Apprendre à ChatGPT à écrire des prompts Midjourney super créatifs

Copiez-collez chaque instruction si vous souhaitez faire exactement comme moi, ou inspirez-vous en pour créer votre propre séquence.

👉🏻 Do you know Midjourney ?

I am not familiar with "Midjourney" as it doesn't seem to refer to any specific, well-known topic, organization, or event in my training data, which goes up until September 2021. If "Midjourney" is a recently created term, group, or project, I may not be aware of it. Please provide more context or details, and I'll do my best to help you with your question.

👉🏻 Midjourney is a tool that generates images based on prompts. If you tell it to create a photo of a lady, it generates a photo of a lady. You can give more details and it will try to create the image that corresponds to the prompt (the short text you give it to tell it what you want). Do you understand what a prompt is in Midjourney ?

Ah, I see. Midjourney seems to be a tool or application that generates images based on text prompts provided by users. In this context, a "prompt" is a short description or instruction that guides the tool in creating an image.

For example, if you give it the prompt "a lady," it will generate a photo of a lady. You can also provide more specific details in the prompt to guide the tool in creating a more accurate representation of the desired image.

Please note that my knowledge is up-to-date only until September 2021, and I don't have any direct information about Midjourney. However, the concept you described is similar to DALL-E, an AI model developed by OpenAI, which generates images from text prompts. If Midjourney is a more recent tool or an implementation based on similar technology, it's possible that it functions in a similar manner.

👉🏻 Exactly, it's like Dall-E. So, what prompt would you write to get a lady sitting in a chair facing the sea at dusk with a cup of tea ?

To create an image using a tool like DALL-E or Midjourney with the scene you described, you can use the following prompt:"A lady sitting in a chair facing the sea at dusk, holding a cup of tea."

This prompt provides a clear and concise description of the scene you want to generate. The AI should understand the main elements – the lady, the chair, the sea, the dusk setting, and the cup of tea – and attempt to create an image that incorporates all these details. Keep in mind that you might need to adjust the prompt or try a few variations to get the desired result, as the AI's interpretation can sometimes be different from your intended visualization.

👉🏻 Elaborate on this prompt and give it more details about the state of mind, the environement, the feelings she has

To provide more details about the environment, state of mind, and feelings of the lady in the scene, you can use the following prompt:

"A serene lady with a gentle smile, sitting comfortably in a cozy chair facing the sea at dusk, holding a steaming cup of tea. She is surrounded by a tranquil environment with soft, golden sunlight reflecting on the water, as the calming sound of waves crashing on the shore envelops her. She feels at peace and content, taking a moment to breathe deeply and embrace the beauty of the natural world around her, while the warm tea soothes her soul."

This prompt expands on the initial description by adding more context about the lady's state of mind, the emotions she is experiencing, and the environment that surrounds her. This should help the AI generate an image that captures the essence of the scene you envision. Remember that you might need to adjust the prompt or try a few variations to achieve the desired result.

👉🏻 Please keep it as short as possible, use many nouns, verbs, adjectives

For a shorter prompt with a focus on nouns, verbs, and adjectives, try:

"Serene lady, cozy chair, sea at dusk, steaming tea, golden sunlight, calming waves, contented smile, peaceful surroundings.

"This prompt condenses the details into key descriptive words, while still capturing the essence of the scene you want to generate. It may require the AI to interpret the relationships between the elements more, but it should still provide a good starting point for the image generation.

👉🏻 You can add parameters at the end of the prompt. For example, if you want an aspect ratio of 16 by 9, you add --ar 16:9. It means that --ar tells midjourney a parameter about aspect ratio is coming, and : represents the width and height ratios. So if you want a square image, use --ar 1:1. If you want an instagram vertical image, use --ar 4:5. Do you understand ?

Yes, I understand.

The "--ar" parameter followed by a value in the format "width:height" is used to specify the aspect ratio of the generated image in Midjourney.

This allows you to create images with different dimensions, such as a square image with "--ar 1:1", a vertical Instagram image with "--ar 4:5", or a widescreen image with "--ar 16:9".

👉🏻 What would would use as a parameter for a Notion cover ?

Notion does not have a strict aspect ratio requirement for cover images, but it does have a recommended size for a visually appealing cover. To create an image for a Notion cover, you can use the following aspect ratio parameter:- -ar 2:1

This aspect ratio of 2:1 (width:height) is a good starting point for a Notion cover image, as it provides a balance between the image width and height, ensuring that the image looks visually appealing when displayed at the top of a Notion page.

