW`コードを使ってキャラクターの服装を調整できます。
• 🎨 スタイルを統一するための機能も追加—特定の色味やスタイルを維持することができます。
• 🔊 11 Labsは新しい音声エフェクトツールを発表—音声を自動的に動画に追加できるようになりました。
• 🎥 Pābの音声エフェクト機能と比べ—11 Labsのエフェクトの方が少し優れていると感じました。
• 📜 研究論文「Style Gaussian」が公開—アップロードされた参照画像のスタイルをGaussian Splatに投影できる技術です。
• 🎧 Open AIのCTOがSoraの発売についてコメント—今年後半に発売される可能性が高く、音声も組み込まれる予定です。
• 🤖 Elon MuskがGrockをオープンソースに発表—XのAIモデルのポジショニングと訴訟の進行が注目されています。
• 📹 AIが映画制作の未来を変える可能性—ライブアクションとAI生成のビジュアルが混ざる新しいメディア「Live Gen」が登場しました。
• 🏆 今週のAI映画ニュースで紹介された作品—Zahir Khanの「The Mermaid」、Uncanny Harryの「Uncanny Harriet」、Marcoの「Escape」が特に好評です。

Q & A

新しいキャラクター一貫性機能のMid Journeyはどのように使われるのか？

Mid Journeyの新しいキャラクター一貫性機能を使うと、同じキャラクターを繰り返し作成することができます。これはストーリーテリングや映画、広告プロジェクトで非常に役立ちます。デスクトップバージョンを使って使い方を示すことができます。

Mid Journeyでキャラクターを一貫させる際に注意すべき点は何ですか？

Mid Journey内で作成された画像を使用する場合、より良い結果が得られます。直接Mid Journeyで作成されたキャラクターを使用することをお勧めします。

キャラクターの衣装を変えるために使用されるコードは何ですか？

キャラクターの衣装を変えるために使用されるコードは「--CW」です。これは「Character Weight」の略で、0から100のスライダーで、0はキャラクターが衣装や髪型を変えることを可能にし、100はデフォルトで同じ衣装を繰り返し使います。

Mid Journeyでconsistent Stylesをブレンドすることの重要性は何ですか？

consistent Stylesをブレンドすることは、一貫性のある映画的なルックを維持することが重要であるため、非常に重要です。これにより、シーン間のキャラクターが一致し、全体的に美しい視覚効果が得られます。

Discord版のMid Journeyでキャラクター一貫性をどうやって得るのですか？

Discord版のMid Journeyでは、キャラクターの画像をチャットチャンネルに投稿し、右クリックしてリンクをコピーし、`/imag`コマンドを使用してキャラクター参照を設定します。また、`--CF`コマンドを使用してキャラクター参照を、`-D CW`コマンドを使用してキャラクターの重量を指定します。

11 Labsのサウンドエフェクトツールはどのように機能するのですか？

11 Labsのサウンドエフェクトツールは、プロンプトに沿ってサウンドエフェクトを自動的に追加する機能があります。画像やビデオにサウンドエフェクトを追加することができます。また、過去に生成したビデオにもサウンドエフェクトを追加できます。

Open AIのSoraのリリースタイミングは何ですか？

Soraは遅くとも今年後半にリリースされる予定です。具体的には、数ヶ月以内にリリースされる可能性があると言われています。また、Soraはオーディオを組み込む予定であり、P labsのようなサウンドエフェクト機能に似たものになります。

最近公開された研究論文「Style Gaussian」について教えてください。

「Style Gaussian」は、アップロードされた参照画像からスタイルを投影し、ガウシアンスプラットに適用することができる研究です。これは、Runway Gen1のように、拡張現実でスタイルを投影できる技術の早期版であり、将来のインタラクティブエンターテインメントに大きな影響を与える可能性があります。

最近リリースされたポッドキャストエピソードの内容は何ですか？

最新のポッドキャストエピソードでは、Renard Jenkinsがゲストとして登場し、AIが基本的なレベルでどのように動作するか、そしてAI生成画像においてより良い表現をどのように共创できるかについて洞察を提供しました。

