[Runway] The arrival of the era where you can become a film director, Part 2. A true virtual film set. New features in video generation AI & Midjourney image generation AI prompts and V1 Video models.
Hello! This is Sakista!
I wrote this note the other day!
This time, I will introduce more
new technology for Runway Gen-4 References that I couldn't cover last time!
Once you learn this, you can freely reuse compositions and characters...!
Your creative expression will literally expand significantly!
If you learn this along with Part 1, it's like being on a virtual film set...
Release news!!!
Chat Mode was released by Runway on June 12, 2025!
I will write about Chat Mode in a future note as well!
Now, let's get to the main topic.
First, I will introduce how to process unwanted backgrounds using Runway.
This is what's known as chroma key compositing using a green screen.
That said, you don't need to prepare a physical green screen... you can process everything using Runway.
Just by adding this extra step, the accuracy of image referencing improves dramatically!
1. Make the background a green screen
What is a green screen
The effect of a green screen is a technique for freely replacing the background of videos or images, mainly used in movies, television, YouTube, and photography. Since green is the color furthest from human skin tone (reddish), it is less likely to cause detection errors.
In Runway, when referencing material to use in other images, it is easier to generate images as intended if there is no unnecessary information with a green screen.
If other elements like backgrounds are included, it can affect the reference.
First, prepare the image you want to use.

Make the background a green screen with Runway Gen-4 References.
I named the original image "Cast_A".
You can call up images registered with @.
Please see the previous note for detailed explanations.

It is also fine to leave it in the default form like IMG_1.

Prompt
Remove the background and ground from @Cast_A and place the girl against a flat green background (hex #04F404). Maintain the exact composition of the scene, and remove all contact shadows.
Japanese translation: Remove the background and ground from @Cast_A and place the girl against a flat green background (color code: #04F404). Maintain the exact composition of the scene, and remove all contact shadows.
I was able to cleanly green-screen even the fine hair.

Press the Reference button to use the completed image as a reference.

Registered as Cast_A_GS.

Prepare a background image.

Prompt
@Cast_A_GS is sitting at the table in @Location_1. Please ensure she is placed naturally within the scene, maintaining lighting and perspective consistent with @Location_1.
English translation: @Cast_A_GS is sitting at the table in @Location_1. Please ensure she is placed naturally within the scene, maintaining lighting and perspective consistent with @Location_1.
Completed image
Generated with an aspect ratio of 16:9.

Because the person image was green-screened to cut out background element information,
the cafe image wasable to be blended cleanly with only the person without being affected.
Try placing pancakes on the table
First, prepare the materials.

Prompt
Remove the background and surface from @Pancakes and place the plate of pancakes against a flat green background (hex #04F404). Maintain the exact composition of the scene and remove all contact shadows.
Remove the background and surface from @Pancakes and place the plate of pancakes against a flat green background (hex #04F404). Maintain the exact composition of the scene and remove all contact shadows.

Prompt
Keep the original composition of @Scene_1, and place a small version of the two-stack pancakes from @Pancakes_GS on the table. Make sure the placement looks natural and matches the lighting and perspective of the scene.
Keep the original composition of @Scene_1, and place a small version of the two-stack pancakes from @Pancakes_GS on the table. Make sure the placement looks natural and matches the lighting and perspective of the scene.

Convert to video

Completed video
Video generation prompt
The woman gazes out at the horizon. Behind her, the ocean waves roll in and out in a rhythmic flow. Nearby, people play and splash in the shallow water. A curtain above the rooftop sways gently in the wind. The camera remains fixed, capturing the entire scene with a subtle sense of atmosphere. Smooth, cinematic live-action style.
Translation: The woman gazes out at the horizon. Behind her, the ocean waves roll in and out in a rhythmic flow. Nearby, people play and splash in the shallow water. A curtain above the rooftop sways gently in the wind. The camera remains fixed, capturing the entire scene with a subtle sense of atmosphere. Smooth, cinematic live-action style.
I used "ElevenLabs" for the ambient sound.ElevenLabs" for the ambient sound.
Prompt for music generation
From a beachside cafe, the gentle sound of waves rolling in and out with a calm, rhythmic pattern. Only the soft, soothing sound of the sea continues peacefully. No voices, no music just the ocean.
From a beachside cafe, the gentle sound of waves rolling in and out with a calm, rhythmic pattern. Only the soft, soothing sound of the sea continues peacefully. No voices, no music just the ocean.
2. How to arrange characters (composition) using images
This time, I will show you how to create a prompt using images to decide "what kind of composition to use" while increasing the number of characters.
First, prepare the image of the character you want to use.

