Minor Release: December 18

General Enhancements

This release includes support for FLUX canny and depth control models compatible with FLUX.1 [dev] models, new ways to edit Control Layers on Canvas, and improvements to LoRA model uploads.

Minor Release: December 12

General Enhancements

This release includes improvements for easily adding FLUX models from Hugging Face, an assortment of fixes, and other minor improvements.

Minor Release: December 9

General Enhancements

This release includes access to model training for Indie-tier users, layout and navigation updates, plus an assortment of fixes and other minor improvements.

Minor Release: December 4

General Enhancements

This release includes initial support for FLUX Regional Guidance, arrow key nudge on Canvas, plus an assortment of fixes and other minor improvements.

Minor Release: November 20

General Enhancements

This release includes image batching for workflows, additional composition nodes, and other minor improvements. 

Minor Release: November 19

General Enhancements

This release includes updates to improve job routing and support for FLUX.1 [dev] model uploads for users with a commercial license. You can read more about uploading FLUX.1 [dev] models here.

Minor Release: November 8

General Enhancements

Control Layer UX Improvements

This release updates Control Layers to automatically filter based on the model selected, rather than requiring a Filter to be manually applied. Users can continue to use advanced filter functions by clicking "Advanced" when they choose a control model.

Denoising Strength Updates

The Denoising Strength value has been moved to the top of the Layers Tab, and been updated with some helpful tooltips and visuals to improve the communication of how this feature works.

Minor Release: November 5

General Enhancements

This release expands layer merging capabilities in the Canvas and makes a number of other fixes and enhancements.

Minor Release: November 4

General Enhancements

Organizations can now disable system models within Project Settings

Minor Release: October 29

✨ New Feature: Select Object

This release introduces a new, powerful feature in Invoke's Control Canvas: Select Object. This tool makes it easy to select specific objects within an image and instantly convert them into editable layers. With it, you can fine-tune specific image elements through inpainting, control layers, and image transformations directly on raster elements.

Other Enhancements

We have added support for XLabs FLUX IP Adapters, support for more conversions between Canvas Layers, and various UI improvements.

You can find a demo of the features in this release here: Invoke 5.3 introduces a new Select Object Tool and Flux Support for Global Reference Images 

Minor Release: October 18

✨ New Feature: Support for FLUX ControlNets

This release includes support for both XLabs and InstantX ControlNets for FLUX. We've found the Union Pro model substantially outperforms the other models and it has been added to starter models. FLUX ControlNets can be accessed used in both Workflows and the Linear UI.

Please note: You will not be able to use FLUX Controlnets without an active FLUX Dev license. For more information on sourcing a commercial FLUX Dev license, please reach out to us.

Other Enhancements

We have improved the user experience for img2img by adding 3 modes for transforming on the Control Canvas. Other enhancements include starter model bundles in Model Manager, support for bulk image uploads, and the ability to sort boards by name or date.

Minor Release: October 9

✨ New Feature: Drawing tablet support

This release includes drawing tablet support on Canvas for both touch and pen input devices (e.g. drawing tablets). Pressure sensitivity for pen input devices can be controlled in Canvas settings. You can read more about tablet support here.

General Enhancements

We have improved Canvas layout to better fit smaller screens, added middle-mouse panning for Canvas, updated the workflows list to function similar to the style presets list, and made an assortment of other fixes and smaller enhancements.

Major Release: September 24

Invoke 5.0: Introducing Control Canvas, Flux Integration, and Enhanced Prompt Templates

We are excited to introduce Invoke 5.0, our biggest update of the year! This release unveils Control Canvas, a revolutionary tool for generating, iterating, and refining images with unparalleled control. We’ve also integrated the highly popular Flux models for both commercial and non-commercial use, and improved Prompt Templates to make saving, sharing, and reusing prompts easier than ever.

✨ New Feature: Control Canvas

Control Canvas is a major upgrade to our renowned canvas, designed for artists who need both flexibility and precision. This new workspace integrates generation and editing tools in one interface, enhancing the experience for professional users.