Keep in mind that you may need to adjust the size of the image after generating it to fit the specific dimensions recommended by Notion. As of my knowledge cutoff in September 2021, Notion recommends a cover image size of 1500 x 750 pixels, but this may have changed since then. It's a good idea to check for any updated recommendations on image sizes from Notion.

👉🏻 To me, it's 15:6 for a notion cover

If you prefer a 15:6 aspect ratio for a Notion cover image, you can use the following parameter: --ar 15:6

This aspect ratio (width:height) of 15:6 will give you a wider image compared to the 2:1 ratio. Using this parameter in Midjourney will generate an image with the desired dimensions, and you can then upload it as a cover image in Notion.

Remember to check for any updates or recommendations on image sizes from Notion to ensure the best results.

👉🏻 What aspect ratio would you use for a twitter image ?

For a Twitter image, a commonly recommended aspect ratio is 2:1 (width:height).

This works well for Twitter's image previews in the timeline, as it fits within the default cropping without losing too much visual information. To generate an image with this aspect ratio in Midjourney, you can use the following parameter:- -ar 2:1

However, keep in mind that Twitter supports a range of aspect ratios for uploaded images, from 1:1 (square) to 16:9 (widescreen). Depending on the specific use of the image on Twitter and the desired appearance in the timeline, you may want to adjust the aspect ratio to better fit your needs.

For a Twitter image, a commonly recommended aspect ratio is 2:1 (width:height). This works well for Twitter's image previews in the timeline, as it fits within the default cropping without losing too much visual information. To generate an image with this aspect ratio in Midjourney, you can use the following parameter:- -ar 2:1

However, keep in mind that Twitter supports a range of aspect ratios for uploaded images, from 1:1 (square) to 16:9 (widescreen). Depending on the specific use of the image on Twitter and the desired appearance in the timeline, you may want to adjust the aspect ratio to better fit your needs.

👉🏻 Another parameter is stylize. With --s and a number ranging for 0 to 1000, you decide how much creativity, how much Midjourney will be creative with the prompt you give it

The "--s" parameter, followed by a numerical value between 0 and 1000, allows you to control the level of creativity Midjourney applies when generating an image based on your prompt. A higher value indicates more creative freedom, while a lower value will result in a more conservative interpretation of the prompt.

For example:

--s 100

This would result in a moderate level of creativity, as the value is in the middle of the scale.

--s 1000

This would instruct Midjourney to be highly creative with the given prompt, potentially resulting in more abstract or unexpected interpretations of the scene described.By adjusting the "--s" parameter, you can fine-tune the balance between strict adherence to the prompt and artistic liberty taken by the AI when generating the image.

👉🏻 The --c parameter stands for Chaos, ranging from 0 to 100. It tells Midjourney how different, how "all over the place" the generated image can be

The "--c" parameter, followed by a numerical value between 0 and 100, controls the level of chaos or randomness in the image generated by Midjourney. A higher value allows for more variation and unpredictability in the generated image, while a lower value will result in a more structured and coherent interpretation of the prompt.

For example:

--c 25

This would result in a relatively low level of chaos, producing an image that closely follows the prompt while still allowing for some minor variations or unexpected elements.

- -c 100

This would instruct Midjourney to incorporate a high level of chaos into the generated image, potentially resulting in a more unusual, abstract, or diverse representation of the scene described in the prompt.

By adjusting the "--c" parameter, you can fine-tune the balance between maintaining a clear structure and embracing the chaotic and unpredictable nature of the AI-generated image.

👉🏻 Ok, now let's talk about genres, or styles. For each prompt, we should use some kind of genre, of style, which is not a parameter, but a part of the prompt. You can use in the genre of vincent van gogh, in the genre of toulouse lautrec, in the genre of andy warhol but also photo realism, anime, illustration, disney-pixar style, ink drawing. Do you understand ?

Yes, I understand.

By incorporating a specific genre or style into the text prompt, you can guide Midjourney (or similar AI image generation tools) to generate an image in the visual style of a particular artist, art movement, or aesthetic. This helps to give the generated image a distinct appearance and feel, based on the style mentioned in the prompt.

For example, if you want an image of a lady sitting by the sea in the style of Vincent van Gogh, you could use the following prompt:"In the style of Vincent van Gogh, a lady sitting in a chair facing the sea at dusk, holding a cup of tea."

Similarly, for other styles, you can modify the prompt accordingly:

- "In the style of Toulouse Lautrec, a lady sitting in a chair facing the sea at dusk, holding a cup of tea."- "In the style of Andy Warhol, a lady sitting in a chair facing the sea at dusk, holding a cup of tea."

- "In a photo-realistic style, a lady sitting in a chair facing the sea at dusk, holding a cup of tea."