Remove the background to create a green screen, just like before.
This time, I registered the names "Yuki" and "Kana".

Prepare an image for composition
A simple image like the one below is fine.

Once you have your materials ready, enter the prompt.

Prompt
@Yuki and @Kana are standing back-to-back as shown in @IMG_1, with both shown in profile (side view). The background is a vast grassy field stretching to the horizon under a cloudy sky.
English translation: @Yuki and @Kana are standing back-to-back as shown in @IMG_1, with both shown in profile (side view). The background is a vast grassy field stretching to the horizon under a cloudy sky.

I was able to generate it with the layout exactly as the reference image, but...!
I feel like I could generate this much just with the prompt alone! Haha
Let's make it a bit more complex


Prompt
@Yuki and @Kana are standing back-to-back as shown in @IMG_2, with both shown in profile (side view). The background replicates @IMG_2, featuring a sandy beach, an ocean stretching to the horizon, and a yacht floating in the upper right corner.
English translation: @Yuki and @Kana are standing back-to-back as shown in @IMG_2, with both shown in profile (side view). The background replicates @IMG_2, featuring a sandy beach, an ocean stretching to the horizon, and a yacht floating in the upper right corner.
Generation result

Try a different composition

Prompt
@Yuki and @Kana are sitting on the hood of a yellow classic car, as shown in @IMG_3. Reference @IMG_3 to place an American diner in the upper left of the background. The scene is set at night, with cinematic lighting and atmosphere.
English translation: @Yuki and @Kana are sitting on the hood of a yellow classic car, as shown in @IMG_3. Reference @IMG_3 to place an American diner in the upper left of the background. The scene is set at night, with cinematic lighting and atmosphere.

3. How to maintain consistency in images
Next, I will introduce generation using the Styles feature!
In creative work, maintaining consistency (tone and manner) is extremely important!
With that in mind, Runway Gen 4's Styles is a very powerful tool!
First, press Styles
There are default styles, but we will create a new one this time.
Please press the '+' button.

The create screen is now displayed.
There are two methods: entering a prompt directly, or using a reference image.
This time, we will upload the previous image as a style.

Image analysis has started.

The style prompt has been displayed!
It is very convenient because the AI analyzes it automatically.

The prompt you created is now displayed in 'Styles'.
Neonoir uses the automatically generated name. Of course, you can change it. Perhaps it is an expression combining 'neo' and 'noir' in French.

Insert a reference image and enter a prompt.
Please check 'Styles'.
By the way, when using styles, it is better to avoid expressions related to style in the prompt as much as possible.Runway Academy confirmed this.
Prompt
@Yuki and @Kana are sitting at the counter of an American diner. Behind them, a window reveals a view of a nighttime parking lot.
@Yuki and @Kana are sitting at the counter of an American diner. Behind them, a window reveals a view of a nighttime parking lot.

Generation result

Thanks to using 'Styles', the prompt is very simple and clean.
This allows me to focus on the composition, making it easier to make adjustments!
Completed video
Video generation prompt
The white-haired woman quietly speaks while the black-haired woman nods gently in response. A slow left-to-right dolly movement reveals the two seated at a reflective glass table in a retro American diner at night. Outside the neon-lit window, a vintage car rolls by, its headlights glinting on the glass. Purple-tinged smoke drifts lazily through the room as moody neo-noir lighting casts shifting shadows across the booth. The atmosphere feels still and cinematic, accompanied by a subtle background of soft jazz. Smooth live-action with a dreamy, stylized flow.
English translation: A white-haired woman speaks quietly, and a black-haired woman nods gently in response. A slow dolly shot from left to right reveals the two sitting across from each other at a reflective glass table in a retro American diner at night. Outside the neon-lit window, a vintage car passes by, its headlights glinting on the glass. Purple-tinged smoke drifts slowly through the shop, and moody neo-noir lighting casts flickering shadows inside the booth. The air is still, a cinematic atmosphere hangs in the air, and soft jazz plays in the background. The footage is smooth, live-action, and has a dreamy, stylized flow.
The ambient sound and background music were generated using "ElevenLabs".
Here is how to generate realistic models that don't look like AI.
That's all for Runway Gen-4 References for now!
Did you enjoy it?
I would be very happy if you found it even a little bit helpful!
…but actually, there is more.
Some "interesting ways to use it" that I couldn't cover this time!
If I put it all together, the content would get cluttered,
so I would like to deliver it again as a separate note.
If you'd like,I would be happy if you followed me and waited for it!
I introduce Midjourney V7 references in the following note.
Finally, here are the Midjourney NEWS and prompts!
Midjourney V1 Video Model is coming soon!
On June 12, 2025, the Video Rating Party started from Midjourney.
A Rating Party is a customary event held before a Midjourney release.
Simply put, you just tap to choose "which of the two works is better?". New features will be released based on global rating results.