Control Canvas includes the following new capabilities:

• Raster Layers: Draw, paint, and manipulate shapes and images on the canvas, with layers functioning like those in traditional photo editing tools. Resize and adjust elements without impacting others.
• Editable Control Layers: Modify Control Layers directly with Invoke’s drawing and design tools. These layers can be fine-tuned for precise guidance.
• Canvas Layer Recall: Save and recall previous canvas states with metadata to ensure consistency and accuracy in your creative process.

✨ New Feature: Flux.1 Integration

We’re thrilled to announce the integration of Flux.1 models, developed by Black Forest Labs. Since its release, Flux.1 has become one of the most popular AI image generation models available.

• Flux for Non-Commercial Use: The Flux.1 [schnell] and [dev] models are available for non-commercial projects in Invoke’s Community Edition.
• Flux for Commercial Use: If you're using the Professional Edition, you have the option to use Flux.1 [dev] for commercial purposes. For commercial access, contact our sales team.

Flux supports text-to-image, image-to-image transformations, LoRA support, and inpainting. We’ll continue integrating Flux with Control Canvas as the ecosystem grows.

✨ Enhanced Feature: Prompt Templates

While previously released, Prompt Templates are now easier to use than ever. Save, share, and reuse prompts to ensure consistent results and efficiency. Use the built-in templates or create your own for specific styles or effects, making it simple to recreate specific looks or processes across projects.

Getting Started with Invoke 5.0

• Professional Edition Users: Invoke 5.0 is live for all users on the Indie, Premier, and Enterprise plans. If you're interested in Flux models for commercial use, reach out to our sales team.
• Community Edition Users: Download the latest version of Invoke 5.0 on GitHub.

We can’t wait to see what you create with this latest version. As always, your feedback is essential in helping us improve and support your creative journey.

Watch the full release video

Minor Release: September 5

✨New Feature: Updated UX for Project Page

We have redesigned the project page for a more intuitive and streamlined user experience. The new layout enhances usability, making it easier to organize and manage your projects efficiently - You'll see your activity as well as your team's in a single panel, with quick access to common features like Prompt Templates and Workflows.

✨New Feature: New Default Workflows

Introducing six new default workflows to help streamline the creative process - Whether you want to start from a sketch or use references images, these default workflows provide a pre-set path, allowing for quicker onboarding to core capabilities of Invoke. You'll find these workflows in your Default Workflows on the Workflows tab.

Major Release: August 23

✨New Feature: Prompt Templates

Prompt Templates can be used to quickly produce assets in a pre-configured style using a templated prompt, letting you focus on subject and composition in your prompts - but you can also use them in other creative ways.

Minor Release: August 1

General Bug Fixes

We have resolved a number of minor issues that could cause issues during log-in, and updating and managing galleries and boards.

Minor Release: July 18

✨New Model Management Feature: Hugging Face Model Validation

Users will be prevented from importing invalid models using the Hugging Face model form - Uploaded models now need to be formatted with a valid repository ID in order to import models.

Minor Release: July 3

✨New Gallery Feature: Board Pagination

Images are now organized by pagination instead of an infinite scroll - This is primarily intended to service users who have created extremely large galleries/boards, as they can now navigate to specific pages and identify information more easily with this.

✨New Gallery Feature: Metadata Search

Users can now use the search bar at the top of their selected board's gallery to easily search for specific images using a general metadata search -- Type in a prompt word, a model name, or any other information stored in the image metadata, and easily filter your images to only those you're actively looking to see.

Minor Release: June 11

✨New Admin Feature: SSO Auto-Provisioning

Enterprise: When adding users through SSO systems, users who log-in will automatically be provisioned an Invoke account with MEmber access unless the Enterprise seat limit has been reached.

✨New Admin Feature: Usage Breakdowns

Administrators for Premier & Enterprise accounts can now view breakdowns of usage by user, allowing for you to determine the activity and generation frequency of individual users.

Minor Release: June 5

Aligned with v4.2.4 of the Community Edition

✨New Feature: Image Comparison UI

Introducing a new image comparison UI with multiple modes for better visual analysis of generated assets. You can now use slider, side-by-side, and hover comparison modes to compare images more effectively. To get started, simply ALT+Click an image in your gallery, or right click the image and select "Select for Compare"

Bug Fix: Bulk Download UI

We have resolved authorization issues that users were experiencing with the bulk download UI if not logged in while accessing images.

