【AIツール】Midjourney - ミッドジャーニーで写真を元に画像生成する方法。
TLDRThe video script introduces an AI tool called Midjourney, which generates images based on specific keywords and reference photos. It explains the process of using Midjourney, including creating a Discord account, joining the Midjourney server, and executing commands to generate images. The script emphasizes the importance of using 'spells' or commands to influence the output and offers tips for achieving higher quality results, such as using multiple reference images and specific keywords. It also discusses the free and premium plans available for Midjourney, highlighting the commercial use of generated images with premium subscriptions.
Takeaways
- 🌟 Introduction to AI tool Midjourney, which generates images based on specific keywords and can also incorporate reference photos for more accurate results.
- 📸 Midjourney can be used by creating a Discord account, receiving an invitation from Midjourney, joining a room, and executing commands to generate images.
- 💡 The free plan allows for up to 25 image generations, with paid plans offering more generations and commercial use of the images.
- 🛠️ The quality of generated images can be influenced by the 'spells' or commands used, and trial and error can help refine the process.
- 🌐 Reference images can be uploaded and used to guide the AI in creating images closer to the desired outcome.
- 🎨 Background color of reference images can affect the final result, so it's important to consider this aspect during the image generation process.
- 🔄 The generated images can be further refined using various buttons provided after the initial generation, which offer options like high-quality enhancement and re-generation.
- 📌 When using Midjourney, it's beneficial to include specific details about the reference image, such as style or medium, to improve the quality of the generated image.
- 🔍 Looking at other people's posts and using their keywords can provide inspiration and potentially yield better results.
- 🔧 Parameters like aspect ratio and exclusion of certain keywords can be specified to have more control over the output.
- 🌿 The presenter experimented with combining multiple reference images, such as a portrait and a plant photo, to create a composite image.
Q & A
What is the AI tool introduced in the script?
-The AI tool introduced in the script is called Midjourney, which generates images based on specific keywords or text descriptions.
What is the limitation of generating images with text descriptions alone?
-The limitation of generating images with text descriptions alone is that it can be quite challenging to fully express one's imagined image, often resulting in generated images that differ from the intended concept.
How does Midjourney allow users to generate images closer to their imagination?
-Midjourney allows users to generate images closer to their imagination by sending reference photo data along with the text, which helps to create images that more closely align with the user's vision.
What do users call the commands used for image generation in Midjourney?
-Users refer to the commands used for image generation in Midjourney as 'incantations' or 'spells'.
What is the basic process of using Midjourney?
-The basic process of using Midjourney involves creating a Discord account, receiving an invitation from Midjourney, joining the room, executing commands, and generating images.
What are the pricing plans for Midjourney?
-Midjourney offers a free plan that allows for up to 25 image generations. For more than that, users need to subscribe to a paid plan. The cheapest paid plan is the Basic Plan at $10 per month, which allows for up to 200 image generations.
What are the commercial usage rights for images generated by Midjourney?
-Images generated by Midjourney are not allowed for commercial use by default. However, if users subscribe to a paid plan, they gain the rights to use the images commercially.
How does the background color of a reference image affect the generation process?
-The background color of a reference image can significantly impact the generation results. For example, a transparent background (PING) versus a white background (JPEG) can lead to different outcomes in the generated images.
What are some tips for improving the quality of generated images?
-To improve the quality of generated images, users can specify the style of the reference image (e.g., anime style), use other people's successful keywords, and experiment with various 'incantations' or parameters to fine-tune the results.
How can users refine the generated images?
-Users can refine the generated images by using buttons below the image, such as U1 to UFO for high-quality upscaling, V1 to V4 for regenerating the image based on the same incantation, and the recycle mark to再生 the image with the same incantation. Note that each action consumes a generation count.
What parameters can be used to modify the image generation process?
-Parameters such as -AR for aspect ratio adjustment, - for excluding specific keywords, and including certain keywords or words like 'high quality' or 'beautiful' can be used to modify the image generation process and achieve desired results.
How can using multiple reference images enhance the generation process?
-Using multiple reference images can enhance the generation process by providing more detailed visual cues to the AI, resulting in a more accurate and richly detailed final image that reflects the combined elements of all reference images.