Dave Clarkによって作られた短編映画「Another」の特長は何ですか？

Dave Clarkの短編映画「Another」は、実写映像とAI生成のビジュアルを混ぜ合わせた作品で、AIと実写の境界がはっきりしないほどの素晴らしい仕事をしています。この新興のメディア「Live Gen」は、将来の映画の方向性を示すプロトタイプのようなものです。

Elon MuskがオープンソースとしてリリースしたGrockとは何ですか？

Elon MuskがオープンソースとしてリリースしたGrockは、Open AIのAIモデルと関連する訴訟の進行をどのように影響するか見守るに値する興味深いものです。

Outlines

00:00

🎬 AI映画制作の新機能

本段落では、AI映画制作の分野で注目すべき新しい機能が紹介されています。Mid Journeyがキャラクター一貫性機能をリリースし、11 Labsは音声エフェクトツールを発表しました。CTO of Open AIもSoraの市場投入について更新情報を提供しています。特に、Mid Journeyの新しい機能はストーリーテリングや映画、広告プロジェクトにおいて、同じキャラクターを簡単に再作成できるようになりました。この機能の使い方の詳細が説明されています。

05:00

🎨 一貫性のあるキャラクターとスタイルの作り方

この段落では、Mid Journeyを使って一貫性のあるキャラクターとスタイルをどのように作り出すかが説明されています。キャラクターの体重（character weight）というスライディングスケールを使用して、キャラクターの衣装を変更することができ、0から100までの範囲で調整できます。また、一貫性のあるスタイルを作り出すために、特定の色グレードやスタイルを検索する方法も紹介されています。

10:01

🎵 音声エフェクトの追加方法

本段落では、11 LabsとP laabsが新たにリリースした音声エフェクトの追加方法について説明されています。音声エフェクトをAI生成の映像に追加する方法や、既存のビデオにエフェクトを追加する方法が紹介されています。また、11 Labsの音声エフェクトツールのベータ版も触れられており、その使い方や効果の例が提供されています。

15:02

📰 AI技術の最新動向と作品紹介

最後の段落では、Open AIのCTOがSoraのリリースタイミングについて語ったインタビューや、新しい研究ペーパー「Style Gaussian」の内容が紹介されています。また、AI技術の最新動向として、Elon MuskがGrockをオープンソースにすることを発表したことや、AIとロボット技術の統合デモンストレーションが行われたことも触れられています。さらに、AI技術を使用した映画作品の紹介が行われており、その作品の特徴や使用した技術についても詳細が提供されています。

Keywords

💡AI film making

AI映画制作は、人工知能技術を用いて映画を作成するプロセスです。このビデオでは、AI技術が物語tellや視覚効果を作り出す際にどのように活用されているかが示されています。例えば、AIによって生成されたキャラクターやシーンが、映画や広告プロジェクトにおいて一貫性を維持する方法が紹介されています。

💡Mid Journey

Mid Journeyは、AI映画制作ツールのことで、ユーザーが一連のプロンプトや画像を入力して、AIが自動的に映画的な画像やシーンを生成する機能があります。このツールは、キャラクターデザインや衣装の一貫性を維持し、映画や広告の制作において効果的です。

💡Character consistency

キャラクターの一貫性とは、映画やストーリーテリングにおいて、キャラクターの見た目や服装がシーンをまたいで同じであることを指します。これにより、視聴者はキャラクターを追いかけ、ストーリーに引き込まれることができます。AI映画制作ツールでは、この一貫性を維持するための特別な機能が提供されています。

💡Discord version

Discord versionは、Discordというオンラインコミュニケーションプラットフォーム上で利用できるAI映画制作ツールのバージョンを指します。このバージョンを利用することで、ユーザーはテキストベースのコマンドを入力して、AIが画像や動画を生成するように指示できます。

💡Sound effects tool

サウンドエフェクトツールは、映画や動画に音効果を追加するアプリケーションです。AI技術を利用することで、特定のシーンや状況に合わせて自然な音を生成することができ、視聴者の没入感を高めることができます。

💡Paa

Paaは、AI技術を使用してサウンドエフェクトを自動的に追加するツールです。これにより、ユーザーは動画に適切な音を追加することができ、例えば犬の吠え声や火事の音などをリアルに再現できます。

