Video summary

Advanced Midjourney Features — AI Art Sellers Masterclass

Main summary

Key takeaways

Art and Creativity

Summary

The video demonstrates advanced Midjourney tools for controlling image generation, editing prompts, combining references, and producing variations at scale. It distinguishes features that make outputs more precise from those that create more diverse or unusual results.

Advanced features and techniques

  • Text in images: Put the desired wording in quotation marks in the prompt. This can help generate logos, graphic overlays, and information cards, though results may still contain small errors.
  • --style raw: Reduces Midjourney’s default aesthetic and beautification, making results reflect the prompt more directly. It can help with icons or specific styles but requires careful prompting. It can also be enabled in /settings.
  • --no: Explicitly excludes subjects from an image. Multiple exclusions can be entered as a comma-separated list.
  • Multi-prompting with ::: Splits a prompt into separate concepts and allows them to be weighted. For example, space::2 ship gives “space” twice the weight of “ship.”
  • Remix mode: Enable it in /settings to edit a prompt or its parameters after generating an image, including the subject, composition, medium, aspect ratio, and exclusions. The presenter says he generally leaves this mode off.
  • /shorten: Offers four shortened versions of a prompt, helping users see which words Midjourney considers important.
  • /blend: Combines two to five uploaded images into one result. An output aspect ratio can be specified, but there is otherwise little control over the result.
  • Image prompts: Upload an image and place it at the beginning of an /imagine prompt to guide the generation. The presenter advises caution with copyrighted images and suggests using stock or non-copyrighted images.
  • /describe: Analyzes an uploaded image and produces four prompts that might recreate it. The presenter warns about copyright concerns and suggests using public-domain artwork as a safer reference.
  • --tile: Creates a tile intended to repeat seamlessly. Repeating the tile in a grid can produce patterns for wallpaper, fabric, or prints. Crop the assembled result to the desired aspect ratio.
  • Permutations: Put alternatives in braces in a prompt to generate multiple versions from one template—for example, by swapping subjects or scene descriptions. This is useful for generating designs in bulk.
  • --chaos or --c: Controls how varied the results are, from 0 to 100. Higher values produce less predictable images and can make them depart further from the prompt.
  • --weird or --w: Controls how unusual the results are, from 0 to 1000. Unlike chaos, which changes the diversity among generated results, weirdness pushes images toward more unusual outcomes.

The video encourages experimenting with these tools while using judgment, especially when working with image references and copyright-sensitive material.

Creator/Presenter: Alec Phelps.

Rate this summary

Your feedback will help improve summaries.

Improve this summary

Reprocess with a stronger model when the summary feels incomplete or inaccurate.

Pro

Translate summary in another language

Pro

Ask questions to this video

Chat for follow-up questions, clarifications, and source-backed answers.

Coming soon

Share this summary

Original video