Krita AI Trick to Removing Objects With AI diffusion Plugin

Streamtabulous
12 Dec 202321:31

TLDRIn this informative video, the creator discusses a method to remove objects from images using the Krita AI diffusion plugin. The process involves manipulating AI diffusion with a static image to build unique images, and uploading a file to the cloud for users to download and utilize. The video provides a step-by-step guide on how to use the static diffusion pattern to effectively remove unwanted elements from a photo, highlighting the benefits of using open-source tools like Krita over paid options like Adobe.

Takeaways

  • 🎨 The video discusses a method to remove objects from images using Krita's AI diffusion plugin.
  • 🔍 Krita lacks a dedicated AI background removal tool, but the AI diffusion can be manipulated to achieve similar results.
  • 🌐 The presenter uploads a static diffusion file to the cloud for viewers to download and use with the AI.
  • 🖼️ The AI diffusion works by creating a static image and using it to build a unique image each time, sometimes getting caught in loops.
  • 📱 The presenter explains how to install the static diffusion file into the Krita app by accessing the data roaming CER folder.
  • 💻 The presenter's system is old and slow, using an old GTX 1070, and discusses the limitations when compared to Adobe's AI capabilities.
  • 🎨 The presenter prefers Krita's AI diffusion due to the ability to select models, live preview, and the graphic user interface.
  • 🚀 The presenter demonstrates how to remove a vehicle from an image by using the static diffusion pattern and adjusting settings.
  • 🌳 The AI's ability to render natural elements like trees and bushes is discussed, with a focus on Australian trees.
  • 🔧 The presenter shares tips on adjusting padding and saturation for better blending and realistic results.
  • 💡 The presenter encourages viewers to explore different models on CD AI for tasks like photo restoration or nature rendering.
  • 🤝 The video ends with a call to action for viewers to support open-source projects and provide feedback to plugin creators.

Q & A

  • What is the main topic of the video?

    -The main topic of the video is a tutorial on how to use the AI diffusion plugin in Krita to remove objects from images.

  • What issue does the video address with Krita's AI background removal tool?

    -The video addresses the issue that Krita doesn't have a decent AI background removal tool, and the creator explores ways to manipulate the AI to achieve the desired effect.

  • What is the purpose of the static image uploaded by the creator?

    -The static image uploaded by the creator is used as a base for the AI diffusion process, helping to manipulate the AI into creating a unique image without the unwanted object.

  • What hardware limitation does the creator mention?

    -The creator mentions that they are using an old GTX 1070, which limits their ability to use the AI diffusion tool at full resolution.

  • How does the creator plan to handle the limitation of the AI model's training on specific image sizes?

    -The creator plans to handle this by using a lower resolution for the AI model rendering and then upscaling the image afterward, taking advantage of AI upscaling to maintain detail.

  • What is the advantage of using the AI diffusion tool in Krita over Adobe, according to the creator?

    -The advantage of using the AI diffusion tool in Krita over Adobe is the ability to select different models, the live preview feature, and the fact that it is free and open source.

  • How does the creator suggest using the static diffusion pattern?

    -The creator suggests using the static diffusion pattern as a base layer in Krita, then adding a new paint layer and flooding it with static colors to help the AI generate a new image without the unwanted object.

  • What is the creator's opinion on Adobe's AI model training?

    -The creator believes that Adobe's AI model training has a better language model associated with the diffused images, which gives it an advantage in understanding user input like the word 'remove'.

  • What does the creator recommend for users looking to improve their AI diffusion results?

    -The creator recommends users look for models suited to their specific artwork, experiment with different settings like padding and saturation, and provide positive feedback to the creators of plugins to encourage further development.

  • How can users access the static pattern file mentioned in the video?

    -Users can access the static pattern file by downloading it from the link provided in the video description and placing it in their Krita data folder's patterns directory.

Outlines

00:00

🎥 Introduction to AI Manipulation in Stream Tabulous

The speaker introduces the topic of using AI in the video series, specifically focusing on manipulating AI diffusion to achieve desired outcomes. They mention a previous video about adding elements like a cat to an image and express intent to explore object removal from backgrounds. The speaker acknowledges their lack of formal AI training but is eager to experiment with the tools available. They introduce a method involving a static image that the AI uses to build unique images, likening it to 'magic eye' pictures. The speaker also discusses the limitations of their system, an old GTX 1070, and the advantages of Adobe's AI model, while expressing a preference for Cryer diffusion due to its customizability and open-source nature.

05:01

🖼️ Adjusting Resolution and Removing Objects

The speaker discusses the process of adjusting the image resolution to accommodate their system's limitations and shares their intent to remove a vehicle from the image. They explain the challenges of using the AI to ignore selected objects and the need to experiment with different settings. The speaker introduces a new tool, 'static diffusion,' which they have uploaded to their cloud for viewers to download. They explain that this tool allows users to add a static image to influence the AI's output, and they guide viewers on how to integrate this tool into their Cryer data file. The speaker also compares the AI diffusion process to painting, where the AI generates static to create images.

