Midjourney Office Hours (16th Oct 2024)
A new image editor and multiple personalization will be available soon. Depth References allow users to re-texture everything within the image. More releases are expected over the next six months.
Get a free email alert when a new article is published
Dear readers,
Here is the summary of the latest Midjourney Office Hours:
The top two upcoming features are the new editor, which allows users to edit their images outside of Midjourney, and the multiple and fast personalization features, which would enable users to quickly get up to speed with personalization.
It is unclear whether the new editor will be available to all users or only those on the yearly Pro Plan or higher. The developers limit the release to keep an eye on the moderation.
A new feature called "Depth References" (exact name to be determined) will be available in the editor, allowing users to re-texture everything inside the image while keeping its shape.
David mentioned a new conceptual model that allows users to train the model rather than simply create images.
The developers are currently training 3D and video models. The 3D development is being delayed because the team is on sick leave.
The video model aims to be fast, good, and inexpensive. However, there will be no Relax mode at the start of its release because it is unknown whether there will be enough free computers to support it. However, the developers will look into it again later. Niji is also working on the video.
Midjourney V7 is currently undergoing training and is unlikely to be released soon. It is estimated that 99% of V7 will be different from V6. Multi-character, color, and drawing are all possible features, but they have not been confirmed. A better /describe will be worked on in V7.
David mentioned that Midjourney's strategy is to share what they're working on, with the exception of secret projects, and then release it as soon as possible for users to try.
Other potential releases in the next six months include a storytelling feature and three or more "crazy, weird, and secret" projects.
Some explanations for the existing Midjourney V6.1 features:
Why doesn't the bot follow the prompt?
It's because the bot has to strike a balance between aesthetics and prompt adherence. Images that strictly adhere to the prompt may not have a good aesthetic.
Could a prompt affect the outcome of the Sref random?
No. Sref random will simply bring the bot to a high-dimensional space at random in order to generate an aesthetic. It's not related to the prompt.
What happens if you combine two Sref codes? Does it generate new Sref code behind the scenes?
No. Combining Sref codes will not result in a new code. The bot will simply calculate the average of the two codes to generate the aesthetic style.
Cover prompt
/imagine prompt: a detailed illustration of a man, dog, cat, and a cyborg chasing each other around an endless cycles of computer. This artwork exudes elegance through its line art design, capturing the essence of both realism and technology. --sref 1753827134 --ar 16:9 --v 6.1
Free Geeky Animals' Sref code: 1753827134
This Sref code was used to create this article's cover image.
