Consistent Characters Soon | Midjourney Office Hours Recap Feb 28th 2024 | Midjourney News
TLDRIn the Midjourney Office Hours recap from February 28th, 2024, the team is actively working on three major updates: a new character reference feature, a new style reference system, and a new describe function. They are currently testing two versions of the character reference and are excited about the potential for anime characters to appear in real-life photography. The updates are expected to launch soon, with improvements to be made in version 7. The team is also addressing issues with Discord links and image prompting, and they are exploring better body coherence. In addition to these updates, they are considering a random style generator and a style map for improved style exploration. Version 6 is set to become faster with server upgrades, aiming to reduce image generation time significantly. There is also a mention of potential API openings for certain chatbot companies. The social aspect of their website is currently a work in progress, focusing on bug fixes and mobile experience improvements.
Takeaways
- 🎨 The Midjourney team is working on three main features: a new character reference feature, a new style reference system, and a new describe feature, with a potential launch next week.
- 🧐 They are testing two versions of the character reference and are excited about the possibility of integrating anime characters into real-life photography.
- 🔍 The character reference feature may not include small details like outfits or facial features but aims to be more accurate in future versions.
- 🚫 New moderation policies might be introduced, possibly requiring positive imagery for real person images.
- 🔄 The new describe feature is expected to be more accurate and closer to the source image.
- 🔄 The style reference update will improve the style transfer process, although it may initially break the old system.
- 💻 Issues with Discord links and image prompting are acknowledged and being worked on.
- 🤔 Challenges in achieving better body coherence are being addressed, possibly pushing the limits of version 6.
- 🚀 Version 7 is anticipated to be a significant step forward, with no signs of diminishing returns.
- 🔧 A random style generator and style map for better exploration of style space are in the pipeline.
- 🚀 Version 6 is expected to get faster with server upgrades, aiming to reduce image generation time significantly.
- 📱 The website's social features are being improved, with a focus on bug fixes and enhancing the mobile experience.
- 🤖 An API opening for chatbot companies is being explored, although specific companies are not yet disclosed.
Q & A
What are the three main features the Midjourney team is currently working on?
-The Midjourney team is working on a new character reference feature, a new style reference system, and a new describe feature.
What is the current status of the character reference feature?
-They are testing two versions of the character reference feature and are hopeful to start testing the describe feature soon.
What limitations might there be with the character reference feature?
-Small details such as the outfit a character is wearing or small facial features might not work with the character reference feature initially.
What exciting prospect does David, from the Midjourney team, mention about the character reference feature?
-David is excited about the possibility of having anime characters appear in real-life photography through the character reference feature.
How does David describe the process of improving the character reference feature?
-David mentions that the process is a big step forward and that it will probably improve in version 7 as they continue to figure out how to pull out smaller details.
What kind of moderation might be introduced with the new features?
-There might be a new moderation policy where real images of people need to be positive imagery only. It is unclear if this will apply to realistic-looking Midjourney images.
What improvements are expected in the new describe feature?
-The new describe feature is expected to be more accurate and get users closer to the source image.
What issues have been encountered with the style reference update?
-The style reference update will break the old one, but it is expected to be more accurate at transferring the style. There have been some issues with links on Discord related to image prompting.
What is the current status of body coherence in Midjourney's version 6?
-David mentioned that they are trying to improve body coherence, but it has been challenging and they might be pushing the limits of version 6.
What are the upcoming improvements for version 6 in terms of speed?
-Version 6 will likely get faster as they are upgrading their servers, aiming to reduce the generation time from about 35 seconds to 15 or 20 seconds.
What is the current focus regarding the social aspect of the Midjourney website?
-The team is currently working on fixing bugs and improving the mobile experience for the social features on their website.
Is Midjourney considering opening an API for certain companies?
-Yes, they are exploring the possibility of opening an API for some chatbot companies, though they did not specify which ones.
Outlines
📢 Mid-Journey Office Hours Recap and Patreon Support
The video script provides a recap of the Mid-Journey office hours held on February 28th. The narrator emphasizes their effort to keep the channel free from sponsors and invites viewers to support the channel on Patreon for $5 a month, which grants access to multiple prompts weekly. Currently, there are nearly 50 prompts available. The Mid-Journey team is actively working on three main features: a new character reference feature, a new style reference system, and a new describe function, with a potential launch planned for the following week. They are testing two versions of the character reference and are also preparing to test the describe feature soon. The character reference is expected to allow anime characters to appear in real-life photography, although small details like outfits or facial features might not be captured. The team is also considering moderation policies for imagery, possibly requiring positive imagery only. The new describe feature is anticipated to be more accurate in matching the source image, and the style reference update will improve style transfer accuracy. The script also mentions ongoing issues with Discord links and image prompting, as well as efforts to enhance body coherence. The team is excited about the upcoming version 7, which is expected to bring significant improvements without hitting diminishing returns. Additionally, they are considering a random style generator and a style map for better exploration of style options. The video concludes with a mention of server upgrades to speed up generation times and potential plans to open an API for certain bot companies. The narrator also shares the prompt used for the art in the video and invites viewers to follow them on Instagram for more content.
Mindmap
Keywords
💡Midjourney Office Hours
💡Character Reference Feature
💡Style Reference System
💡Describe
💡Moderation
💡Body Coherence
💡Version 7
💡Random Style Generator
💡Style Map
💡API
💡Server Upgrade
Highlights
Midjourney team is working on three main features: a new character reference feature, a new style reference system, and a new describe feature.
All three features are expected to launch next week.
Two versions of character reference are currently being tested.
The character reference feature may allow anime characters to appear in real-life photography.
Small details like outfits or facial features may not be captured in the character reference feature.
David, from Midjourney, is excited about the potential of the new features and anticipates improvements in version 7.
New moderation policies may be introduced, requiring positive imagery for real images of people.
The new describe feature is expected to be more accurate and closer to the source image.
The style reference update will be more accurate but may break the old system.
There have been issues with Discord links and image prompting, which are being addressed.
Midjourney is exploring better body coherence, although it has been challenging.
Version 7 is anticipated to push the limits beyond version 6.
A random style generator similar to SRF random is planned as the next feature after the current three.
A style map for better exploration of the style space is also in the works.
Version 6 will likely become faster as server upgrades are implemented, aiming to reduce generation time.
Social features on the website are being improved, with a focus on bug fixes and mobile experience.
Midjourney is considering opening an API for certain chatbot companies, although specifics are not disclosed.
The art in the video was created using a specific prompt and settings in Midjourney.
The speaker encourages following on Instagram for more art and updates.