Think Like an AI & Prompt Better in Midjourney v6

Tokenized AI by Christian Heidorn
20 Jan 202415:13

TLDRThe video discusses the intricacies of creating detailed prompts for Midjourney version 6, an AI image generation tool. The host compares the new version to Dolly 3 in a prompt coherence challenge and shares tips on how to replicate specific photos. The process involves starting with a basic prompt and iteratively adding details to refine the output. The video showcases examples, including replicating a studio photo and a candid image of a woman on a balcony, emphasizing the need for a detailed and specific language in the prompts. The host also highlights the challenges with hands and feet in the generated images and the potential of adjusting the stylization level for different effects. The video concludes with an invitation for viewers to share their thoughts on Midjourney version 6 and to explore the host's course for mastering the tool.

Takeaways

  • 📈 **Prompt Coherence Improvement**: Midjourney v6 has significantly improved prompt coherence, making it easier to generate more coherent images.
  • 🤔 **Detail-Oriented Prompting**: Achieving great results in v6 requires detailed and specific prompts, which can be challenging and requires a good vocabulary and imagination.
  • 🖼️ **Iterative Approach**: Start with a basic prompt and incrementally add details to refine the output, using trial and error to reach the desired outcome.
  • 🧑 **Subject Focus**: Positioning the subject correctly and focusing on their details, such as clothing and pose, is crucial for replicating specific images.
  • 🌄 **Background Adjustments**: Paying attention to the background and setting the right atmosphere can greatly enhance the final image and bring it closer to the reference.
  • 👕 **Clothing and Accessories**: Including specific details about clothing and accessories helps in generating images that closely match the reference material.
  • 🤲 **Challenges with Hands and Feet**: Midjourney v6 still struggles with accurately rendering hands and feet, which can be a significant issue in some prompts.
  • 🎨 **Stylization Level Impact**: Adjusting the stylization level can have a strong impact on the final image, making it more or less subject-centric.
  • 🌐 **Cultural and Ethnic Considerations**: Specifying ethnicity and cultural elements can help in creating more authentic and diverse images.
  • 📱 **Modern Props**: Including modern elements like smartphones can make the image more relatable and current, but the accuracy of these details can vary.
  • 🌞 **Lighting and Positioning**: The direction of light and the positioning of the subject can significantly affect the mood and the realism of the image.

Q & A

  • What is the main challenge in creating great prompts in Midjourney version 6?

    -The main challenge is to unlock the full power of the improved prompt coherence by being extremely specific with the prompts, which requires a lot of imagination and a good vocabulary.

  • How does the speaker begin the process of replicating a specific photo in Midjourney version 6?

    -The speaker starts with a bare minimum prompt and then iteratively adds more details, inching towards the final outcome through a process of trial and error.

  • What was the first detail added to the initial prompt about the young man on an orange bean bag?

    -The first detail added was that the image should be against a light teal background.

  • How does the speaker describe the process of iteratively refining the prompt to match the reference image?

    -The process involves starting with a basic prompt and progressively adding more specific details about the composition, subject, and atmosphere to achieve a closer match to the reference image.

  • What issue does the speaker identify with Midjourney version 6 regarding the depiction of hands?

    -The speaker notices that Midjourney version 6 has a bit of an issue with hands, which seems to have gotten worse from version 5 to version 6.

  • How does the speaker suggest improving the prompt to get the desired ethnicity and atmosphere in the image?

    -The speaker suggests specifying the ethnicity, such as 'Czech woman,' and adding a sentence about the desired atmosphere and vibe, such as 'Serene capturing the essence of a quiet personal retreat.'

  • What is the speaker's opinion on the importance of adding details about the overall atmosphere and vibe of the image?

    -The speaker believes that adding details about the atmosphere and vibe gives the images a nice polish and, although some people argue it doesn't do much, the speaker finds it to be a subtle but important aspect.

  • What is the speaker's observation about the variations in different rerolls of a prompt in Midjourney version 6?

    -The speaker notes that there will be typical variations between different rerolls of a prompt, such as slight changes in the position of objects or the subject, and these variations are to be expected.

  • How does the speaker plan to address the issue of hands and feet depiction in Midjourney version 6?

    -The speaker acknowledges the issue and suggests that it might be possible to tweak and improve the results with a few rerolls, also expressing hope that Midjourney will improve these aspects in future updates.

  • What is the impact of adjusting the stylization level in Midjourney version 6?

    -Adjusting the stylization level can significantly impact the output, with higher stylization levels generally making the images more subject-centric.

  • What advice does the speaker give for those who want to learn more about prompting effectively in Midjourney version 6?

    -The speaker is working on a new sub-module within their course to teach an entirely new prompting framework for Midjourney version 6 and encourages interested individuals to check out the links in the video description.

Outlines

00:00

🎨 Replicating Specific Photos in Midjourney V6

The video discusses the challenge of creating high-quality images with Midjourney V6, emphasizing the importance of prompt coherence. After a prompt coherence challenge with Dolly 3, the focus shifts to replicating a specific studio photo using a reference image. The process involves starting with a basic prompt and incrementally adding details to refine the image, such as the subject's position, clothing, and background. The video highlights the iterative nature of prompting, the need for a good vocabulary, and the use of specific language to achieve desired results. Despite improvements in coherence, the video also points out issues with hands and poses in the generated images.

