How to update Krita.Ai plugin and use the best AI model for generative fill

GraphicalFinity
31 Aug 202405:59

TLDRThis video tutorial guides viewers through the process of updating the Krita AI plugin to leverage the latest AI model, Flux. It covers manual upgrade steps, especially for versions 1.15 and older, emphasizing the need to backup server settings. The video demonstrates downloading the latest plugin, installing it via the 'Import Python Model' menu in Krita, and resolving server outdated issues by downloading necessary components. It also introduces new features like multi-language support and tag completion. The tutorial further explores the experimental Flux model, detailing its setup, potential performance issues, and tips for faster generation, such as adjusting NVIDIA settings. It concludes with practical advice on using different models for various tasks within Krita AI.

Takeaways

  • 😀 The video demonstrates how to update the Krita AI plugin to utilize the latest AI model for generative fill.
  • ✅ Krita AI does not have an automatic update feature, so updates must be done manually.
  • 🔍 The latest version of Krita AI supports the new Flux model, which is explored in the video.
  • 📖 Users are advised to check the website for upgrade instructions, especially if using version 1.15 or older.
  • 🔄 For versions 1.15 and older, it's crucial to copy server settings before updating.
  • 📌 The video shows how to check the current version of Krita AI within the settings.
  • 📂 The update process involves downloading the latest plugin file and importing it into Krita via the 'Import Python Module' menu.
  • 💻 After updating, users may need to install additional components and restart the server within the plugin's configuration.
  • 🆕 The updated plugin introduces new features such as multi-language support and tag completion.
  • 🌐 Flux model support is experimental and may require manual downloading of the model file.
  • 🚀 For better performance with the Flux model, especially with limited VRAM, the video suggests adjusting NVIDIA settings to prefer no SIMS fallback.

Q & A

  • What is the main topic of the video?

    -The main topic of the video is how to update the Krita AI plugin and use the new Flux model for generative fill.

  • Does Krita AI have an auto-update feature?

    -No, the video mentions that Krita AI does not have an auto-update feature, so the update process must be done manually.

  • What is the latest model supported by the updated version of Krita AI?

    -The latest model supported by the updated version of Krita AI is the Flux model.

  • What should users do if they are using Krita AI version 1.15 or older?

    -If users are on version 1.15 or older, they need to copy their server settings before upgrading.

  • How can users check their current version of Krita AI?

    -Users can check their current version of Krita AI by opening the settings of the Krita AI plugin and looking at the bottom of the window.

  • Where can users find the download link for the latest version of Krita AI?

    -Users can find the download link for the latest version of Krita AI on the releases page of the plugin's homepage.

  • What is the process to install the new version of Krita AI after downloading it?

    -After downloading the new version, users should go back to Krita and use the 'Tools' -> 'Script' -> 'Import Python module' menu to select and install the downloaded file.

  • What happens when the server is out of date in Krita AI?

    -When the server is out of date, a message appears in the Krita AI panel stating that the server is missing components, and the user is prompted to install them.

  • How can users download and use the Flux model in Krita AI?

    -Users need to manually download the Flux model, set the style to Flux in the plugin configuration, and move the downloaded model to the models folder within the plugin settings.

  • What is the recommended model to test within the Flux models?

    -The recommended model to test within the Flux models is the 'Flux Chel' model as it is less resource-heavy.

  • What is a tip to speed up generation time when using the Flux model?

    -A tip to speed up generation time when using the Flux model is to open the Nvidia configuration panel and set the 'Max Tessellation' fallback policy to 'Prefer no tessellation fallback'.

  • What should users do if they encounter issues with the Flux model during the fill or refine process?

    -If users encounter issues with the Flux model during the fill or refine process, they should use the SD 1.5 or the XL model styles instead.

Outlines

00:00

🔄 Upgrading Creta AI and Exploring Flux Model

This paragraph outlines the process of upgrading an existing Creta AI version and introduces the new Flux model. The speaker mentions that there isn't an auto-update feature, so manual updating is necessary. The latest version supports the Flux model, and the speaker checks the website for upgrade instructions. A warning is highlighted for users on version 1.15 or older to copy their server settings before upgrading. The speaker verifies their Creta AI version (1.17) and decides to reinstall to upgrade. The process involves downloading the latest version, using the 'import python model' menu in Creta AI, and overriding the existing installation. After the upgrade, the speaker encounters a message about an outdated server, which is resolved by installing missing components and restarting the server. New features like multi-language support and tag completion are mentioned, with documentation available in the change log on the releases page. The speaker expresses particular interest in the experimental Flux model and guides viewers on how to manually download and configure it within Creta AI.

