Skip to content
InvokeAI Logo InvokeAI Documentation

Releases

v5.1.0

This release includes drawing tablet support and middle-mouse panning for Canvas, support for GGUF FLUX models, and an assortment of other fixes and enhancements.

❗ We are investigating an issue with Apple Silicon devices and SDXL. Users with Apple Silicon devices may wish to hold off on updating until we resolve this. See the Known Issues section below for more information.

Be sure to review the v5 release notes if you haven’t already upgraded to v5.

Enhancements

  • Canvas tablet support for both touch and pen input devices (e.g. drawing tablets).
  • Pressure sensitivity for pen input devices. This can be disabled in the Canvas settings.
  • Revised Canvas layout to better fit on smaller screens.
  • Improved image context menu layout. Thanks @joshistoast!
  • Improved Apple Pencil support.
  • Added long-press trigger for context menus. The canvas itself does not have a long-press trigger, but the same actions are accessible via menu button at the top-right corner of the canvas.
  • Middle-mouse button panning on Canvas. Thanks to @FloeHetling for getting this moving!
  • Support for GGUF FLUX models.
  • Create a new “session”, resetting all settings and Canvas to their defaults (except for model selection). These functions are in the menu next to the cancel button.
  • Crop to bbox on Canvas. You can crop an individual layer, or the whole canvas. Accessed via right-click menus.
  • Allow for a broader range of guidance values for flux models. Thanks @rikublock!
  • Updated translations. Thanks @rikublock @Ery4z @Vasyanator @Harvester62 @Phrixus2023!

Fixes

  • Duplicating a regional guidance layer with a reference image causes an error during graph building, preventing generation from working.
  • Recalling LoRAs can create duplicate LoRAs.
  • Fixed color picker tool edge case where wrong color could be detected while moving the cursor quickly.

Perf

  • Throttled color picker sampling to improve performance.

Internal

  • Bump all UI deps to latest.
  • Bump torch and xformers versions to latest.
  • Gracefully handle promise rejections in the UI’s metadata handlers.
  • Updated docker-compose.yml to use GHCR latest image. Thanks @jkbdco!

Docs

  • Added Ollama node to community nodes. Thanks @Jonseed!
  • Added FLUX support to docs. Thanks @aakropotkin!

Known Issues

Apple Silicon devices will output mushy noise on SDXL unless Regional Guidance or an IP Adapter is used.

The issue appears to be related to us bumping torch to v2.4.1, which was needed for GGUF FLUX support. The SDXL generation code wasn’t changed in this release. We are investigating the issue.

As a workaround, users can set their attention type to torch-sdp in their invoke.yaml configuration file. This will result in some increased memory utilization, but allow for generations to proceed as normal.

Installation and Updating

To install or update to v5.1.0, download the latest installer and follow the installation instructions

To update, select the same installation location. Your user data (images, models, etc) will be retained.

What’s Changed

New Contributors

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v5.0.2…v5.1.0

v5.0.2

This minor release fixes a regression with FLUX LoRAs introduced in v5.0.1.

Be sure to review the v5 release notes if you haven’t already upgraded to v5.

Fixes

  • Fix regression with FLUX LoRAs introduced in v5.0.1.

Enhancements

  • Show CLIP Vision models in the model manager UI, allowing for them to be deleted in case of corruption.

Installation and Updating

To install or update to v5.0.2, download the installer and follow the installation instructions

To update, select the same installation location. Your user data (images, models, etc) will be retained.

What’s Changed

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v5.0.1…v5.0.2

v5.0.1

This minor release includes support for Kohya FLUX LoRAs and a handful of fixes and enhancements.

Be sure to review the v5 release notes if you haven’t already upgraded to v5.

Enhancements

  • Support for Kohya FLUX LoRAs with lora_te1 layers (i.e. CLIP LoRA layers)
  • Default scheduler is now dpmpp_3m_k
  • First round of v5 translations @Harvester62 @Vasyanator @Atalanttore @Ery4z @rikublock
  • Improved FLUX img2img/inpainting
    • ❗ This is a breaking change. The trajectory_guidance_strength field on FluxDenoiseInvocation was removed in favor of a simpler solution that doesn’t need the extra field. More details in #6938.
  • Revised hotkeys focus tracking
  • Recall hotkeys work when viewer is closed and gallery is open
  • Added enter and esc hotkeys to apply and cancel filters and transforms
  • Model default settings support for FLUX guidance parameter

Fixes

  • Hide brush preview while staging
  • Prevent error toasts from being so large they cannot be closed
  • UI crash with TypeError: i.map is not a function
  • Canvas erroneously interactable after refresh while staging

Internal

  • More efficient image selection handling
  • More efficient newly-generated image handling
  • Revised resizable panels logic

Installation and Updating

To install or update to v5.0.1, download the installer and follow the installation instructions

To update, select the same installation location. Your user data (images, models, etc) will be retained.

