Midjourney v6 Tips and Tricks. Text. Chaos. Weird. Stylize. Part 2
TLDRIn this video, the creator explores Midjourney v6's new features, highlighting the improved text-to-image capability. They discuss the use of quotation marks for text input and the parameters for chaos, weird, and stylize, providing insights on their effects. The video showcases how these features can enhance image generation, create logos, and even new fonts, while noting the necessity for trial and error.
Takeaways
- 📌 Midjourney v6 introduces significant improvements in generating text within images, which was a challenge in previous versions.
- 🔍 To include text in images, enclose the desired text within quotation marks in the prompt; v6 recognizes them differently from previous versions.
- 🎯 The accuracy of text generation is not 100%, but it yields better results than before, especially with more common words or phrases.
- ⚙️ The chaos parameter (--chaos or -c) adds variety to the outputs, with higher values leading to more extreme variations.
- 🌀 With higher chaos values, the outputs may stray from the original prompt, so using lower values is recommended for more predictable results.
- 🤔 The weird parameter (--weird or -w) controls the unusualness of the images, with values ranging from 0 to 3000，默认值为0意味着该参数通常不被使用，除非在提示中特别提及。
- 🎨 The stylize parameter (--stylize or -s) adjusts the application of Midjourney's default style to the images, with a range from 0 to 1000，默认值为100。
- ✨ The stylize parameter makes images more aesthetically pleasing, in contrast to the weird and chaos parameters which can make them appear uglier.
- 🔎 The default and raw styles in v6 seem more connected, as overlaying images from both styles on the same seed number reveals similar compositions.
- 🔄 Other features available in v6 include Das Dash, tile, and image blending, which may be explored in future videos for their utility and differences.
- 🙏 The video creator encourages viewers to share their experiences with chaos and weird parameters, and whether they find them useful in their own experimentations.
Q & A
What is the main focus of the video?
-The main focus of the video is to discuss the features of Mid Journey's latest version 6, particularly the improvements in generating text within images and other features like chaos, weird, and stylize.
How has the text generation in images improved in version 6?
-In version 6, text generation in images has improved significantly. It is now possible to get text to appear in the images by placing the desired text in quotation marks in the prompt. Although not 100% accurate, it can produce good results occasionally.
What are some challenges with generating text in images using version 6?
-Some challenges include difficulties with more obscure words or phrases, as the text may not be recognized correctly. Additionally, longer sentences may be harder to get right due to the higher chance of words getting scrambled.
How can users direct the appearance of text in the generated images?
-Users can direct the appearance of text by specifying how the text has been created, such as 'typed', 'written', or other phrases. They can also direct the medium where the text appears, like on a sign, billboard, paper, or post-it.
What is the chaos feature and how does it affect the outputs?
-The chaos feature, controlled with '-- chaos' or '--c' followed by a value from 0 to 100, introduces variety in the outputs. Higher values result in more diverse and unexpected options, but may also stray from the prompt more significantly.
What are the differences between the 'weird' and 'chaos' parameters?
-The 'weird' parameter, controlled with '-- weird' or '--w', accepts values from 0 to 3,000 and controls how unusual the images will look. It seems to have subtler effects compared to the 'chaos' parameter, which introduces more significant changes to the outputs.
How does the 'stylize' feature affect the images?
-The 'stylize' feature, controlled with '-- stylize' or '--s', ranges from 0 to 1,000. It controls how heavily Mid Journey's default style is applied to the images. Higher values add significant beautification, while lower values result in less appealing images.
What are some additional features available in version 6?
-Additional features in version 6 include '-- dash', 'tile', and 'image blending'. The 'tile' feature creates seamless edges for images, while 'image blending' is supposed to work better in this version, although it hasn't been extensively tested by the speaker.
How can users get better results with the new text feature in version 6?
-Users can improve results by using more common words, keeping sentences short and simple, and using the '-- style Raw' option if the text isn't showing up correctly.
What is the speaker's recommendation for using the chaos and weird parameters?