Outlines
🖼️ Introduction to AI Image Generation with Midjourney
This paragraph introduces the concept of using AI tools, specifically Midjourney, to generate images based on specific keywords and reference photos. It explains that while Midjourney typically generates images from text keywords alone, there are limitations to expressing detailed personal visions with text. The speaker aims to demonstrate how to generate images closer to one's own imagination by combining text and reference photo data. The paragraph also touches on the 'incantations' or commands used in image generation and how they can significantly alter the resulting image. The speaker shares their experience with trial and error in finding the right 'incantations' to create images that closely match their vision.
📸 Preparing for Image Generation with Reference Photos
The speaker discusses the process of preparing reference images for generating images with Midjourney. They mention the importance of choosing appropriate images, such as a sample photo they took for a thumbnail, and the impact of background color on the generation results. The speaker emphasizes the need to use images that are suitable for public disclosure. The paragraph outlines the steps to upload the reference image to Discord and how to obtain its URL for use in the image generation process.
🛠️ Using Commands and Parameters for Image Generation
In this paragraph, the speaker dives into the specifics of using commands, or 'incantations,' to generate images with Midjourney. They explain the basic workflow of joining a Midjourney room on Discord, executing commands, and waiting for the image to be generated. The speaker also discusses the different plans available for using Midjourney, including a free plan and paid plans that allow for more image generation and commercial use of the generated images. Additionally, the paragraph covers various parameters that can be used to modify the aspect ratio and exclude specific keywords, as well as the potential impact of including certain keywords like 'High Quality' or 'Beautiful' to enhance the image quality.
🌿 Combining Multiple Photos for Enhanced Image Generation
The speaker explores the possibility of using multiple reference photos to generate an image, aiming to create a more complex and detailed result. They demonstrate this by combining a previous photo with a plant photo and adding relevant keywords to the 'incantation.' The resulting image reflects the combination of the two photos, with noticeable changes in facial expressions and a slightly more 'foreign' appearance. The paragraph concludes with the speaker's reflection on the simplicity and power of using AI tools like Midjourney to generate high-quality images and their anticipation for a future where AI becomes a common tool in various tasks.
Mindmap
Keywords
💡AI工具
💡ミッドジャーニー (Midjourney)
💡画像生成
💡キーワード (Keyword)
💡参考写真 (Reference Photo)
💡コマンド (Command)
💡Discord
💡無料プラン (Free Plan)
💡有料プラン (Paid Plan)
💡商用利用 (Commercial Use)
💡呪文 (Spell)
💡パラメーター (Parameter)
Highlights
Introduction to AI tool, Midjourney, which generates images based on specific keywords.
Midjourney typically generates images from text keywords, but sometimes the generated images may not match the imagined image due to the limitations of text expression.
To generate images closer to one's own image, the method of sending reference photo data along with text is introduced.
The use of 'commands' or 'incantations' in image generation can significantly change the resulting image.
A brief overview of how to use Midjourney, including creating a Discord account, joining the Midjourney server, and executing commands.
Information about the free and paid plans of Midjourney, including the number of images that can be generated and the cost of the Basic Plan.
Commercial use of images generated by Midjourney is not allowed in the free plan, but it is permitted in paid plans.
Preparation of reference images, including the selection of a suitable image and the impact of background color on the generation result.
Uploading the reference image to Discord and obtaining its URL for use in Midjourney.
Entering the /imagine command followed by the image URL and keywords in the Midjourney input form to generate an image.
Explanation of the waiting time for image generation and the approximate time it takes.
The process of reviewing the generated image and the options available for high-quality, re-generation, and recycling with the same incantation.
Points to consider when selecting keywords, such as specifying the style or type of image desired, and using references from other people's posts.
Introduction to parameters that can be used during image generation, such as -AR for aspect ratio and - for excluding specific keywords.
The impact of including words like 'high quality' or 'beautiful' in the keywords to potentially increase the level of the generated image.
Demonstration of generating an image using multiple reference photos and the resulting combination of elements from the photos.
Reflection on the ease of generating high-quality images with AI and the potential for AI to become a common tool in the near future.
Encouragement for those who have never used AI to try Midjourney and experience the capabilities of AI.