💡Sora

Soraは、Open AIが開発しているAIツールの名前で、公開される予定です。このツールは、音声を組み込むことができるとされており、AI技術を用いた新しいタイプのインタラクションを可能にするかぎります。

💡Character weight

キャラクターウェイトは、AI映画制作ツールで使用されるパラメーターで、0から100までのスケール上で設定されます。0はキャラクターの服装や髪型が大きく変わる可能性を意味し、100はデフォルトで同じ服を繰り返し着用的ことを示します。この機能を使うことで、一定の範囲内でキャラクターの見た目を柔軟に変化させることができます。

💡Style weight

スタイルウェイトは、AI映画制作ツールで使用されるパラメーターで、0から1000までのスケール上で設定されます。これは、生成される画像やシーンに適用されるスタイルの強さを制御するために使用されます。デフォルトでは100と設定されており、これにより基準となるスタイルが維持されます。

💡Cinematic still

シネマティックスティルとは、映画的な静止画を指します。これは、映画やテレビ番組の静止した画像から作成される宣传資料や、映画の特定のシーンを表現するために使用されるものです。AI映画制作ツールを使用することで、ユーザーはそのようなシネマティックな画像を簡単に生成することができます。

💡Live gen

Live genは、実時間のライブアクションとAI生成されたビジュアルを混在させて新しいメディアフォーマットを創出することを指します。この技術は、実世界の映像とAIが生成した視覚効果を組み合わせて、一種のエクステンシブなリアリティ体験を提供する可能性があります。

Highlights

Mid Journey releases a new character consistency feature.

The new feature allows for the creation of the same character with the same clothes repeatedly.

The feature is particularly useful for storytelling and film and advertising projects.

Characters created within Mid Journey yield better results with the feature than uploaded images.

The demonstration shows creating consistent characters in various scenes with ease.

The character weight (CW) slider allows for variability in character's clothing and appearance.

Character weight can be adjusted from 0 to 100 for different levels of consistency.

Blending consistent characters with consistent styles is crucial for a uniform cinematic look.

Styles can be locked in with a cinematic still and character reference for a cohesive visual experience.

The Alpha version of Mid Journey is now accessible to users who have generated over a thousand images.

Discord can be used to achieve similar functionality with specific commands.

11 Labs introduced a sound effects tool that can be added automatically to footage.

Sound effects can be added to both text prompts and existing images or videos.

11 Labs' sound effects tool is in limited release but offers promising capabilities.

A game is presented where viewers must identify the source of monster roaring sound effects.

The research paper 'Style Gaussian' introduces a method for projecting styles into augmented reality.

A new podcast episode features Renard Jenkins discussing AI representation.

A short film 'Another' by Dave Clark blends live-action with AI-generated visuals, introducing the term 'live gen'.

The CTO of Open AI provides an update on Sora, hinting at a release within the next few months.

Chat GPT 4.5 Turbo is hinted to be released in June 2021, reshuffling the Open AI board.

Elon Musk announces that Grock will be open source, positioning it in relation to Open AI.

A demo shows a robot integrated with Open AI, capable of interpreting its surroundings.

AI films of the week showcase 'The Mermaid', 'Uncanny Harriet', and 'Escape', highlighting the best in AI filmmaking.

