Web App Launch
The web app is launched with features such as a gallery component, and the ability to view and track users' own image creations. The web app can be accessed at app.bluewillow.ai, with more features coming soon.
The AI Edge Podcast
Last week, we started AI Edge podcast, a weekly live event on Discord that will also be published on podcast platforms. The podcast will include a mix of prompt-guiding sessions and interviews with AI guests. Listeners can catch the podcast live or later on over 20 platforms, including Apple Podcasts, Google Podcasts, and Spotify.
The V3 improvements include the deployment of prompt improvement tools and the upcoming release of tools to help users with short prompts. We encourage everyone to test it out and provide feedback on the improvements.
Blue Willow Blog
We also have a Blue Willow Blog, which is available on the company's official website, bluewillow.ai. The blog contains useful information and updates about the company.
Pay What You Can Model
Soon we will release a Pay What You Can model, allowing users to contribute and support the platform and its mission of staying free for everyone. We expect to launch it this week, stay tuned for more information!
Q: Will V4 bring back artist styles that were in V2?
A: We do understand that some people liked the artist styles in V2. We are trying to incorporate more features with every new model generation, but it's difficult to satisfy every factor. We are making every effort to ensure compatibility.
Q: Can users get access to a V2.5, and is there a way to get V2 style variations from the early V3 release?
A: There wasn't a V2.5, just some minor updates to V2. Users can still try V2 if they prefer its output. To access V2 compatibility, users should use the command "--v 2" (with a space between "v" and "2"). We plan to improve compatibility and reduce confusion soon.
Q: Will there ever be a function to merge existing images?
A: We acknowledge the request and plan to add more functionality to the model. However, our top priority is to increase the model's general quality and ensure basic functionality works well. We are a small team with limited resources, so it's difficult to prioritize every feature request.
Q: Will there be an option to delete images?
A: We are working on this, but we suggest users create their own server and add the bot there for more control over image deletion.
Q: Are there plans to add weights for prompts?
A: We would love to add this feature, but it's challenging to maintain compatibility between models with more constraints. We are aware of the request, but it's hard to predict the timeline for its implementation.
Q: Will it be possible to hide prompts in the upcoming gallery for privacy?
A: We suggest users create their own server and add the bot there for better privacy.
Q: When will the gallery be fixed for users experiencing issues?
A: We suggest completely closing the browser and logging back in as a temporary solution. The team is working on fixing the overall problem.
Q: Is it possible to change the command prefix for the BlueWillowAI bot?
A: It cannot be changed on Discord due to the platform's limitations.
Q: How long has the team been working on the Blue Willow project?
A: The team has been working on the project for about four months, with some additional time spent on it before the project went live on Discord.
Q: Will there be more aspect ratios available when creating images?
A: Yes, the team is actively working on providing more aspect ratio options and ensuring image quality is consistent across different ratios.
Q: Will the team fix V2 duplication or elongation issues?
A: The team is currently not actively improving V2 because we are focused on new models. However, we will keep compatibility with V2.
Q: Are there plans to improve NSFW censoring?
A: The team is continuously working on improving NSFW censoring and sees it as both a technical and community challenge. We appreciate the help from moderators and encourage users to report inappropriate content.
Q: Is it possible that Blue Willow AI could degrade output quality due to bad training and incorrect user preferences?
A: This is a recognized risk, and the team is aware of it. However, we believe that with more users participating, even with some bad actors, reasonable ratings can still be achieved. We are also careful not to overfit the model on user ratings to maintain the model's generative capabilities.
Q: Is there a development road map for the company?
A: There isn't a comprehensive road map, as the field is constantly changing with new techniques being developed every day. The team's priority is to get new features out as soon as possible.
Q: How seriously does the company take members feedback?
A: The company takes member feedback very seriously. The team goes through the feedback channel at least once a week to consider suggestions for features and improvements.
Q: Will there be high-resolution font images?
A: Yes, high-resolution font images are a feature request seen from users, and it is on their timeline for incorporation.
Q: What model should users try for creating cyberpunk content?
A: Users should try using the dash-based V2 model if they are having trouble with the current model. The team is working on improving this aspect.
Q: How can someone become a BlueWillow mod?
A: To become a BlueWillow mod, users can contact Tree Admin directly, who will send them a form to fill out. Alternatively, users can contact any of the mods for assistance. Please fill out our Volunteer Discord Mod & Other Jobs Application Here: https://form.jotform.com/form/230618636259058