What’s Changed

New Contributors

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v5.0.0…v5.0.1

v5.0.0

Invoke 5.0 Release: Our Biggest Update Yet — A new canvas with Layer & Flux support

Invoke 5.0 Update Video

Today, we’re thrilled to launch the most significant update to Invoke this year: Invoke 5.0. This release introduces Control Canvas, a new tool to generate, iterate, and refine images with unprecedented control and flexibility. We’re also announcing integration with the hugely popular Flux models.

You may notice that some things have moved around. We’ve created a few resources for you to get to know the new Control Canvas and everything new in Invoke 5.0:

Control Canvas

creation-tools-home Building on our Unified Canvas, the new Control Canvas integrates generation, iteration, and refinement into a single workspace. Highlights include:

  • Raster Layers: Easily draw, paint, and manipulate shapes or guides that can be moved and resized independently, giving you more flexibility in editing.
  • Editable Control Layers: Directly edit Control Layers (e.g., ControlNets) on the canvas, offering precise control over generation tasks without needing to re-upload images.
  • Canvas Layer Recall: Save and recall the canvas state of your creation, ensuring that works can be regenerated and saved with the proper metadata.

Flux.1 Model Integration

We’ve partnered with Black Forest Labs to integrate the highly popular Flux.1 models into Invoke. You can now:

  • Use Flux’s schnell model for commercial or non-commercial projects and dev models for non-commercial projects in your studio.
  • Generate with text-to-image, image-to-image, inpainting, and LoRA support, with more features coming soon.

If you are looking to use Flux [dev] for commercial purposes, you’ll need to obtain a commercial license from Black Forest Labs. This is included as an add-on option in our Professional Edition. If you are interested in learning more, you can get in touch with us.

  • Note: Flux Model support is currently only available for local inference on Windows and Linux systems.

Installation and Updating

To install or update to v5.0.0, download the installer and follow the installation instructions To update, select the same installation location. Your user data (images, models, etc) will be retained.

What’s Changed

New Contributors

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v4.2.9…v5.0.0

v4.2.9

FLUX

Please note these nodes are still in the prototype stage and are subject to change. This Node API is not stable!

We are supporting both FLUX dev and FLUX schnell at this time in workflows only. These will be incorporated into the rest of the UI in future updates. At this time, this is an initial and developing implementation - we’re bringing this in with the intent of long-term stable support for FLUX.

Default workflows can be found in your workflow tab: FLUX Text to Image and FLUX Image to Image. Please note that we have not added FLUX to the linear UI yet, LoRAs and Img2Img are not yet supported, but will be added soon.

Required Dependencies

Screenshot 2024-09-05 at 4 48 24 PM

In order to run FLUX on Invoke, you will need to download and install several models. We have provided options in the Starter Models (found in your Model Manager tab) for quantized and unquantized versions of both FLUX dev and FLUX schnell. Selecting these will automatically download the dependencies you need, listed below. These dependencies are also available for adhoc download in Starter Models list. Currently invoke only supports unquantized models, and bitsandbytes nf4 quantized models.

  • T5 encoder
  • CLIP-L encoder
  • FLUX transformer/unet
  • FLUX VAE

Considerations

FLUX is a large model, and has significant VRAM requirements. The full models require 24gb of VRAM on Linux — Windows PCs are less efficient, and thus need slightly more, making it difficult to run the full models.

To compensate for this, the community has begun to develop quantized versions of the DEV model - These are models with a slightly lower quality, but significant reductions in VRAM requirements.

Currently, Invoke is only supporting NVidia GPUs. You may be able to work out a way to get an AMD GPU to generate, however we’ve not been able to test this, and so can’t provide committed support for it. FLUX on MPS is not supported at this time.

Please note that the FLUX Dev model is a non-commercial license. You will need a commercial license to use the model for any commercial work.

