SREF RANDOM released for Midjourney. 25 example styles.
TLDRThe video by Th explores the 'Midjourney' style reference feature, showcasing 25 unique styles discovered using the 'sref random' function. Th discusses the recent updates and shares their excitement for finding new styles. They explain how to apply styles to new prompts using the image link after the dash-sref and detail the four different style versions (sv1 to sv4), each offering a distinct look. Th encountered an issue with seeds not working with style random but discovered that specifying a seed number for dh-sref random locks in both seed and style seed numbers. The video presents a variety of styles, from vibrant colors and fantastical landscapes to gothic ambiances and surrealistic elements, each with its own distinct characteristics and color schemes. Th concludes by expressing their preference for exploring the impacts of words on prompts over style referencing for sharing purposes.
Takeaways
- ๐จ The introduction of the 'sref random' feature by Midjourney allows for a wide variety of styles to be generated with each prompt, enhancing the discovery of new and unique styles.
- ๐ To apply a preferred style to new prompts, use the image link followed by a dash and 'sref'.
- ๐ There are four official style reference versions (sv1 to sv4), each offering a slightly different visual outcome.
- ๐ง An issue was encountered where traditional seeds did not work with 'style random', leading to identical images regardless of seed number used.
- ๐ข A 'style seed number' was discovered which, when used with a dash and 'sref random', locks in both the seed and style seed numbers, resulting in the same image each time.
- ๐ The unofficial way to control random testing is by specifying a number for 'dh-sref random', allowing for้ข็ฅ (advance knowledge) of the style before committing to using it.
- ๐ The process of combining styles from different images via links was discussed, although it's noted as unofficial and subject to change.
- ๐ The speaker shares a personal preference for vibrant and rich color palettes, leaning towards fantasy and surrealism in style.
- โจ A variety of styles were showcased, each with a distinct description of its visual characteristics and the emotions it evokes.
- ๐ The number of styles that can be discovered is vast, with 4,294 million,96725 styles available for version 6.0, indicating a wide range for exploration.
- ๐ The speaker expresses a desire to test as many styles as possible but acknowledges the challenge due to the sheer number of options.
- โ๏ธ The video transcript concludes with a note on the preference for exploring the effects of words on prompts rather than relying solely on style references for sharing purposes.
Q & A
What is the latest feature released by Midjourney that the speaker is excited about?
-The latest feature released by Midjourney that the speaker is excited about is the 'sref random' feature.
How does the 'sref random' feature work in Midjourney?
-The 'sref random' feature works by allowing users to add 'D- sref random' to their prompt, resulting in a different style each time the prompt is used.
What is the official way to copy a style that you like in Midjourney?
-The official way to copy a style you like is by specifying the link to the image after the Dash - sref in new prompts.
How many different versions of style reference are mentioned in the script?
-Four different versions of style reference (SV1, SV2, SV3, and SV4) are mentioned in the script.
What does the speaker discover about using seeds with the 'style random' feature?
-The speaker discovers that using the same seed number with the 'style random' feature does not change the style with every prompt, and that there is a style seed number that locks both seed number and style seed number in place.
How does the speaker suggest controlling random testing of styles?
-The speaker suggests controlling random testing by specifying a number for 'dh- S ref' in advance, which allows for the randomization to be done manually.
What did the speaker find challenging about using image links for style referencing?
-The speaker found using image links for style referencing to be quite tedious for sharing purposes.
What is the speaker's approach to sharing styles with the audience?
-The speaker prefers to explore the impacts of words on prompts instead of using srf for Mid Journey, as it is less tedious for sharing purposes.
How many styles did the speaker discover with the random feature?
-The speaker discovered 25 styles with the random feature.
What is the speaker's impression of the styles found in Midjourney version 6.0?
-The speaker is impressed with the styles found in Midjourney version 6.0, finding many of them to be really cool and diverse.
What is the speaker's intention with the experiment involving different sref versions?
-The speaker's intention with the experiment is to show the same prompt in four image grids but with versions ranging from 1 to 4, to observe the differences between the style versions.
