Krita AI diffusion, A look at v1.15 update, bugs on older hardware and new cool features
TLDRThe video discusses the AI diffusion 1.15 update for Krita, highlighting its powerful new features that rival Adobe Photoshop's generative fill. It provides an update process, showcases the ability to edit artwork with more control, and discusses bugs encountered on older hardware. The creator also mentions a new lightning model from Civ AI and shares workarounds for certain issues.
Takeaways
- 🎨 The AI diffusion 1.15 update by akley for Cryer introduces many new features, making it more powerful and comparable to Adobe Photoshop's generative fill.
- 🔄 Users can now edit their artwork with more control, which helps prevent glitching and allows for better manipulation of the generated content.
- 📚 The update includes the ability to copy prompts and a suggestion for future versions to include a feature to save prompts as text files.
- 🔧 A new feature allows users to discard rendered parts they don't like by right-clicking, providing more control over the final output.
- 🔍 Cryer tools and Auto selection with outlining are mentioned as handy tools for users who may not have them, enhancing the workflow.
- 🌟 Civit AI's new furnace Model, specifically the lightning model, is highlighted for its ability to render both fantasy creatures and humans effectively.
- 📈 The update adds new core components to the plugin, improving its performance and offering better results.
- 🖌️ A new feature called 'expand' is introduced, which takes the entire image into account rather than just the prompt, leading to more controlled and relevant image generation.
- ⚙️ The user interface now includes options to adjust the diffusion settings, allowing for better control over how the AI paints outside the selection area.
- 🐞 The video discusses bugs encountered when using the update on older hardware, such as the GTX 1070, and suggests workarounds like adding a 'magic dot' to resolve certain errors.
- 🆕 Despite the bugs, the update is praised for its new capabilities and potential for photo restoration and digital artwork, offering more power and options to users.
Q & A
What is the main focus of the Krita AI diffusion 1.15 update by akley for Cryer?
-The Krita AI diffusion 1.15 update focuses on introducing new features that make the software more powerful and versatile, allowing users to edit their artwork in ways that were not possible before. The update brings it on par with Adobe's Photoshop for generative fill, but with more control to prevent glitching.
How can users update their Krita AI diffusion to version 1.15?
-To update Krita AI diffusion to version 1.15, users can visit akley's GitHub page, download the zip file, extract it, and overwrite the existing files. Upon opening Krita, the software should prompt the user to update the configuration for the AI, ensuring all new features are included.
What is the significance of the new core components introduced in the 1.15 update?
-The new core components in the 1.15 update enhance the software's capabilities, making it better than ever before. These improvements allow for more precise editing and control over the generative AI, resulting in better artwork manipulation and fewer glitches.
What is the purpose of the 'discard' feature in the 1.15 update?
-The 'discard' feature enables users to easily reject a rendered part of their artwork that they are not satisfied with. By right-clicking on the unwanted element, users can discard it and make changes as needed.
How does the 'copy prompts' feature work in the 1.15 update?
-The 'copy prompts' feature allows users to duplicate the prompt used for generating a model. This is useful for making quick adjustments or reusing the same prompt for multiple iterations without having to retype it.
What is the 'furnace model' mentioned in the script, and how does it perform?
-The 'furnace model' is a newly released model by Civ AI, designed for generating fantasy creatures. However, it has been found to perform exceptionally well with human subjects as well, offering users a versatile tool for various rendering tasks.
What samplers are currently available in the 1.15 update, and what is the user's opinion on them?
-In the 1.15 update, the available samplers include DPM Plus+, SD, and SGM. The user expresses a desire for additional samplers, such as a 3M version, but acknowledges that the current samplers are well-implemented and likely optimized for the in-painting workflows.
How does the 'expand' feature in the 1.15 update work, and what are its benefits?
-The 'expand' feature allows users to extend an image while considering the entire context, rather than just the prompt information. This results in a more controlled and accurate generation of content, avoiding random or mismatched elements in the expanded areas.
What issue does the user encounter when using the 'face' tool in the 1.15 update?
-The user experiences a runtime error when using the 'face' tool in the 1.15 update. The error message indicates a mismatch between expected data types, which the user suspects is related to memory allocation issues previously fixed in another software version.
How does the user propose to address the bugs found in the 1.15 update?