Below are additional details on which model to use based on your system:

  • FLUX dev quantized starter model: non-commercial, >16GB RAM, ≥12GB VRAM
  • FLUX schnell quantized starter model: commercial, faster inference than dev, >16GB RAM, ≥ 12GB VRAM
  • FLUX dev starter model: non-commercial, >32GB RAM, ≥24GB VRAM, linux OS
  • FLUX schnell starter model: commercial, >32GB RAM, ≥24GB VRAM, linux OS

Running the Workflow

You can find a new default workflow in your workflows tab called FLUX Text to Image. This can be run with both FLUX dev and FLUX schnell models, but note that the default step count of 30 is the recommendation for FLUX dev. If running FLUX schnell, we recommend you lower your step count to 4. You will not be able to successfully run this workflow without the models listed above as required dependencies installed.

  • Navigate to the Workflows tab.
  • Press the Workflow Library button at the top left of your screen.
  • Select Default Workflows and choose the FLUX workflow you’d like to use.

The exposed fields will require you to select a FLUX model ,T5 encoder, CLIP Embed model, VAE, prompt, and your step count. If you are missing any models, use the “Starter Models” tab in the model manager to download and install FLUX Dev or Schnell.

Screenshot 2024-09-04 141124

We’ve also added a new default workflow named Flux Image to Image. This can be run vary similarly to the workflow described above with the additional ability to provide a base image.

Screenshot 2024-09-04 140846

Other Changes

  • Enhancement: add fields for CLIPEmbedModel and FluxVAEModel by @maryhipp
  • Enhancement: FLUX memory management improvements by @RyanJDick
  • Feature: Add FLUX image-to-image and inpainting by @RyanJDick
  • Feature: flux preview images by @brandonrising
  • Enhancement: Add install probes for T5_encoder and ClipTextModel by @lstein
  • Fix: support checkpoint bundles containing more than the transformer by @brandonrising

Installation and Updating

To install or update to v4.2.9, download the installer and follow the [installation instructions](https://invoke-ai.github.io/InvokeAI/installation/010_INSTALL_AUTOMATED/).

To update, select the same installation location. Your user data (images, models, etc) will be retained.

What’s Changed

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v4.2.8…v4.2.9

v4.2.8

v4.2.8 brings Prompt Templates to Invoke, new schedulers and a number of minor fixes and enhancements.

Prompt Templates

Prompt templates are often used for commonly-used style keywords, letting you focus on subject and composition in your prompts - but you can use them in other creative ways.

Thanks to @maryhipp for implementing Prompt Templates!

Creating a Prompt Template

Create a prompt template from an existing image generated with Invoke. We’ll add the positive and negative prompts from the image’s metadata as the template, and the image will be used as a cover image for the template.

You can also create a prompt template from scratch, uploading a cover image.

How it Works

Add a positive and/or negative prompt to your template. Use the {prompt} placeholder in the template to indicate where your prompt should be inserted into the template:

  • Template: highly detailed photo of {prompt}, award-winning, nikon dslr
  • Prompt: a super cute fennec fox cub
  • Result: highly detailed photo of a super cute fennec fox cub, award-winning, nikon dslr

If you omit the placeholder, the template will be appended to the end of your prompt:

  • Template: turtles
  • Prompt: i like
  • Result: i like turtles

Default Prompt Templates

We’re shipping a number of templates with the app, many of which were contributed by community members (thanks y’all!). We’ll update these as we continue developing Invoke with improvements and new templates.

Import and Export

You can import templates from other SD apps. We support CSV and JSON files with these columns/keys:

  • name
  • prompt or positive_prompt
  • negative_prompt

Export your prompt templates to share with others. When you export prompt templates, only your own templates are exported.

Preview and Flatten

Use the Preview button to see the prompt that will be used for generation. Flatten the prompt template to bake it into your prompts.

Compatible with Dynamic Prompts

You can use dynamic prompt in prompt templates, and they will work with dynamic prompts in your positive prompt box.

Other Changes

  • Enhancement: Added DPM++ 3M, DPM++ 3M Karras, DEIS Karras, KDPM 2 Karras, KDPM 2 Ancestral Karras and UniPC Karras schedulers @StAlKeR7779
  • Enhancement: Updated translations - Italian is 100%! Thanks @Harvester62!
  • Enhancement: Grounded SAM node (text prompt image segmentation) @RyanJDick
  • Enhancement: Update DepthAnything to V2 (small variant only) @blessedcoolant
  • Fix: Image downloads with correct filename
  • Fix: Delays with events (progress images will be smoother)
  • Fix: Jank with board selection when hiding or deleting boards
  • Fix: Error deleting images on systems without a “trash bin”
  • Fix: Upscale metadata included in SDXL Multidiffusion upscales @maryhipp
  • Fix: invoke.sh works with symlinks @max-maag
  • Internal: Continued work on the modular backend refactor @StAlKeR7779

Installation and Updating

To install or update to v4.2.8, download the installer and follow the installation instructions.