Outlines
๐จ Exploring Mid Journey's Style Reference and Random Feature
The speaker, an AI art enthusiast, introduces the topic of Mid Journey's style reference feature and its recent updates, particularly the 'sref random' feature. This feature allows users to discover new styles by adding 'D- sref random' to their prompt, resulting in different styles each time. The speaker explains how to apply a preferred style to new prompts using an image link and the Dash - sref command. Four different style versions are mentioned, each offering a unique look: sv1 (original), sv2 (follow-up), sv3 (more vibrant), and sv4 (current default with improved performance). The speaker also discusses an unofficial way to control style randomness using a style seed number and the potential to combine style numbers for further customization. The paragraph ends with the speaker's intent to share discovered styles and seek feedback on how to present them.
๐ A Spectrum of Styles: From Vibrant to Gothic Ambiance
This paragraph delves into a variety of artistic styles discovered using the 'sref random' feature. Each style is characterized by its use of color, light, and technique, ranging from vibrant and energetic with rich color schemes to more subdued and ethereal styles with soft transitions and muted tones. The speaker describes styles that evoke fantasy, surrealism, modern interpretations of classic techniques, and futuristic motifs. The paragraph highlights the diversity of styles, from those with a monochromatic palette and strategic use of light and shadow to others that blend digital artistry with classical aesthetics. The speaker also mentions an experiment with different style reference versions and provides a glimpse of the styles to be showcased in the video.
๐๏ธ Artistic Techniques and Color Palettes in Mid Journey Styles
The paragraph continues to describe the intricate details and techniques associated with the discovered styles. It covers a range of styles from expressionism with aggressive brushwork to impressionism with modern twists. Each style is defined by its color palette, use of light and shadow, and the emotional impact it conveys. The speaker touches on styles that blend historical art movements, Gothic and surreal elements, and digital painting techniques. The paragraph emphasizes the contrast between light and dark, the use of vibrant and subdued hues, and the interplay between the real and the imagined in these styles. It concludes with a note on the speaker's preference for exploring the impacts of words on prompts rather than style referencing for sharing purposes.
๐ Reflecting on Discovered Styles and Future Exploration
In the final paragraph, the speaker reflects on the styles discovered and shares personal observations. They note a significant improvement in the variety and usability of styles from Mid Journey version 5.2 to version 6.0, where many cool styles were easily found. The speaker expresses gratitude to the viewers and encourages continued exploration and prompting in the world of AI art. The paragraph concludes with a call to action for viewers to engage and share their preferences on how styles should be presented in future videos.
Mindmap
Keywords
Midjourney
Style Reference (sref)
Prompt
Random Feature
Seed Number
Style Seed Number
SV (Style Version)
Viby
Impressionism
Chiaroscuro
Digital Painting
Highlights
Introduction of the SREF RANDOM feature by Midjourney, a favorite new release for discovering new styles.
The ability to add 'D- sref random' in prompts for varied style outputs with each prompt.
Official method to copy a liked style by specifying the image link after the Dash - sref.
Four different style reference versions (SV1 to SV4) each offering a slightly different look.
SV1 being the original style reference algorithm, SV2 its successor, SV3 more 'viby', and SV4 the current default with improved performance.
Unofficial way to reference styles using a style seed number instead of an image link.
Incompatibility of style random with traditional seeds, as they lock both seed and style seed numbers.
Finding a workaround by specifying a seed number for dh- sref random to control the style.
The discovery of 4.294 million, 96725 styles to explore in version 6.0 of Midjourney.
Personal preference for majority of the styles found using the srf random feature.
Experimenting with different s ref versions on the same prompt to compare styles.
Vibrant colors, fantastical landscapes, and celestial wonders characterize the mesmerizing style with rich and intense hues.
A modern interpretation of classic techniques with rich and varied color schemes applied for texture and depth.
Fusion of vibrant contrasts and futuristic motifs with neon hues reminiscent of urban nightlife.
Immersion into an alternate reality with a monochromatic palette and strategic use of light and shadow.
Gothic ambiance with a chiaroscuro technique, emphasizing drama and intensity with a restrained color palette.
Vivid and eclectic color blending with surrealism and futurism elements offering depth and dimension.
Fusion of fantasy and realism with vivid and intense use of color creating a dreamlike quality and golden tones.
Concluding thoughts on the exploration of styles and the preference for sharing styles using numbers instead of image links.