Invoke - 4.0 Open Source Release - New Model Manager Interface, Gradient Denoising & More...

Invoke
2 Apr 202406:43

TLDRInvoke 4.0 has been released, marking the beginning of a new era for the platform. The update introduces a new model manager interface, enhancing the user experience with a solid foundation for future evolution. It supports new model architectures and provides a streamlined way to manage and customize models. Users can now add models via local path or URL, and Invoke has a conversion cache to store the most recent 20 GB of models in their converted state for faster reuse. The new model manager allows for multiple simultaneous installations without blocking the UI. Additionally, it features a trigger phrases feature for main and concept models, making it easier to reuse common prompts. The unified canvas now includes a gradient denoising feature that reduces generation time and seamlessly integrates inpainting results. The release notes provide a full list of updates, and the partnership with Nvidia for the AI workbench project is also mentioned. The video concludes with an invitation to join the Discord community for more information and updates.

Takeaways

  • 🚀 **New Model Manager Interface**: Invoke 4.0 introduces a revamped model manager with a user-friendly interface for managing various models.
  • 🔍 **Customizable Default Settings**: Users can now customize default settings for main models and easily switch between models without issues.
  • 📚 **Direct Model Installation**: New models can be added via local path or URL, including direct access to Hugging Face repositories and the ability to scan and select individual files.
  • 🔗 **In-Place Install Option**: A new feature that allows models to be referenced directly from the hard drive without copying, saving space.
  • 💾 **Conversion Cache**: Invoke has a conversion cache that stores the most recent 20 GB of models in their converted state, making reuse easier and faster.
  • ⏱️ **Improved Performance**: The new gradient denoising feature reduces the time for inpainting results on the canvas, providing seamless generation in a single cycle.
  • 📈 **Non-Blocking Downloads**: Users can queue multiple model downloads without interrupting the user interface, allowing for immediate work on available models.
  • 📝 **Trigger Phrases Feature**: A new feature that allows users to configure and store common prompts or phrases for easy reuse with selected models.
  • 🎨 **Unified Canvas Update**: The unified canvas now leverages novel technology for a more efficient denoising process, particularly around masked areas.
  • 🔄 **Nvidia AI Workbench Partnership**: Invoke has partnered with Nvidia for the AI workbench project, with the 4.0 version to be updated soon for this integration.
  • 📋 **Release Notes and Roadmap**: For a detailed list of changes, users are encouraged to check the release notes, and join Discord for updates on the future roadmap.

Q & A

  • What is the main focus of Invoke's 4.0 update?

    -The main focus of Invoke's 4.0 update is the introduction of a new model manager interface, which includes a strong foundation for evolving Invoke as the ecosystem evolves, support for new model architectures, and a user-friendly interface for managing and customizing models.

  • How can users customize main models with default settings in the new model manager?

    -Users can customize main models with default settings by updating the default settings in the new model manager and enabling them. When the default settings button is hit in the main studio, all settings will update to what is ideal for the selected model.

  • What is the benefit of using the 'in place install' option when adding new models?

    -The 'in place install' option allows Invoke to make a direct reference to the file on the user's hard drive instead of copying the files into the Invoke model folder. This saves storage space and makes it easier to use the same model again without having to convert it each time.

  • How does the conversion cache work in the new model manager?

    -The conversion cache in the new model manager stores the most recent 20 GB of models in their converted state on the user's hard drive. This allows for faster reuse of models without needing to convert them every time. The cache size can be configured, and it can be set to zero if no additional hard drive space is desired for conversions.

  • What is the trigger phrases feature in the model manager?

    -The trigger phrases feature allows users to configure and store common prompts or phrases that are compatible with their main models or concept models. These can be easily referenced and reused in the studio, providing a quick way to input prompt terms that work with the selected model.

  • How has the gradient denoising feature improved the unified canvas experience?

    -The gradient denoising feature has streamlined the generation process on the unified canvas by performing the denoising process in a blurred area around the mask within a single generation cycle. This eliminates the need for two generation passes, reducing the time to generate an inpainting result and minimizing seams.

  • What is the significance of the partnership with Nvidia for Invoke?

    -The partnership with Nvidia involves building the Nvidia AI workbench project for Invoke, which will provide an updated 4.0 version soon. This collaboration aims to enhance the capabilities and user experience of Invoke, especially for those utilizing Nvidia's AI workbench.

  • How can users stay updated on the roadmap and future developments for Invoke?

    -Users can join the Invoke Discord community to hear more about the roadmap and what's coming next in terms of updates and features.

  • What is the process for users who want to start training their own custom AI models?

    -Users interested in training their own custom AI models can look into Invoke's training repositories, where they can find resources and get started with creating models for professional studios.

  • How can users customize prompt triggers for the stock models that come with Invoke Studio?

    -Users can customize prompt triggers for both the stock models that come with Invoke Studio and any custom models they upload into their account. By configuring the model, users can set specific settings and prompt triggers that the rest of the team can use with a click of a button.

  • What should users do if they want to see a full list of changes in the 4.0 version of Invoke?

    -Users who are interested in a comprehensive list of changes should refer to the release notes, which will detail all the updates that contribute to the enhanced functionality of Invoke 4.0.

  • How does the new model manager allow for simultaneous installation of multiple models?

    -The new model manager enables users to queue up multiple downloads without blocking the user interface. This allows for the immediate start of work on models that are already on the machine while other models are being downloaded.