To update, select the same installation location. Your user data (images, models, etc) will be retained.

Missing models after updating from v3 to v4

See this FAQ.

Error during installation ModuleNotFoundError: No module named 'controlnet_aux'

See this FAQ

What’s Changed

New Contributors

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v4.2.7post1…v4.2.8

v4.2.7post1

🚨 v4.2.7post1 resolves an issue with Windows installs. 🚨

v4.2.7 includes gallery improvements and some major features focused on upscaling.

Upscaling

We’ve added a dedicated upscaling tab, support for custom upscaling models, and some new nodes.

Thanks to @RyanJDick (backend implementation), @chainchompa (frontend) and @maryhipp (frontend) for working on this!

Dedicated Upscaling Tab

The new upscaling tab provides a simple and powerful UI to Invoke’s MultiDiffusion implementation. This builds on the workflow released in v4.2.6, allowing for memory-efficient upscaling to huge output image sizes.

We’re pretty happy with the results!

image

4x scale, 4x_NMKD-Siax_200k upscale model, Deliberate_v5 SD1.5 model, KDPM 2 scheduler @ 30 steps, all other settings default

Requirements

You need 3 models installed to use this feature:

  • An upscale model for the first pass upscale
  • A main SD model (SD1.5 or SDXL) for the image-to-image
  • A tile ControlNet model of the same model architecture as your main SD model

If you are missing any of these, you’ll see a warning directing you to the model manager to install them. You can search the starter models for upscale, main, and tile to get you started.

image

Tips

  • The main SD model architecture has the biggest impact on VRAM usage. For example, SD1.5 @ 2k needs just under 4GB, while SDXL @ 2k needs just under 9GB. VRAM usage increases a small amount as output size increases - SD1.5 @ 8k needs ~4.5GB while SDXL @ 8k needs ~10.5GB.
  • The upscale and main SD model choices matter. Choose models best suited to your input image or desired output characteristics.
  • Some schedulers work better than others. KDPM 2 is a good choice.
  • LoRAs - like a detail-adding LoRA - can make a big impact.
  • Higher Creativity values give the SD model more leeway in creating new details. This parameter controls denoising start and end percentages.
  • Higher Structure values tell the SD model to stick closer to the input image’s structure. This parameter controls the tile ControlNet.

Custom Upscaling Models

You can now install and use custom upscaling models in Invoke. The excellent spandrel library handles loading and running the models.

spandrel can do a lot more than upscaling - it supports a wide range of “image to image” models. This includes single-image super resolution like ESRGAN (upscalers) but also things like GFPGAN (face restoration) and DeJPEG (cleans up JPEG compression artifacts).

A complete list of supported architectures can be found here.

Note: We have not enabled the restrictively-licensed architectures, which are denoted with a + symbol in the list.

Installing Models

We’ve added a few popular upscaling models to the Starter Models tab in the Model Manager - search for “upscale” to find them.

image

You can install models found online via the Model Manager, just like any other model. OpenModelDB is a popular place to get these models. For most of them, you can copy the model’s download link and paste in into the Model Manager to install.

Nodes

Two nodes have been added to support processing images with spandrel - be that upscaling or any of the other tasks these models support.

image
  • Image-to-Image - Runs the selected model without any extra processing.
  • Image-to-Image (Autoscale) - Runs the selected model repeatedly until the desired scale is reached. This node is intended for upscaling models specifically, providing some useful extra functionality:
    • If the model overshoots the target scale, the final image will be downscaled to the target scale with Lanczos resampling.
    • As a convenience, the output image width and height can be fit to a multiple of 8, as is required for SD. This will only resize down, and may change the aspect ratio slightly.
    • If the model doesn’t actually upscale the image, the scale parameter will be ignored.

Thanks to @maryhipp and @chainchompa for continued iteration on the gallery!