Minor Release: May 31

✨ Uploading Model Improvements

You should find that uploading models is more streamlined and easier with updates to how we evaluate and analyze models upon upload.

✨ Training Improvements

A number of minor enhancements have been made to improve the ease of use for dataset management in the Training application. You can now manage and caption larger Datasets more easily.

Minor Release: May 21

Aligned with v4.2.2 of the Community Edition

✨ Undo/redo in Workflows

Undo/redo redo now available in the workflow editor. Currently, we group like changes together. Please let us know any feedback you have on how undo actions are grouped, so we can continue to make this experience as intuitive as possible!

✨ Load Workflow from Any Image

With this release, graphs are embedded in all images generated by Invoke. Images generated in the workflow editor also have the enriched workflow embedded separately. The Load Workflow button will load the enriched workflow if it exists, else it will load the graph.

You'll see a new Graph tab in the metadata viewer showing the embedded graph.

Graph vs Workflow

Graphs are used by the backend and contain minimal data. Workflows are an enriched data format that includes a representation of the graph plus extra information, including things like:

• Title, description, author, etc
• Node positions
• Custom node and field labels

This new feature embeds the graph in every image - including images generated on the Generation or Canvas tabs.

Canvas Caveat

This functionality is available only for individual canvas generations - not the full composition. 

? Other Fixes and Enhancements

• Min/max LoRA weight values extended (-10 to +10)
• Denoising strength and layer opacity are retained when sending image to initial image
• SDXL T2I Adapter only blocks invoking when dimensions aren't multiple of 32 (was erroneously 64)
• Improved UX when manipulating edges in workflows
• Connected inputs on nodes collapse, hiding the nonfunctional UI component
• Use ctrl/cmd-shift-v to paste copied nodes with input edges
• Fix: visible seams when outpainting
• Fix: edge case that could prevent workflows from loading if user hadn't opened the workflows tab yet
• Fix: minor inefficiency with control adapter auto-process (control layers only)

Major Release: May 9

Aligned with v4.2.0 of the Community Edition

Control Layers

Control Layers give you control over specific areas of the image by using the new Regional Guidance Layer type. Draw a mask and set a positive prompt, negative prompt, or any number of IP Adapters to be applied to the masked region. Global layers like Control Adapters (ControlNet & T2I Adapters), IP Adapters, and Initial Images are visualized on the canvas.

You can learn more about Control Layers by reviewing the help article.

❗ Interface Changes

A few important changes have been made to the interface as part of this release.

• Control Adapters and IP Adapters are now added to generations by using the Control Layers tab in the options panel.
• The Image to Image tab has been deprecated, and this functionality has been embedded into the Generation Tab - Simply add a Global Initial Image Layer
• With the addition of the Control Layers interface, the Image Viewer and Control Layers interface can be swapped between manually on the Generation tab by clicking the Switch View button on the top right of your workspace, or using the Z hotkey.

Patch Notes for v4.2.0

Enhancements

• Control Layers
• Added TCD scheduler for latest generation of SDXL hypermodels.
• Image viewer & UI updates

Fixes

• Fixed inpainting models on canvas
• Fixed SDXL Checkpoint Inpainting models
• Fixed Control Adapter processors' image size constraints

Major Release: April 10

This release introduces minor updates to how the system processes image generation requests, which improves the responsivity of user requests, as well as the speed of image generations.

Improvements

• (Beta) Regional Prompting: As a beta feature, Regional Prompting capabilities have been added to both IP-Adapter and Denoise Latents nodes in the workflow editor. More information (and an interface) coming soon!
• Refiner Inpainting: For users utilizing SDXL, the Refiner now is utilized effectively during inpainting processes.
  
• Project Invites: You can now invite users to projects directly - Get started easier and quicker than ever!
  

Major Release: April 4

This release introduces major changes to our model management system offering advanced customization capabilities on a per model level, as well as some updates to how models are tracked in order to provide additional model-level visibility and control.