Points to note
Please do not choose videos that are clearly broken (even if they are funny)
You can select using the "1 (left)" and "2 (right)" keys on your keyboard
If you are unsure, please press "Skip (3 key)"
If you see NSFW (inappropriate) content, please "report" it
The videos this time do not represent the quality of the final model
Low-quality videos are intentionally included, and we are collecting data to learn preferences
I'm looking forward to Midjourney's aesthetically excellent Video model!
Midjourney Prompts
Midjourney V7 has been released, and updates have been continuing every few weeks.
Omni Reference, EXP, and Q4 are particularly noteworthy features.
Parameter: EXP
For information on what EXP is, please refer to the following note!
Parameters known for their detail, creativity, and dynamism also pair well with motion blur.

Prompt
On a cloudy beach, an 18-year-old Japanese girl and a cat appear as a two-piece pair in a full-body shot. The girl has a bob haircut, and their motion is captured with strong blur. Shot from a low angle with shallow depth of field, the photo is in color with low contrast and natural tones. It was taken on 35mm film with prominently visible grain, giving it a raw, analog texture. --ar 2:3 --exp 5 --v 7
On a cloudy beach, an 18-year-old Japanese girl and a cat appear as a two-piece pair in a full-body shot. The girl has a bob haircut, and is in a scene with motion alongside the cat. Shot from a low angle, the strong motion blur and shallow depth of field are impressive. The photo is in color with low contrast and natural tones. It features the prominent grain characteristic of 35mm film, giving it an analog texture.
Pairs well with high-quality, dynamic expressions

In the clear waters of the Caribbean, a group of meerkats swims toward the camera in a selfie-style composition, making direct eye contact. Sunlight pierces the surface, casting magical beams throughout the scene. A colorful coral reef stretches behind them, and a Caribbean island is faintly visible above the waterline. Captured with a Canon EOS R5 and RF 15-35mm F2.8L lens inside a Nauticam housing, the photo is ultra-realistic and highly detailed, showcasing the textures of the meerkats’ fur and the serene, weightless atmosphere of the underwater world. --ar 3:2 --exp 15 --v 7
In the clear waters of the Caribbean, a group of meerkats swims toward the camera. It is a selfie-style composition, and the meerkats are making eye contact with the camera. Sunlight pierces the water surface, illuminating the underwater scene with magical beams of light. A colorful coral reef spreads out in the background, and a Caribbean island is faintly visible above the water surface. Captured with a Canon EOS R5, RF 15-35mm F2.8L lens, and Nauticam housing, this photo is ultra-realistic and highly detailed. It delicately depicts everything from the meerkats' fur and movement to the sense of weightlessness underwater.
New Parameter: Q4
The "--q [X]" parameter is also called the quality parameter.
While it was --q 2 up until V6, --q 4 has been released for V7.
If you want to increase the quality of your generated images, please try using it!

Prompt
A hyper-realistic, high-resolution view of a Caribbean beach, shot from just above the water surface. The clear sea reveals coral and fish beneath the surface, while the white sandy beach is dotted with large rocks and palm trees. Gentle afternoon sunlight illuminates the entire scene, with distant clouds and the horizon visible in the background. The image features vivid but natural colors, evoking a calm, tropical atmosphere. --ar 3:4 --quality 4 --v 7
An ultra-realistic, high-definition beach scene shot from just above the surface of the clear Caribbean sea. Large rocks and palm trees are scattered across the white sandy beach, and coral and fish can be seen inside the transparent sea. Gentle afternoon sunlight softly illuminates the entire scene, with clouds and the horizon spreading out in the distance. Depicted in vivid yet natural colors, this composition has a calm, tropical atmosphere.
Summary of prompts using parameters EXP and Q4