Transcripts

it's been a really interesting week in

the world of AI film making mid Journey

finally released the feature that we've

all been waiting for paa and 11 Labs

came out with a sound effects tool but

which one is better and the CTO of open

AI gave us an update on when Sora is

going to be hitting the market this is

your AI film news of the week I want to

kick things off with the big lead and

that is of course the new character

consistency feature inside of mid

Journey character consistency see was

really tricky up to this point in time

but with this new feature you can easily

create the same character with the same

clothes again and again this is

obviously huge for storytelling and

working on film and advertising projects

I want to show you how easy it is to

create consistent characters inside of

mid Journey so I'll show you how to use

it on the desktop version then we'll hop

over to the Discord version now I will

say that this feature works better with

images that were created inside of mid

Journey so if you have to upload an

image you will not get results that are

as good as if you use a character that

was created directly in mid Journey so

let's go ahead and start things out here

I'll go to search props and we'll just

say a man so something really generic

and uh let's find a man that we want to

use here and we will say how about this

guy wearing this ridiculous outfit so

this would typically be very challenging

to prompt for again and again but with

the new feature it actually is pretty

darn easy so what we're going to do is

hit the image button here so use image

and now we will use the image in our

prompt so there's some really small

icons here they're hard to make out but

basically right now it's being used as

an image reference but if you go all the

way to the left there's this person icon

we'll click it and basically now it is

being used as a character reference so

we can say a cinematic of a man in New

York City and go ahead and click enter

and just like that we have some shots of

this man and it looks photorealistic

he's in New York City and he's wearing

that crazy outfit so this would have

been basically impossible up to this

point in time we have a shot of him from

behind from the side and oh my God I

have no idea what's happening here I

think he has been in a horrible accident

cuz his head is looking the wrong way so

there's some really interesting things

that you can do especially when you

begin to use this tool in conjunction

with some of the other features that are

available inside of mid Journey so I

want to show you how you can get

consistent characters with different

clothing options as well it's really

interesting so if you go to search

prompts we'll just say wide shot woman

and we'll just find a wide shot of a

woman where we can see all of her

clothing and just everything that she's

wearing maybe this woman right here you

see she has this brown jacket and the

blue undershirt so if we use the image

and we go over here to the character

reference button we can just say a woman

stands in we'll say downtown Los

Angeles and go ahead and click enter you

can see that yes of course we have this

woman in downtown Los Angeles and she's

basically wearing the same thing in each

shot now what if you actually wanted

this woman to be wearing something

different maybe you're doing some

exploration and wanting to do basically

a virtual photo shoot you can actually

do that very easily using mid Journey

all you have to do is click this use

button so now we have our inputed

reference image again right here and

what we will do is type in a code

called-- CW that stands for character

weight and character weight is on a

sliding scale of 0 to 100 with zero

being the character can basically change

their clothing their hairstyle and there

can be much more variability in what the

character is wearing versus a character

weight of 100 which is the default the

character will generally be wearing the

exact same clothes again and again so

I'm going to change this to a character

weight of zero and we'll say a woman in

a sci-fi film there we go and let's see

what we get okay let's look at our

results here we have the woman and in

this one she's wearing like futuristic

headphones and a yellow jacket in the

next one she's wearing this kind of worn

out coat here has an earring the next

one she's wearing I think a leather

jacket and and this has a completely

different coat as well so you can use

the character weight to change the

costume of your character to do a little

bit more exploration and when you find

the costume that you want your character

to wear again and again you can change

the character weight to 100 and be able

to prompt it to get the same character

in your scenes between your shots you

also have the ability to blend

consistent characters with consistent

Styles which is really important

whenever you are trying to have a

consistent cinematic look so here's how

you do it I'm just going to go to our

prompt here we'll just say a woman and

we'll just find someone that we want to

use here let's say that we want to use

this woman here we'll go ahead and click

the image button and set her to

character reference so we also want to

figure out a style so to find a style

we'll just go to our search bar here

type in a cinematic

still and let's find kind of a color

grade or Style that we want to use for

our film I kind of like this kind of

hazy sepia tone over here so I'm going

to go ahead and click on this image and

we will go to style so now we have a

consistent character because we hit the

icon here and we have a consistent style

so now we can prompt in a cinematic

still of a woman in New York City and

the cool thing is your codes actually

work in here as well so the style weight

by default is 100 if you wanted to