05:00

🚀 Using the Flux Model in Creta AI

The second paragraph focuses on the practical use of the Flux model within Creta AI. The speaker notes that the model is resource-intensive, suggesting a tip to speed up generation time by adjusting NVIDIA settings. After downloading and configuring the Flux model, the speaker demonstrates its use with various prompts. However, due to the model's experimental status, some features like 'fill' and 'refine' do not work well with Flux. The speaker advises using SD 1.5 or XL model styles for these features and shows an example where the 'Cinematic photo' style effectively removes a person from an image. The paragraph concludes with a thank you to viewers and an invitation to watch the next video.

Mindmap

Keywords

💡Krita.Ai Plugin

The Krita.Ai plugin is a tool integrated into the Krita software that allows users to utilize AI models for tasks such as generative fill. The video focuses on updating this plugin to its latest version, ensuring access to the most current features and models, like the Flux model.

💡Flux Model

The Flux model is a new, experimental AI model that can be used within the Krita.Ai plugin for generative tasks. It is resource-intensive, requiring significant VRAM, and the video demonstrates how to download and configure this model within Krita.Ai, despite some limitations in its current experimental stage.

💡Generative Fill

Generative fill refers to the process of using AI to generate content that fills or replaces parts of an image. The video discusses how to utilize different AI models in Krita.Ai for generative fill, particularly highlighting the challenges and benefits of using the Flux model for this purpose.

💡Manual Update

Manual update refers to the process of updating the Krita.Ai plugin manually, as it does not support automatic updates. The video guides users through the steps of checking their current version, downloading the latest version, and reinstalling the plugin to complete the update.

💡Server Settings

Server settings in Krita.Ai are configurations related to the plugin's connection and operation. The video advises users to copy these settings before updating from an older version to avoid losing important configurations during the manual update process.

💡Installation Instructions

Installation instructions are the steps provided by the Krita.Ai documentation to properly install or update the plugin. The video emphasizes following these instructions carefully, especially when updating from older versions, to ensure a smooth upgrade.

💡Configuration Window

The configuration window in Krita.Ai is where users adjust settings for the plugin, including selecting AI models and managing server components. The video highlights how to navigate this window to fix errors and enable the Flux model after updating the plugin.

💡Model Folder

The model folder is the directory where AI models like Flux are stored for use in Krita.Ai. The video instructs users on how to download the Flux model and move it to this folder, enabling its use within the plugin.

💡VRAM

VRAM (Video Random Access Memory) is a type of memory used by graphics cards to store image data for fast access. The video mentions the high VRAM requirements for running the Flux model effectively in Krita.Ai, offering a tip to adjust Nvidia settings to improve performance.

💡Experimental Phase

The experimental phase refers to the testing stage of the Flux model in Krita.Ai, where the model is still being refined and may not work perfectly. The video acknowledges this status, warning users that certain features may not function smoothly and advising them on how to work around these issues.

Highlights

How to manually update the Krita AI plugin

Latest version supports the new Flux model

Upgrade instructions can be found on the website

Warning for users on version 1.15 or older to copy server settings

Checking current Krita AI version in settings

Reinstalling to upgrade if on version 1.17 or newer

Downloading the latest Krita AI plugin from the releases page

Installing the plugin via the 'Import Python Module' menu in Krita

Server out of date message upon opening Krita AI panel

Fixing server issues by installing missing components

New features in the updated plugin: multi-language support, tag completion

Flux model support is experimental and may not work smoothly

Downloading the Flux model for testing

Setting the style to Flux Chel in plugin configuration

Moving the downloaded Flux model to the models folder

Refreshing the configuration to recognize the new model

Restarting Krita to resolve experimental phase warnings

Optimizing generation time by adjusting Nvidia settings

Using the Flux model for generative fill in Krita AI

Limitations of the Flux model in experimental phase for Fill and Refine

Using SD 1.5 or XL model styles for Fill and other features

Cinematic photo style effectively removes unwanted elements

Thank you note and sign-off for the video