Prompt
On a boat under bright sunlight, an 18-year-old Japanese girl with a pink bob haircut poses for a selfie, looking directly at the camera. She wears a Korean-style fashion, sitting with her legs widely spread and leaning forward in a dramatic pose. The scene is softly lit with natural light. The photo is ultra-realistic, with vivid colors and a shallow depth of field that brings strong focus to her expression and posture. The background is a grassy field and a cloudy sky. --ar 16:9 --exp 20 --v 7
On a boat bathed in bright sunlight, an 18-year-old Japanese woman with a pink bob haircut stares straight at the camera, taking a selfie. Dressed in Korean-style fashion, she sits with her legs spread wide and leaning forward, striking a dramatic pose. The scene is softly illuminated by natural light, and the vivid colors and shallow depth of field make for an ultra-realistic photo that highlights her expression and posture. A grassy field and cloudy sky spread out in the background.

Prompt
An 18-year-old Japanese girl stands on a bright sandy beach, facing the camera with her arms extended outward. She has pale skin and white bob-cut hair, with a neutral but slightly cheerful expression. She wears a wet, oversized white t-shirt and a dark gray plaid pleated mini skirt. The background features a vivid tropical seascape with calm turquoise waters, distant landforms, and a bright blue sky dotted with cumulus clouds. Strong natural sunlight highlights the vibrant colors and casts minimal shadows, giving the image a youthful, bright, and carefree feel. --ar 2:3 --exp 3 --quality 4 --v 7
An 18-year-old Japanese woman stands on a bright sandy beach, facing the camera head-on with her arms spread wide. She has fair skin and white bob-cut hair, with a neutral yet slightly cheerful expression. She is wearing a wet, oversized white T-shirt and a dark gray plaid pleated skirt. The background features a calm, emerald-blue sea, distant land, and a vivid blue sky with cumulus clouds; strong natural light accentuates the color contrast, with almost no shadows. The overall atmosphere is bright, youthful, and free.

A close-up portrait of an 18-year-old Japanese woman taken on a hot summer day at a sports ground. Her black bob haircut gently moves in the breeze, while beads of sweat glisten naturally on her forehead and cheeks. She's wearing a simple white T-shirt, reflecting a casual summer sports style. Her skin is illuminated by the strong midday sun, creating a natural, realistic glow with slight redness on her cheeks. The background shows a softly blurred sports field with fence lines and hints of dust or chalk, evoking a calm and candid everyday moment. --ar 2:3 --quality 4 --v 7
A close-up portrait of an 18-year-old Japanese woman taken at a sports ground in summer. Her black bob hair sways in the breeze, sweat beads on her forehead and cheeks, and her skin glows naturally under the intense midday sun. She is in a simple white T-shirt, a casual club activity style, with a softly blurred fence and white ground lines in the background. A faint flush on her cheeks adds to the realistic, everyday atmosphere.

A powerful black muscle car is captured mid-air in a dramatic low-angle shot as it jumps over rugged, rocky terrain at high speed. Its front wheels are lifted off the ground while the rear wheels are just beginning to rise, sending clouds of dust and flying debris into the air. Behind it, two grey military helicopters hover low above the treetops, tilted slightly forward in a tense, aggressive pursuit. A massive plume of black and brown smoke billows in the background, suggesting a recent explosion. The terrain is uneven, dotted with green vegetation and exposed brown rock, while mist-covered mountains rise beneath a cloudy sky, adding atmospheric depth. The entire scene conveys a cinematic intensity filled with speed, danger, and adrenaline-fueled action. --ar 3:2 --exp 20 --quality 4 --v 7
An action scene captured from a low angle, showing a powerful black muscle car jumping over rugged, rocky terrain at high speed, suspended in mid-air. The front wheels are lifted high, the rear wheels are just leaving the ground, and dust and rock debris are flying everywhere. In the background, two gray military helicopters hover just above the treetops, pursuing in a tense, forward-leaning posture. Behind the car, a massive plume of black and brown smoke billows, suggesting a recent explosion. Under a cloudy sky, mist-covered mountains, deep green vegetation, and exposed rocks give the scene a wild, natural feel. The entire image depicts a cinematic, intense moment filled with speed, thrill, and tension.
Thank you for reading until the end!
See you in the next note!
いいなと思ったら応援しよう!
よろしければサポートお願いします! いただいたサポートはクリエイターとしての活動費に使わせていただきます!