change that you could just do d-w and

then up it all the way to a th000 if you

wanted to inherit maximum style from

this reference photo I like the default

I think that's totally fine and you also

can do d-r and we can say R5 to repeat

this prompt five times so all of those

codes still work on the website which is

absolutely amazing so go ahead and click

enter so here we go we can take a look

at our photos here and you can see that

they all have generally the same color

grade the style is very similar the

character is consistent and it looks

really good obviously you can use tools

like a cool or Insight base swap to

really help with stabilizing the faces

even further but for a quick character

base this new character consistency tool

inside of mid journey is absolutely

clutch and I should note that the Alpha

version of mid journey is now open to

everyone who has generated over a

thousand images in the platform if you

can't access the alpha just yet don't

worry you still have all of the same

functionalities you just have to use

Discord and some of the codes inside of

Discord in order to unlock all of this

functionality so I'm not going to leave

you hanging don't worry I'm going to

show you how to get consistent

characters inside of Discord so I'm

inside of Discord I have the Discord bot

invited to a private server and I

basically have this photo of this like

hipster guy and I'm going to drop him

into our chat channel here and basically

all you have to do is right click and we

will go to copy link so now that we have

the image link to this character we're

just going to do forward slash imag so

we'll say a cinematic still of a man in

a western film and now we need to add in

our character reference code to do that

you just do--

CF put a space and then paste your link

and for character weight we can do the

exact same thing D- CW space and then

we'll say 100 to have his uh outfit come

through and actually that'd be really

impressive because western films do not

typically feature people wearing this

type of clothing so uh I'll be really

curious to see how this turns out okay

let's take a look at our result here so

we have this guy and yeah all of those

photos look pretty darn consistent and I

love that the color grading has that

just very typical

1960s color grade and film stock style

over it it looks really awesome the team

at P laabs also came out with the

ability to add sound effects automatic

to your footage inside of their tool let

me show you how to use it there's a few

different ways that you can use this

feature I'll show you all of them

basically just go to p. art and you can

go down to the creation window down here

and just paste in your prompt we'll say

a dog barks at the park and go ahead and

click that sound effects button and then

click generate Okay let's listen to our

result

okay okay yeah I mean it sounds like a

dog barking there's something a little

off about it but not too bad you also

have the ability to add sound effects to

videos that you generate using an image

as well which is really helpful so all

you have to do is go to the image or

video button and select your image I

have this picture here of this Rog and

we'll go ahead and upload him and make

sure that your sound effects button is

selected and click generate Okay so

let's listen to our

result that kind of sounds like a cursed

frog so I don't think it did a very good

job on that one now I should also note

that you do have the ability to create

sound effects on previous videos that

you have generated all you have to do is

go to one of those previous videos like

here's the example that we used last

week with the house on fire you can just

click edit and when you click edit make

sure you hit the sound effects button

and you can type in the sound effects

that you want to hear so in this

instance we'll say a home on fire big

fire there we go and click generate cool

it generated three Clips in about 30

seconds let's go ahead and listen to

them this is number

one oh man that's a big

explosion

that one actually sounds really good I

think that one did a great job and this

is number

three so in this instance I think paika

actually did a really good job with the

sound effects we can basically pick one

now and click continue and we can

basically scrub through and pick the

exact spot where we want the sound

effects to be I'll just attach it to the

beginning and click attach to video and

there you go not to be outdone the team

at 11 Labs actually came out with their

own tool for typing in a prompt and

getting sound effects now it's only in

limited release right now but we have

access to the beta and I want to

showcase it with you today so I'm inside

of 11 Labs basically you just go to the

sound effects menu in the top left and

you just type in the sound effects that

you want to hear so for 11 Labs I want

to test it with that Frog video from

earlier so we're going to say a frog in

the park and we will generate sound

effects and after about 10 seconds It

produced five different Generations

let's listen to a few of them here this

is number

one sounds like a

frog sounds like a

frog that's great so as you can see the

sound effects inside of 11 labs in our

opinion sound a little bit better at

this point but I want you to tell me

which sound effects you like better and

that brings us to our game of the week

for our game this week I am going to

play you three sound effects of a

monster roaring I prompted the sound

effects in 11 labs pabs and I also went

to Art list which is a stock music and

sound effects company and downloaded a

monster roaring sound effect I want you

to tell me which one came from which

tool okay are you ready this is number

one this is number

two and this is number

three okay if you think you know the

answer let us know in the comments of

this video we will randomly select one