-The user suggests adding arguments to the configuration file to fix the memory allocation issues, similar to a fix applied in a previous version of another software. They plan to reach out to the developer, suspicious Firefly, on Reddit to report the issue and hope for a resolution.
Outlines
🚀 Introduction to AI Diffusion 1.15 Update
The video begins with an introduction to the AI Diffusion 1.15 update by akley for Cryer, highlighting the new features and improvements that make it more powerful than ever. The update is compared to Adobe's Photoshop in terms of generative fill, but with the added benefit of user control to prevent glitching. The video promises a deeper dive into these features after the intro and mentions a link to a tutorial on how to update the software.
🛠️ Exploring AI Diffusion Features and Samplers
The second paragraph delves into the specifics of AI Diffusion's features, such as auto-selection with outlining and the various samplers available. The speaker expresses a desire for additional samplers like DPM Plus+ and 3M, and discusses the potential reasons for their absence. The paragraph also touches on the lightning model by Civ AI, its applications for both fantasy creatures and humans, and the sharing of prompt information by users for testing purposes.
🎨 Customizing and Enhancing Image Edits
This paragraph focuses on the customization options provided by the AI Diffusion update, such as discarding unwanted renders and copying prompts for regenerating models. The speaker appreciates the new core components and the ability to control the generative process, including the selection of samplers and the adjustment of settings for better image generation. The paragraph also discusses the practical application of these features in editing and enhancing images.
🐞 Addressing Bugs and Workarounds
The speaker discusses encountering bugs in the AI Diffusion 1.15 update, particularly when working with the face tool and the expand feature. They share a workaround involving the addition of a 'magic dot' to the image to trick the AI into generating the desired expansion. The paragraph also touches on the challenges of using older hardware and the hope for future updates to fix these issues.
🌟 Appreciating the Power of AI Diffusion 1.15
In the final paragraph, the speaker reflects on the overall positive impact of the AI Diffusion 1.15 update, despite the presence of some bugs. They express gratitude for the ability to use older hardware and hope that future updates will address the issues encountered. The speaker also encourages viewers to share their experiences with different hardware and to provide feedback on the software's performance.
Mindmap
Keywords
AI diffusion
Krita
Update 1.15
Bug
Older Hardware
Generative Fill
Cryer
GitHub
Auto-selection
Out-painting
Prompt
Highlights
The AI diffusion 1.15 update by akley for Cryer introduces a plethora of new features, enhancing the software's capabilities.
The update allows for better editing and utilization of artwork, rivaling Adobe's Photoshop in generative fill but with more control to prevent glitching.
GitHub is the platform where users can find and download the AI diffusion update, with instructions provided for easy installation.
Civ AI has released a new lightning model designed for rendering fantasy creatures, but it also performs well with human subjects.
The new update includes the ability to discard rendered elements and copy prompts, streamlining the workflow for artists.
The implementation of the 1.15 model has resolved the issue of finding it, Styles lightning, without needing an update.
Samplers like DPM Plus+ SD and SGM are used in the background for better in-painting, though they are not directly accessible to users yet.
The lightning sampler is recommended for use with the lightning model to ensure optimal performance and simplicity.
The ability to select areas and generate custom content within the selected region is a significant addition to the AI diffusion tool.
The update includes a feature to expand images while considering the entire context, leading to more coherent and accurate generative fills.
Users can now adjust the diffusion settings to control how much the AI paints outside of the selection for a better blending effect.
The AI diffusion 1.15 update has introduced some bugs, particularly on older hardware like the GTX 1070.
A workaround for the generative expand bug involves placing a dot in the area to be expanded, tricking the AI into processing the request correctly.
Despite the bugs, the AI diffusion 1.15 update is a powerful tool that offers more capabilities than previous versions, especially for photo restorations.
The new features allow for greater control and precision in editing, such as selecting and removing content or adding specific details like gold coins.
The lightning models from Civ AI have significantly improved rendering speeds, allowing for faster and higher quality outputs.
The video creator plans to reach out to the developers to address the bugs found, hoping for continued support and improvements for older hardware.
The AI diffusion tool's advancements are particularly beneficial for artists and those in photo restoration, providing new levels of control and creative possibilities.
Community feedback and real-world testing are crucial for the developers to understand and fix issues, as well as to further enhance the software.