Mastering Midjourney v6 : A Deep Dive into the New Version!
TLDRMidjourney Version 6, a significant update from 5.2, has been released with enhanced aesthetics, prompt understanding, and new features, albeit in Beta. The update requires relearning prompting techniques, as V6 is sensitive to keywords and avoids 'junk' terms. The video demonstrates the new model's attention to detail, especially in hair and skin textures, and its dynamic composition. Creative upscaling introduces more detail, while text prompts within images are now supported. Despite some commands not functioning yet, V6 offers powerful features like 'very subtle' and 'very strong' variations. The speaker, Tim, encourages viewers to experiment with prompts and stay updated with Midjourney's developments.
Takeaways
- ๐ **Significant Update**: Midjourney version 6 is a notable upgrade from version 5.2, especially in aesthetics and prompt understanding.
- ๐จ **New Features**: Version 6 introduces new features, but as a Beta release, some features are still missing.
- ๐ **Prompting Changes**: Prompting with V6 is different and requires relearning; it is sensitive to the prompt and prefers explicit instructions.
- ๐ซ **Avoid Junk Keywords**: V6 advises against using generic descriptors like 'award-winning' and instead focuses on clear, specific prompts.
- ๐ **Less Opinionated Option**: For a less stylized image, use the '--raw' option.
- ๐ **Creative Prompts**: Even simple prompts like 'toast' can yield interesting results, showcasing the model's versatility.
- ๐ **Image Comparisons**: V6 pays close attention to keywords, as demonstrated by the difference between 'Northeast' and 'Northwest' in image outputs.
- ๐ **Detail Improvement**: V6 is particularly good at rendering detailed elements like skin, eyes, and hair, even down to small details like raindrops.
- ๐ **Text Prompting**: A new feature allows for text to appear in images when enclosed in quotation marks.
- ๐ **Creative Upscaling**: V6 introduces a creative upscaling option that enhances image details but can sometimes alter the original scene's context.
- ๐ ๏ธ **Commands in V6**: Some commands are not yet functional in V6, but 'very subtle' and 'very strong' have become more powerful for image variations.
- ๐ **Resources Available**: Images and a PDF worksheet from the video are available for free on Gumroad, with donations appreciated.
Q & A
What is the significance of Midjourney version 6 release?
-Midjourney version 6 is a significant update over version 5.2, offering improvements in aesthetics, prompt understanding, and new features, although it is a Beta release and some features are missing.
How can one switch to using Midjourney version 6?
-To use version 6, issue the command 'SL settings' and choose 'Mid Journey model V6'. Alternatively, for experimentation, add 'D-V6' to the end of any prompt.
What is the major change with prompting in version 6?
-Prompting with version 6 is significantly different from V5, requiring users to relearn how to prompt. The model is more sensitive to the prompt and prefers explicit instructions over generic keywords.
What is the recommended approach to prompting with version 6?
-Avoid using generic or 'junk' keywords. Be explicit about the desired outcome. However, there is no right or wrong way to prompt; experimentation is encouraged.
How does version 6 handle text in images?
-Version 6 can prompt for text within images by encapsulating the text within quotation marks. It works similarly to other AI image generators but with some improvements.
What are the two features introduced in version 6?
-The two major features introduced in version 6 are 'creative upscaling' and the ability to prompt for text within images.
How does creative upscaling in version 6 differ from regular upscaling?
-Creative upscaling not only increases the resolution but also enhances details and adds creative elements to the image, making it more dynamic and detailed.
What are the limitations of the creative upscaling feature?
-Creative upscaling can sometimes alter the original scene's details, such as reducing the extent of destruction in a scene to make it more intact.
How can one get the images and PDF worksheet from the video?
-All images and the PDF worksheet are available on Gumroad, where they can be accessed for free, with donations appreciated.
Which commands are currently unsupported in version 6?
-Commands that are not yet working in version 6 include pan, zoom, very region, in painting, tuning, and describe (which is expected to be updated to V6 soon).
What are the tips for getting better results with text prompts in version 6?
-Encapsulating the text within quotation marks and using a weight of two or three, or using all caps for the text can improve the chances of getting better results with text prompts.