  • Cleaner boards UI.
  • Improved boards and image search UI.
  • Fixed issues where board counts don’t update when images are moved between boards.
  • Added a “Jump” button to allow you to skip pages of the gallery

<video src=https://github.com/user-attachments/assets/b834cc36-995a-464e-af3f-68cf3b38818f>

Other Changes

  • Enhancement: When installing starter models, the description is carried over. Thanks @lstein!
  • Enhancement: Updated translations.
  • Fix: Model unpatching when running on CPU, causing bad/no outputs.
  • Fix: Occasional visible seams on images with smooth textures, like skies. MultiDiffusion tiling now uses gradient blending to mitigate this issue.
  • Fix: Model names overflow the model selection drop-downs.
  • Internal: Backend SD pipeline refactor (WIP). This will allow contributors to add functionality to Invoke more easily. This will be behind a feature flag until the refactor is complete and tested. Thanks to @StAlKeR7779 for leading the effort, with major contributions from @dunkeroni and @RyanJDick.

Installation and Updating

To install or update to v4.2.7post1, download the installer and follow the installation instructions.

To update, select the same installation location. Your user data (images, models, etc) will be retained.

Missing models after updating from v3 to v4

See this FAQ.

Error during installation ModuleNotFoundError: No module named 'controlnet_aux'

See this FAQ

What’s Changed

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v4.2.6…v4.2.7post1

v4.2.7

v4.2.7 includes gallery improvements and some major features focused on upscaling.

Upscaling

We’ve added a dedicated upscaling tab, support for custom upscaling models, and some new nodes.

Thanks to @RyanJDick (backend implementation), @chainchompa (frontend) and @maryhipp (frontend) for working on this!

Dedicated Upscaling Tab

The new upscaling tab provides a simple and powerful UI to Invoke’s MultiDiffusion implementation. This builds on the workflow released in v4.2.6, allowing for memory-efficient upscaling to huge output image sizes.

We’re pretty happy with the results!

image

4x scale, 4x_NMKD-Siax_200k upscale model, Deliberate_v5 SD1.5 model, KDPM 2 scheduler @ 30 steps, all other settings default

Requirements

You need 3 models installed to use this feature:

  • An upscale model for the first pass upscale
  • A main SD model (SD1.5 or SDXL) for the image-to-image
  • A tile ControlNet model of the same model architecture as your main SD model

If you are missing any of these, you’ll see a warning directing you to the model manager to install them. You can search the starter models for upscale, main, and tile to get you started.

image

Tips

  • The main SD model architecture has the biggest impact on VRAM usage. For example, SD1.5 @ 2k needs just under 4GB, while SDXL @ 2k needs just under 9GB. VRAM usage increases a small amount as output size increases - SD1.5 @ 8k needs ~4.5GB while SDXL @ 8k needs ~10.5GB.
  • The upscale and main SD model choices matter. Choose models best suited to your input image or desired output characteristics.
  • Some schedulers work better than others. KDPM 2 is a good choice.
  • LoRAs - like a detail-adding LoRA - can make a big impact.
  • Higher Creativity values give the SD model more leeway in creating new details. This parameter controls denoising start and end percentages.
  • Higher Structure values tell the SD model to stick closer to the input image’s structure. This parameter controls the tile ControlNet.

Custom Upscaling Models

You can now install and use custom upscaling models in Invoke. The excellent spandrel library handles loading and running the models.

spandrel can do a lot more than upscaling - it supports a wide range of “image to image” models. This includes single-image super resolution like ESRGAN (upscalers) but also things like GFPGAN (face restoration) and DeJPEG (cleans up JPEG compression artifacts).

A complete list of supported architectures can be found here.

Note: We have not enabled the restrictively-licensed architectures, which are denoted with a + symbol in the list.

Installing Models

We’ve added a few popular upscaling models to the Starter Models tab in the Model Manager - search for “upscale” to find them.

image

You can install models found online via the Model Manager, just like any other model. OpenModelDB is a popular place to get these models. For most of them, you can copy the model’s download link and paste in into the Model Manager to install.

Nodes

Two nodes have been added to support processing images with spandrel - be that upscaling or any of the other tasks these models support.

image
  • Image-to-Image - Runs the selected model without any extra processing.
  • Image-to-Image (Autoscale) - Runs the selected model repeatedly until the desired scale is reached. This node is intended for upscaling models specifically, providing some useful extra functionality:
    • If the model overshoots the target scale, the final image will be downscaled to the target scale with Lanczos resampling.
    • As a convenience, the output image width and height can be fit to a multiple of 8, as is required for SD. This will only resize down, and may change the aspect ratio slightly.
    • If the model doesn’t actually upscale the image, the scale parameter will be ignored.