Improvements

• Custom Model Defaults: You can now set a number of defaults on a per-model basis directly within model settings. You'll never have to remember the precise settings you need for any of your models.
• Prompt Triggers: On both Main Models and Concept Models (LoRAs), you can now configure prompt triggers. Whether you use this for required terms in using that model, or are configuring default prompts that you want to use when accessing the model, you can type the < key in the prompt box to get an easy type-ahead menu for adding the prompt to your workflow.

Major Release: March 26

This release introduces significant updates to how the system processes image generation requests, which improves the responsivity of user requests, as well as the speed of image generations.

Improvements

• Speed Increases: Optimizing the system for speed - Everything should feel just a bit snappier, and generation times should be shorter.
  

Major Release: March 21

This release introduces significant updates to the operational efficiency of image generations, as well as novel mechanisms for the Unified Canvas.

Improvements

• Gradient Denoising - Canvas: The Canvas now has vastly improved speed and quality for inpainting on the Unified Canvas. Some updates have been made to the Compositing settings to control this new feature, and the previous process which leveraged two denoising steps has been deprecated in favor of this more efficient and effective workflow.
  

Major Release: February 12

This release introduces significant updates to VRAM cache management, monitoring capabilities, and security enhancements, along with improvements to database interactions and error handling. These changes aim to improve performance, usability, and security of the software.

Improvements

• VRAM Cache management : Enhanced VRAM cache management for improved performance and efficiency in resource utilization.
• Browser Security Enhancements: Implemented additional security protocols into the UI to enhance security against common web vulnerabilities
• Save Finished Queue Items to DB: Improved the reliability of saving finished queue items to the database, enhancing data persistence and integrity.
• Improved Error Handling on Invoke Stats: Enhanced error handling mechanisms in Invoke Stats for more robust and informative error feedback.

Major Release: February 5

This release introduces an entirely new Workflow capabilities - Libraries! With the Library, you have an easy to use mechanism for sharing and deploying workflows across your teams.

New Features

Workflow Library

• Description: Users now have the ability to share workflows with other members of their project, allowing for easy loading/editing of the project workflow library.

Improvements

• Image Generation Speed Enhancement: We've changed the way we store images in order to improve the generation time on your end!
• Canvas Hotkey Enhancements: We've implemented new Canvas hotkeys, helping users change brush size and more with an easier set of interactions (Ctrl+scroll up and down to change your brush size, Press space to move the bounding box from any tool, etc.)
• LoRA Toggle: You can now enable or disable LoRAs without having to remove them entirely!

Major Release: January 30

This release introduces significant updates to VRAM cache management, monitoring capabilities, and security enhancements, along with improvements to database interactions and error handling. These changes aim to improve performance, usability, and security of the software.

Improvements

• UX Feedback: We heard the feedback and tackled a lot of small enhancements to the new UI/UX. We're happy to see it was so well received, and are confident the updates made (things like settings defaults, tooltips, and queue management have all seen updates based on your feedback)
• Width/Height by Model Architecture: Switching models now resets width/height based on the architecture type of this is different than the previous model.

Bug Fixes

• Width/Height Fixes: Updating width/height from images now accurately updates the visual indication in the aspect ratio setting
• Panel Resizing: Now when users resize the browser, panels will maintain the smallest possible size. Previously, they were expanding and taking up more space than desired.

Major Release: January 16

This release marks a significant milestone for our platform as we embark on a journey to better serve businesses and enterprises by addressing their deployment challenges for generative AI. With a major UI/UX overhaul reflecting our evolving brand identity, we reaffirm our commitment to our mission, team, and open-source software (OSS) principles. In addition to aesthetic improvements, this version brings a range of functional enhancements and bug fixes aimed at streamlining operations and enhancing user experience.

New Features

UI/UX Updates

• Description: A comprehensive redesign of our user interface and user experience to streamline the user experience

Improvements

• Placeholder Text for HF Upload: Improved clarity in the UI by introducing placeholder text for the Hugging Face upload feature, guiding users more intuitively through the upload process.
• Color Updates for Team Size Stepper: Enhanced the team size selection step in the UI with updated color schemes for better visibility
• User Context Updates on Name Change: Ensured that user context is automatically updated across the platform when a user changes their name, maintaining consistency and improving user experience.