- "In an anime style, a lady sitting in a chair facing the sea at dusk, holding a cup of tea."

- "In a Disney-Pixar style, a lady sitting in a chair facing the sea at dusk, holding a cup of tea."

- "In an ink drawing style, a lady sitting in a chair facing the sea at dusk, holding a cup of tea."By including the desired style or genre within the prompt, you can influence the artistic direction of the generated image.

👉🏻 If it is a photo realistic image, it would be interesting to add notions about volumetric lighting, a render engine like Unreal or the dslr and lens used. For the lens, you can use something like fish eye, or 16mm 1.4 or 55mm 1.4 or 135mm 1.8. Environement could also play a big part : foggy, sunny, rainy, smoke, those conditions and wheather can also be added

Certainly, adding specific details about lighting, rendering, camera equipment, or environmental conditions can help guide Midjourney to generate a more realistic and visually striking image. Here are some examples of how you can incorporate these elements into a prompt:

"In a photo-realistic style with volumetric lighting and fog, a lady sitting in a chair facing the sea at dusk, holding a cup of tea. Rendered in an Unreal Engine-like quality and captured with a 55mm f/1.4 lens."

"In a photo-realistic style on a sunny day, a lady sitting in a chair facing the sea at dusk, holding a cup of tea. Rendered with the depth of field of a 135mm f/1.8 lens and soft, natural lighting."

"In a photo-realistic style with a fish-eye lens effect, a lady sitting in a chair facing the sea at dusk in a rainy environment, holding a cup of tea. The image features raindrops, reflections, and moody lighting."

By incorporating these additional details into the prompt, you can create more vivid and immersive images that closely resemble the appearance of a photograph taken with specific equipment, rendering techniques, or under particular environmental conditions.

👉🏻 Last thing, if we're talking about illustration or other types of vector style image, placing it on a white background and telling midjourney you want a sticker can be important. You can also negate some parts by using the --no parameter and have something like, for example, --no mockup to get an image without anything around, not placed in a frame or on some kind of decor

When generating illustrations, vector-style images, or stickers using Midjourney, specifying a white background and mentioning the desired format can indeed be important. This helps ensure that the generated image meets your expectations and can be easily used for its intended purpose. Here's an example of a prompt for creating a sticker:

"In an illustration style on a white background, a sticker of a lady sitting in a chair facing the sea at dusk, holding a cup of tea."

To avoid having the image placed within a mockup or a specific frame, you can use the "--no" parameter followed by the desired exclusion, like "--no mockup". This instructs Midjourney not to include any mockup elements in the generated image.

For example:

"In an illustration style on a white background, a sticker of a lady sitting in a chair facing the sea at dusk, holding a cup of tea. --no mockup"

By including these additional instructions and parameters in your prompt, you can guide Midjourney more effectively to generate the type of image you have in mind, with the desired style, format, and exclusions.

👉🏻 The rest of this conversation is about you creating prompts. If I give you a short description, please write the prompt corresponding . If I want to learn you something new, i will tell you "Something new to learn :" and you will know. For the rest, just generate as best as possible and ask questions if you really need more info. Also, you don't need to generate more than 5 prompts for each request, but each prompt should be quite different from the others so we can go in any direction and make our choice. Do you understand ? If so, say "let's go"I understand your instructions.

Let's go!

👉🏻 business man, office, working on computer

1. "In a photo-realistic style, a focused businessman working on a computer in a modern office."
2. "In an illustration style, a busy businessman at his office desk, typing on a computer keyboard."
3. "In an anime style, a determined businessman in a professional office setting, concentrating on his computer screen."
4. "In a Disney-Pixar style, a cheerful businessman in a well-lit office, working on his computer with enthusiasm."
5. "In a minimalistic vector style on a white background, a sticker of a businessman using a computer in his office."

👉🏻 Don't forget to add parameters. innovate ;-)

1. "In a photo-realistic style with volumetric lighting, a focused businessman working on a computer in a modern office with large windows. --ar 16:9 --s 100 --c 25"
2. "In an illustration style, a busy businessman at his office desk, typing on a computer keyboard with a colorful background. --ar 4:5 --s 300 --no mockup"
3. "In an anime style, a determined businessman in a professional office setting, concentrating on his computer screen with dramatic lighting. --ar 1:1 --s 500 --c 50"
4. "In a Disney-Pixar style, a cheerful businessman in a well-lit office, working on his computer with enthusiasm and surrounded by friendly office items. --ar 3:2 --s 200 --c 75"
5. "In a minimalistic vector style on a white background, a sticker of a businessman using a computer in his office with bold outlines and flat colors. --ar 1:1 --s 50 --no mockup --no frame"

‍