Thanks to @maryhipp and @chainchompa for continued iteration on the gallery!

  • Cleaner boards UI.
  • Improved boards and image search UI.
  • Fixed issues where board counts don’t update when images are moved between boards.
  • Added a “Jump” button to allow you to skip pages of the gallery

<video src=https://github.com/user-attachments/assets/b834cc36-995a-464e-af3f-68cf3b38818f>

Other Changes

  • Enhancement: When installing starter models, the description is carried over. Thanks @lstein!
  • Enhancement: Updated translations.
  • Fix: Model unpatching when running on CPU, causing bad/no outputs.
  • Fix: Occasional visible seams on images with smooth textures, like skies. MultiDiffusion tiling now uses gradient blending to mitigate this issue.
  • Fix: Model names overflow the model selection drop-downs.
  • Internal: Backend SD pipeline refactor (WIP). This will allow contributors to add functionality to Invoke more easily. This will be behind a feature flag until the refactor is complete and tested. Thanks to @StAlKeR7779 for leading the effort, with major contributions from @dunkeroni and @RyanJDick.

Installation and Updating

To install or update to v4.2.7, download the installer and follow the installation instructions.

To update, select the same installation location. Your user data (images, models, etc) will be retained.

Missing models after updating from v3 to v4

See this FAQ.

Error during installation ModuleNotFoundError: No module named 'controlnet_aux'

See this FAQ

What’s Changed

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v4.2.6post1…v4.2.7

v4.2.6post1

v4.2.6post1 fixes issues some users may experience with memory management and sporadic black image outputs.

Please see the v4.2.6 release for full release notes.

💾 Installation and Updating

To install or update to v4.2.6post1, download the installer and follow the installation instructions.

To update, select the same installation location. Your user data (images, models, etc) will be retained.

Missing models after updating from v3 to v4

See this FAQ.

Error during installation ModuleNotFoundError: No module named 'controlnet_aux'

See this FAQ

What’s Changed

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v4.2.6…v4.2.6post1

v4.2.6

v4.2.6 includes a handful of fixes and improvements, plus three major changes:

  • Gallery updates
  • Tiled upscaling via MultiDiffusion
  • Checkpoint models work without conversion to diffusers

We’ve made some changes to the gallery, adding features, improving the performance of the app and reducing memory usage. The changes also fix a number of bugs relating to stale data - for example, a board not updating as expected after moving an image to it.

Thanks to @chainchompa and @maryhipp for working on this major effort.

Pagination & Selection

Infinite scroll is dead, long live infinite scroll!

The gallery is now paginated. Selection logic has been updated to work with pagination. An indicator shows how many images are selected and allows you to clear the selection entirely. Arrow keys still navigate.

https://github.com/invoke-ai/InvokeAI/assets/4822129/128c998a-efac-41e5-8639-b346da78ca5b

The number of images per page is dynamically calculated as the panel is resized, ensuring the panel is always filled with images.

Boards UI Refresh

The bulky tiled boards grid has been replaced by a scrollable list. The boards list panel is now a resizable, collapsible panel.

https://github.com/invoke-ai/InvokeAI/assets/4822129/2dd7c316-36e3-4f8d-9d0c-d38d7de1d423

Search for boards by name and images by metadata. The search term is matched against the image’s metadata as a string. We landed on full-text search as a flexible yet simple implementation after considering a few methods for search.

https://github.com/invoke-ai/InvokeAI/assets/4822129/ebe2ecfe-edb4-4e09-aef8-212495b32d65

Archived Boards

Archive a board to hide it from the main boards list. This is purely an organizational enhancement. You can still interact with archived boards as you would any other board.

https://github.com/invoke-ai/InvokeAI/assets/4822129/7033b7a1-1cb7-4fa0-ae30-5e1037ba3261

Image Sorting

You can now change the sort for images to show oldest first. A switch allows starred images to be placed in the list according to their age, instead of always showing them first.

https://github.com/invoke-ai/InvokeAI/assets/4822129/f1ec68d0-3ba5-4ed0-b1e8-8e8bc9ceb957

Tiled Upscaling via MultiDiffusion

MultiDiffusion is a fairly straightforward technique for tiled denoising. The gist is similar to other tiled upscaling methods - split the input image up in to tiles, process each independently, and stitch them back together. The main innovation for MultiDiffusion is to do this in latent space, blending the tensors together continuously. This results in excellent consistency across the output image, with no seams.