05:01

🖼️ Refining the Image to Match a Reference

The video script details a step-by-step process of refining a prompt to generate an image closer to a given reference. It starts with a simple prompt and progressively adds details about the subject's position, clothing, and background to align with the reference image. The process includes specifying the subject's ethnicity, adjusting the background color, and refining the prompt to capture the desired atmosphere. The video acknowledges the limitations of Midjourney V6, particularly with hands and feet, and suggests that higher stylization can impact the focus and subject of the image.

10:03

🌟 Achieving Specificity with Detailed Prompting

The paragraph outlines a more complex example of using Midjourney V6 to create a specific image of a woman on a balcony, assumed to be in Prague. The process involves starting with a broad prompt and iteratively adding details to refine the image, including the subject's focus, clothing, hairstyle, and the architectural background. The video discusses the importance of detailed prompting for achieving specific results, the challenges with hands and feet rendering, and the potential for variation between different prompt rerolls. It concludes with the impact of stylization levels on the final image and an invitation for viewers to share their thoughts on Midjourney V6.

15:03

📝 Conclusion and Engagement Invitation

The final paragraph serves as a conclusion to the video, summarizing the process of achieving specific image results using detailed prompts in Midjourney V6. It acknowledges the ongoing development and potential improvements in future versions of the software. The speaker invites viewers to share their experiences and challenges with Midjourney V6 in the comments section and promotes further learning through a course and additional resources linked in the video description.

Mindmap

Keywords

Prompting

Prompting in the context of AI refers to the process of providing input or instructions to an artificial intelligence system to generate a specific output or response. In the video, the term is used to describe how to effectively communicate with an AI to produce desired images. The video emphasizes that while basic prompting is easy, creating great prompts requires a detailed and iterative approach.

Coherence

Coherence in the context of AI-generated content refers to the logical and meaningful connection between different parts of the output. The video discusses how Midjourney version 6 has improved in prompt coherence, meaning that the generated images are more likely to be logically connected and relevant to the prompt provided.

Midjourney

Midjourney is the name of an AI system that is being discussed in the video. It is likely a reference to a specific AI model or software used for generating images based on textual prompts. The video focuses on how to use Midjourney version 6 effectively to replicate specific photos.

Iteration

Iteration in this context refers to the process of repeatedly making small changes to a prompt and reviewing the results until the desired output is achieved. The video script emphasizes the importance of an iterative approach when prompting AI to refine the generated images closer to a specific reference image.

Reference Image

A reference image is a specific example or sample that serves as a guide for the AI to follow when generating new images. In the video, the creator uses reference images from the internet to guide the AI in producing images that closely resemble the originals.

Aesthetic

Aesthetic in this context refers to the visual appeal or beauty of the images generated by the AI. The video mentions that even with good prompt coherence, the images need to be aesthetically pleasing to be considered successful.

Parameters

Parameters are the specific details or aspects that are added to the prompt to guide the AI in generating the desired output. The video provides examples of how adding various parameters, such as background color, clothing, and pose, can influence the final image produced by Midjourney version 6.

Stylization

Stylization refers to the artistic or visual style applied to the generated images. The video discusses how adjusting the stylization level in Midjourney version 6 can impact the focus and subject-centric nature of the images, making them more or less abstract.

Hands and Feet

The video points out that Midjourney version 6 has some challenges with generating accurate and realistic hands and feet in the images. This is an area where the AI's output is noted to have issues and is subject to improvement.

Ethnicity

Ethnicity in the context of the video refers to the specific cultural or national characteristics of a person that the AI can be prompted to include in its generated images. The video script includes an example where the ethnicity of the subject is specified as 'Czech' to match the desired setting.

Atmosphere

Atmosphere in this context refers to the mood or feeling that the generated image is intended to convey. The video script describes adding phrases to the prompt that describe the desired atmosphere, such as 'Serene capturing the essence of a quiet personal retreat', to influence the emotional tone of the image.

Highlights

Prompting something great in Midjourney v6 can be incredibly hard, despite improved prompt coherence.

Version 6 of Midjourney requires a new prompting framework that is more specific and detailed.

The process of achieving a desired image involves starting with a basic prompt and iteratively adding details.

The example demonstrates replicating a studio photo by adding specific elements like background color, clothing, and positioning.

Midjourney v6 has shown issues with rendering hands and feet accurately compared to v5.

Adding details such as 'Caucasian male' and 'studio photo' to the prompt helps in achieving a closer match to the reference image.

The importance of specifying the overall atmosphere and vibe of the image for a polished result.

An iterative approach with specific language allows for producing surprisingly specific results in Midjourney v6.

The need for a rich vocabulary and imagination when creating detailed prompts for Midjourney v6.

A new sub-module is being developed for mastering the prompting techniques of Midjourney v6.

A more complex example involves creating an image of a woman on a balcony with specific architectural and environmental details.

The stylization level in Midjourney v6 can significantly impact the output, making images more subject-centric.

Despite improvements, Midjourney v6 still encounters challenges with rendering certain body parts like hands and feet.

The use of 'extreme detail' in prompts is a new style of prompting made possible in Midjourney v6.

The video provides a course and additional resources for mastering Midjourney v6.

Viewers are encouraged to share their experiences and challenges with Midjourney v6 in the comments.

The video concludes with a demonstration of how to achieve extremely specific outcomes using detailed prompts in Midjourney v6.