Outlines

00:00

🚀 Introducing Invoke 4.0: New Model Manager and Features

The video introduces Invoke 4.0, highlighting its significance as the start of a new era for the platform. The update includes a new model manager and user interface, with an underlying robust foundation that supports evolving model architectures. The new model manager allows users to manage all models, including embeddings, VAEs, and control nets, directly from a panel. It also enables customization of main models with default settings, which can be updated and applied across the system. The video demonstrates the addition of new models through local paths or URLs, including the ability to scan repositories and select individual files. It also mentions Invoke's conversion cache, which stores the most recent 20 GB of models in a converted state to facilitate reuse. The feature to install multiple models simultaneously without blocking the user interface is also covered. Lastly, the video touches on the new trigger phrases feature, which lets users set specific prompts for their models for easier access during the creative process.

05:00

✨ Efficiency and Innovations in Invoke 4.0

This paragraph delves into the performance improvements and additional features of Invoke 4.0. It discusses the reduced time required to generate inpainting results on the canvas, thanks to a new single-generation cycle that eliminates seams. The video also mentions the release notes as a resource for a comprehensive list of updates in version 4.0. A partnership with Nvidia is announced, focusing on the Nvidia AI Workbench project for Invoke. The video encourages viewers to join Discord for updates on the roadmap and to explore the invoke training repositories for creating custom AI models. It concludes by noting that the changes discussed are already available, and users can customize prompt triggers for both stock and custom models in Invoke Studio, benefiting the entire team with a single configuration.

Mindmap

Keywords

💡Invoke 4.0

Invoke 4.0 refers to the latest version of the Invoke software, which is a significant update to the previous versions. It introduces a new model manager interface and other enhancements that aim to improve the user experience and the software's capabilities. The update is positioned as the beginning of a new era for Invoke, suggesting that it will continue to evolve with the ecosystem.

💡Model Manager Interface

The Model Manager Interface is a new feature in Invoke 4.0 that allows users to manage all their models directly from a single panel. It includes the ability to customize main models with default settings, which can be updated and applied across the system. This interface is crucial for streamlining the process of working with different models and settings.

💡Default Settings

Default settings in the context of Invoke 4.0 refer to the pre-configured parameters for models that users can apply when they want to use a model in its ideal state. By setting and enabling default settings, users can ensure that all their settings are optimized for a particular model, which is particularly useful when customizing models like 'sdxl lightning'.

💡Hugging Face Repo

A Hugging Face Repo is a repository on the Hugging Face platform where users can store, share, and manage their machine learning models. In Invoke 4.0, users can directly access and install models from a Hugging Face repository, which simplifies the process of adding new models to the system.

💡In-Place Install

The in-place install feature in Invoke 4.0 allows users to install models without copying the files into the Invoke model folder. Instead, it creates a direct reference to the file on the user's hard drive. This feature is beneficial for saving storage space and ensuring that the original files remain unchanged.

💡Conversion Cache

The Conversion Cache is a new feature in Invoke 4.0 that stores the most recent converted models to reduce the time it takes to use the same model again. By default, it keeps the last 20 GB of models in their converted state, which speeds up the loading time for frequently used models. Users can configure this cache size according to their storage preferences.

💡Trigger Phrases

Trigger phrases are configurable prompts for models in Invoke 4.0 that users can set up to quickly access and reuse common prompts or those that have been trained into the model. They can be used in the studio to quickly reference and pull in prompt terms that are compatible with the selected model, making it easier to work with complex or specific model requirements.

💡Unified Canvas

The Unified Canvas is a feature in Invoke 4.0 that allows for seamless generation and editing of models within a single interface. It includes the new gradient denoising feature, which improves the quality of generated content by minimizing seams in a single generation cycle.

💡Gradient Denoising

Gradient Denoising is a novel technology implemented in Invoke 4.0 that enhances the quality of generated images by performing a denoising process in a blurred area around the mask. This feature allows for smoother transitions and fewer visible seams in the generated content, resulting in more professional and polished outputs.

💡Nvidia AI Workbench

The Nvidia AI Workbench is a project partnership between Invoke and Nvidia, aimed at integrating Invoke into Nvidia's AI development platform. This partnership is significant as it will make Invoke more accessible and easier to use for users of Nvidia's AI Workbench, potentially expanding its user base and applications.

💡Discord

Discord, mentioned in the context of Invoke 4.0, is a communication platform where the Invoke community can gather to discuss updates, future roadmaps, and other aspects related to the software. It serves as a hub for users to engage with the developers and with each other, fostering a collaborative environment for using and improving Invoke.

Highlights

Invoke 4.0 has been released, marking the beginning of the next era for the platform.

A new model manager user interface has been introduced with a strong foundation for future evolution.

Support for new model architectures is built-in, including those that do not yet exist.

The model manager allows direct management of all models from a single panel.

Main models can be customized with default settings, simplifying the process of switching between models.

Users can add new models via local path, URL, or directly from a Hugging Face repository.

A conversion cache is implemented to store the most recent 20 GB of models in their converted state.

The cache size can be configured, and the user can set it to zero for no storage on the hard drive.

Multiple models can be installed simultaneously without blocking the user interface.

Trigger phrases can be configured for main models or concept models to store and reuse common prompts.

Unified Canvas now features a gradient denoising process that reduces generation time and minimizes seams.

The new gradient denoising technology is contributed by 'dunon' and operates in a blurred area around the mask.

Invoke 4.0 is one of the best and easiest to use versions, according to the release notes.

A partnership with Nvidia has been announced for the Nvidia AI Workbench project.

Invoke training repositories are available for creating custom AI models for professional studios.

Prompt triggers for stock models and custom models can be customized for the team's ease of use.

Stay tuned for more features and updates coming to Invoke.