This feature is exposed as a Tiled MultiDiffusion Denoise Latents node, currently classified as a beta version. It works much the same as the OG Denoise Latents node. You can find an example workflow in the workflow library’s default workflows.

We are still thinking about to expose this in the linear UI. Most likely, we expose this with very minimal settings. If you want to tweak it, use the workflow.

Thanks to @RyanJDick for designing and implementing MultiDiffusion.

How to use it

This technique is fundamentally the same as normal img2img. Appropriate use of conditioning and control will greatly improve the output. The one hard requirement is to use the Tile ControlNet model.

Besides that, here are some tips from our initial testing:

  • Use a detail-adding or style LoRAs.
  • Use a base model best suited for the desired output style.
  • Prompts make a difference.
  • The initial upscaling method makes a difference.
  • Scheduler makes a difference. Some produce softer outputs.

VRAM Usage

This technique can upscale images to very large sizes without substantially increasing VRAM usage beyond what you’d see for a “normal” sized generation. The VRAM bottlenecks then become the first VAE encode (Image to Latents) and final VAE decode (Latents to Image) steps.

You may run into OOM errors during these steps. The solution is to enable tiling using the toggle on the Image to Latents and Latents to Image nodes. This allows the VAE operations to be done piecewise, similar to the tiled denoising process, without using gobs of VRAM.

There’s one caveat - VAE tiling often introduces inconsistency across tiles. Textures and colors may differ from tile to tile. This is a function of diffusers’ handling of VAE tiling, not the new tiled denoising process. We are investigating ways to improve this.

Takeaway: If your GPU can handle non-tiled VAE encode and decode for a given output size, use that for best results.

Checkpoint models work without conversion to diffusers

The required conversion of checkpoint format models to diffusers format has long been a pain point. The diffusers library now supports loading single-file (checkpoint) models directly, and we have removed the mandatory checkpoint-to-diffusers conversion step.

The main user-facing change is that there is no longer a conversion cache directory.

Major thanks to @lstein for getting this working.

📈 Patch Nodes for v4.2.6

Enhancements

  • When downloading image metadata, graphs or workflows, the JSON file includes the image name and type of data. Thanks @jstnlowe!
  • Add clear_queue_on_startup config setting to clear problematic queues. This is useful for a rare edge case where your queue is full of items that somehow crash the app. Set this to true, and the queue will clear before it has time to attempt to execute the problematic item. Thanks @steffy-lo!
  • Performance and memory efficiency improvements for LoRA patching and model offloading.
  • Addition of a simplified model installation methods to the Invocation API: download_and_cache_model, load_local_model and load_remote_model. These methods allow models to be used without needing them to be added to the model manager. For example, we are now using these methods to load ESRGAN models.
  • Support for probing and loading SDXL VAE checkpoint.
  • Updated gallery UI.
  • Checkpoint models work without conversion to diffusers.
  • When using a VAE in tiled mode, you may now select the tile size.

Fixes

  • Fix handling handling of 0-step denoising process.
  • If a control image’s processed version is missing when the app loads, it is now re-processed.
  • Fixed an issue where a model’s size could be misreported as 0, possibly causing memory issues.
  • Fixed an issue where images - especially large images - may fail to delete.

Performance improvements

  • Improved LoRA patching.
  • Improved RAM <-> VRAM model transfer performance.

Internal changes

  • The DenoiseLatentsInvocation has had its internal methods split up to support tiled upscaling via MultiDiffusion. This included some amount of file shuffling and renaming. The invokeai package’s exported classes should still be the same. Please let us know if this has broken an import for you.
  • Internal cleanup, intending to eliminate circular import issues. There’s a lot left to do for this issue, but we are making progress.

💾 Installation and Updating

To install or update to v4.2.6, download the installer and follow the installation instructions.

To update, select the same installation location. Your user data (images, models, etc) will be retained.

Missing models after updating from v3 to v4

See this FAQ.

Error during installation ModuleNotFoundError: No module named 'controlnet_aux'

See this FAQ

What’s Changed

New Contributors

Full Changelog: https://github.com/invoke-ai/InvokeAI/compare/v4.2.4…v4.2.6

This site was designed and developed by Aether Fox Studio.