-The speaker recommends staying within lower values for the chaos and weird parameters if the goal is to see things that are still recognizable and aligned with the prompt.
How does the default style and raw style compare in version 6?
-In version 6, the default style and raw style seem more connected, with the raw style tending towards photo realism. When using the same seed number, the compositions of the default and raw images are very similar.
Outlines
🖼️ Introduction to Mid Journey's Version 6 and Text-to-Image Features
This paragraph introduces the viewer to Mid Journey's latest version 6, focusing on its new text-to-image capabilities. The speaker explains that while previously it was challenging to get text to appear correctly in images, version 6 has significantly improved this feature, although it is not yet perfect. The viewer is directed to a part one video of the series for basic features and the speaker shares observations on the new text feature, noting that more obscure words or phrases may be difficult to render accurately. The speaker also discusses various syntax elements in version 6, such as quotation marks for text, curly braces for permutations, and double colon for multi-prompting. The paragraph concludes with the speaker's thoughts on the potential for creating logos and fonts with the improved text feature, despite acknowledging other AI tools may offer similar capabilities more efficiently.
🎨 Exploring Chaos, Weirdness, and Stylization Parameters in Version 6
The second paragraph delves into the exploration of various features in Mid Journey's version 6, specifically focusing on the chaos, weirdness, and stylization parameters. The speaker describes the chaos feature, which introduces variety in the outputs, and shares personal experiences with different chaos values, noting that higher values can stray from the prompt. The weirdness parameter is then discussed, with the speaker using the prompt 'chair carved from sandstone' to demonstrate how varying values affect the image's appearance. The speaker finds the weirdness parameter still useful for creating unusual images, albeit with subtle differences between values. Lastly, the stylization feature is examined, which controls the application of Mid Journey's default style to images. The speaker's tests with different stylization values reveal that higher values significantly beautify images, and a comparison between default and raw styles suggests a closer connection in version 6. The paragraph ends with the speaker's reflections on the observed similarities between default and raw styles and hints at potential future content on additional features like Dash, Tile, and image blending.
Mindmap
Keywords
💡Mid Journey's version 6
💡Text in images
💡Quotation marks
💡Chaos feature
💡Weird parameter
💡Stylize feature
💡Seed number
💡Raw style
💡Parameter syntax
💡Multi-prompting
💡Permutations
Highlights
Mid Journey's latest version 6 introduces a new feature that allows text to appear in images, a significant improvement from previous versions.
To include text in images, place the desired text within quotation marks in your prompt; version 6 recognizes them differently from previous versions.
While the text feature is not 100% accurate, it can occasionally produce good results, especially with more common words and phrases.
Obscure words or phrases may be challenging for the AI to generate accurately in images, as they are not commonly written.
Longer sentences may be difficult to render correctly due to the increased chance of letters being scrambled.
Single words are more likely to turn out correctly, as they are simpler and less prone to errors.
The '--style Raw' command may help if the text is not showing up correctly in the image.
With the new text feature, it's possible to create better logos or even entire new fonts, although it requires a lot of trial and error.
The '--chaos' parameter introduces variety in the outputs, with higher values leading to more unexpected results.
The '--weird' parameter controls the unusualness of the images, with higher values resulting in more unconventional visuals.
The 'stylize' feature controls the application of Mid Journey's default style to the images, with higher values adding more beautification.
There seems to be a closer connection between the default and raw styles in version 6, as they share similar seed spaces.
The 'tile' feature allows for seamless edges on images, making them connect to each other without visible breaks.
The 'image blending' feature may have been improved in version 6, although it requires further experimentation to confirm its effectiveness.
Version 6 of Mid Journey is a significant leap forward, especially with the introduction of text in images, despite the need for trial and error.
The video provides a detailed exploration of the new features in Mid Journey version 6, offering insights and recommendations for users.
The presenter encourages viewers to share their experiences with the chaos and weird parameters, indicating an open-minded approach to exploring the AI's capabilities.