of the winners to receive a free AI film

making sweatshirt from Curious Refuge

there was also a research paper this

last week that came out called style

gaussian there are some pretty cool

examples over on the research page

basically the way it works it can take

an uploaded reference image and then

project that style into a gaussian Splat

the reason why that's cool is pretty

soon just like using Runway gen 1 you'll

be able to project a style into your

augmented reality so imagine being in an

apple Vision Pro and as you look around

the entire world is just converted into

an anime style or in the style of Pixar

that's what this early technology is

pushing for and I think it's going to

have huge implications for the future of

interactive entertainment this last week

we also dropped a new podcast episode

with Renard Jenkins who's the president

of simpy Renard shared some really great

insight into how AI operates at a base

level and how we can work together to

create better representation in AI

generated imagery we also dropped a

short film called another by Dave Clark

the short film actually Blends

live-action footage with AI generated

visuals and sometimes it's kind of hard

to tell what's Ai and what's live action

so we're actually calling this new

medium live gen we think that it really

is a prototype to what the future of

Cinema is going to be like Dave did an

awesome job on the project and he

collaborated with some really talented

artists like Rob Neer horor to create

the visual effects for the piece I

highly recommend checking it out the CEO

of Runway also really enjoyed it so uh

congrats Dave that's some really great

exposure we also came across this

interview with the CTO of open AI

talking about Sora and we have an update

on the timeline it actually sounds like

Sora will be coming out later this year

and he actually hinted that it could be

within the next few months he also said

that they are looking to incorporate

audio into Sora so may be very similar

to the way that we have sound effects

inside of P labs and Sora will not be

able to reproduce public figures which

is probably for the best especially

considering we are in an election year

and while we're talking about open AI

there was a page that was leaked only

for a few minutes but a few people found

it online and it basically was the chat

GPT 4.5 turbo release page and basically

now we have hints that it will probably

be released in June of 202 before we

definitely cannot wait to use it and

while we're talking about open AI just

really quickly we should note that Sam

ultman is back on the board and they

have reshuffled their board after all of

that crazy drama from back in November

we actually came across this really

interesting demo over on X where a robot

had been integrated with open Ai and it

actually has the ability to look around

the world and interpret its surroundings

so in the demo the robot is looking

around and it can say that it's looking

at a table with an apple on the table

and this robot is somewhat humanlike so

it's a really interesting use of the

technology and of course we're very

curious to see how this evolves in the

coming months and to flip the script

just a little bit we are going to go

over to Elon Musk who actually announc

that grock will be open source starting

this week it's really interesting to see

how X is positioning its AI model in

relation to open Ai and how that may

play out with the lawsuits that are

underway and that brings us to our AI

films of the week I am very excited to

show you this week's film there's some

of the best films that we've had in this

show up to this point the first one is

this film called The Mermaid by Zahir

Khan and it has some of the most

incredible shots that we've seen up to

this point the VFX is absolutely

breathtaking they clearly went in and

used the motion brush tool inside of

runway for a few shots and it looks

great and I think this is a great

example of working around some of the

limitations that are inside of these

tools they really have a very solid

understanding of narrative storytelling

and how to convey information while

working around some of the challenges

that can pop up whenever you work with

AI tools now they said they used a

mixture of Runway Leonardo sunno chat

GPT After Effects Premiere and art list

which is very interesting because we

actually teach most of those tools

inside of our program so it's cool to

see that some of the best AI filmmakers

are using some of these processes that

we've been able to develop with our

community the next film that I want to

shout out is uncanny Harriet by uncanny

Harry it's basically a music video that

uses a sunno song

and he used a combination of runway for

the visuals Leonardo and the lips were

synced using p Labs he also used Linds

go to add in a style transfer and the

result is one of the most convincing and

interesting music videos that we've seen

using AI up to this point and our final

film of the week is called Escape by

Marco it really features some great

curation and it's basically an AI mood

film that just is exploring an idea and

it has these characters that are in the

woods and there's like this pink smoke I

think it's a really good music video

concept and I think they did a great job

picking and choosing the right shots for

their film okay I hope you enjoyed this

week's episode of AI film news of course

you can get AI film news delivered

directly to your inbox every single

Monday by subscribing over at curious

refuge.com we cover more news in that

newsletter than what we talk about on

this show so if you want the latest and

greatest news be sure to subscribe and

of course you can always subscribe to us

here on YouTube give us a like if you

enjoyed this video thank you so much for

watching we'll see you in the next one