What does the speaker suggest doing with images that have undergone creative upscaling?
-The speaker suggests using an image editor like Photoshop to blend the original and creatively upscaled images, enhancing details by adjusting opacities and layers.
Outlines
๐ Introduction to Mid Journey Version 6
The video script introduces the release of Mid Journey version 6, emphasizing its significant update over version 5.2 in terms of aesthetics and prompt understanding. It mentions that this is a Beta release with some missing features. The speaker plans to dive deep into version 6, discussing prompting, image comparisons, and providing tips and tricks. The audience is instructed on how to run version 6 and the importance of relearning how to prompt for this new version, which is more sensitive to keywords and requires a more explicit description of the desired outcome. The video also touches on the use of the '--raw' option for a less stylized version and shares the speaker's standard prompt formula. It concludes with a comparison of image quality between different versions of Mid Journey, highlighting the improved detail and realism in version 6.
๐ Exploring Version 6's Features and Image Quality
The second paragraph delves into the specifics of version 6, comparing it with previous versions like V4, which is described as weird, gritty, and grimy. It showcases the channel's recurring character, the cyberpunk woman with white hair, in different versions, noting the improved sharpness, pose, attitude, and lifelike hair in V6. The speaker also discusses a new feature in version 6 called 'creative upscaling,' which allows for more detailed and exciting images upon upscaling. The paragraph also addresses the ability to prompt for text within images, providing examples and tips for better results. Lastly, the speaker mentions that some commands are not yet functional in version 6, but others like 'very subtle' and 'very strong' have become more powerful.
๐ Creative Upscaling and Text Prompting Techniques
The final paragraph focuses on the creative upscaling feature, demonstrating its potential with an example transformation of a toast image into a dramatic Godzilla scene. It also discusses the double-edged nature of creative upscaling, where it may alter the original destruction theme in favor of more detailed characters. The speaker shares a technique to combine the subtle and creative upscales in an image editor to achieve the desired effect. Additionally, the paragraph covers the process of prompting for text within images, suggesting the use of quotation marks and weight commands to improve the text appearance. Lastly, it mentions the limitations of unsupported commands in version 6 and the anticipation for a new 'describe' function based on V6.
Mindmap
Keywords
๐กMid Journey version 6
๐กAesthetics and prompt understanding
๐กBeta release
๐กPrompting
๐กCreative upscaling
๐กText in Mid Journey
๐กCommands
๐กImage comparisons
๐กTips and tricks
๐กVersion 6 describe function
๐กUnsupported features
Highlights
Midjourney version 6 has been released with significant updates over version 5.2 in aesthetics and prompt understanding.
This is a Beta release and lacks some features, but offers new features and improved understanding of prompts.
To run version 6, use the command 'SL settings' and choose 'Mid Journey model V6' or add 'D-V6' to the end of any prompt.
Prompting with V6 is significantly different and requires relearning; it is more sensitive to the prompt's keywords.
Avoid using 'junk keywords'; be explicit about what you want for better results with V6.
The new version offers a less stylized option with the '--raw' command for a more neutral output.
Experimentation with different prompt styles is encouraged as the new model responds differently to various inputs.
V6 pays close attention to keywords, as demonstrated by the change in atmosphere when switching 'Northeast' to 'Northwest'.
The character and background details in V6 are more dynamic and have more personality compared to V5.2.
V6 introduces 'creative upscaling', which enhances image details upon upscaling.
Creative upscaling can be a double-edged sword, sometimes prioritizing certain details over others.
V6 now supports text prompts within images, improving on previous versions' capabilities.
There are new tricks for text prompts, such as using quotation marks and specific weights for better results.
Some commands are not yet functional in V6, but others like 'aspect ratio' and 'chaos' are supported.
The 'very subtle' and 'very strong' commands have become more powerful in V6, offering variations close to in-painting.
V6's 'describe' function, based on the V5 model, is still operational, with an updated V6 version forthcoming.
All images and a PDF worksheet from the video are available for free on Gumroad, with donations appreciated.
The speaker, Tim, encourages viewers to experiment with V6 and provides tips and tricks for better image generation.