10:02

🌳 Blending and Selection Adjustments

The speaker shares their experience with previous attempts to add and change objects in images, highlighting the importance of adjusting padding and blending settings for optimal results. They demonstrate how the AI successfully blends the modified image with the original surroundings, despite some initial issues. The speaker also discusses the benefits of using Cryer diffusion over Adobe, particularly the ability to select models suited to specific art styles and the cost-saving aspect of using open-source tools. They emphasize the potential of AI models to improve with time and encourage viewers to explore different models for various tasks, such as photo restoration or nature scenes.

15:05

💧 Experimenting with Reflective Water

The speaker explores the possibility of creating reflective water in images using AI, noting Adobe's proficiency in this area. They express curiosity about the outcome when using Cryer diffusion for such tasks. The speaker shares their experience with reflective water using comfy AI and discusses the impact of different workflows on the final image. They also mention their intention to test this method on buildings and invite viewers to share their experiences. The speaker concludes by providing a link to download a static pattern for use in Cryer, reiterating their belief in the potential of open-source AI tools and the value of community feedback.

20:07

🚀 Encouraging Open Source AI Tools

The speaker expresses excitement about the potential of open-source AI tools, particularly Cryer, and encourages viewers to support the creators of such tools. They mention the importance of positive feedback in motivating developers to improve and expand these tools. The speaker also highlights the benefits of having a diverse range of plugins and tools available, suggesting that community involvement could lead to even more innovative solutions. They conclude the video by encouraging viewers to like, share, and subscribe to support their content and help it reach a wider audience.

Mindmap

Keywords

💡AI diffusion

AI diffusion refers to a process in which artificial intelligence algorithms are used to create or manipulate images by generating and altering pixels. In the context of the video, AI diffusion is employed to either add or remove objects within an image, leveraging the ability of AI to understand and recreate visual elements based on patterns and data it has been trained on.

💡Krita

Krita is a free and open-source digital painting and illustration software. It provides a variety of tools that allow artists to create and edit images. In the video, Krita is used as the platform to apply AI diffusion techniques, specifically with the help of plugins to achieve desired effects like object removal.

💡Cryer

Cryer, as mentioned in the video, appears to be a software or a plugin related to AI diffusion that works in conjunction with Krita. It is used to select and manipulate models and patterns within the AI diffusion process, allowing for more control over the final image.

💡Static image

A static image is a fixed and unchanging visual representation, as opposed to a dynamic or moving image. In the context of the video, a static image serves as a base or starting point for the AI diffusion process, where the AI generates a unique image based on the static image's content and structure.

💡Patterns

In the context of the video, patterns refer to the visual structures or elements that are used within the AI diffusion process. These can be static images or other types of design elements that guide the AI in generating the final image. Patterns are crucial in determining how the AI will interpret and alter the content of the image.

💡Resolution

Resolution refers to the number of pixels in an image, which determines its level of detail and clarity. Higher resolutions have more pixels and thus can display more detail, while lower resolutions have fewer pixels and appear less sharp. In the video, the streamer discusses adjusting the resolution of the image to accommodate for their system's processing capabilities.

💡Adobe

Adobe is a company known for its creative software suite, which includes tools like Photoshop for image editing. In the video, Adobe is mentioned as having certain advantages in its AI models and rendering capabilities, particularly in handling higher resolutions and having a more sophisticated language model for understanding user inputs.

💡SDXL models

SDXL models refer to a type of AI model that is trained on higher resolution images (such as 2K and above). These models are designed to handle more detailed and complex images, providing better results for tasks like image manipulation and restoration. The video suggests that future SDXL models will offer improved performance and capabilities.

💡Reflective water

Reflective water in image editing refers to the simulation of water surfaces that reflect their surroundings. This effect can be challenging to achieve convincingly, as it requires the AI or editing tool to accurately interpret and replicate the way light and color are reflected off the water's surface. In the video, the streamer expresses curiosity about whether the AI diffusion tool can effectively create reflective water effects.

💡Open source

Open source refers to software or tools whose source code is made publicly available, allowing anyone to view, use, modify, and distribute the software freely. The video highlights the benefits of open-source tools like Krita and the AI diffusion plugin, which can be customized and improved by the community.

Highlights

The video discusses a method for removing objects from images using Krita's AI diffusion plugin.

The AI diffusion tool in Krita can be manipulated to achieve specific image alterations.

A static image is used as a base for the AI to build unique images, which can be exploited for object removal.

The presenter uploads a static diffusion file to the cloud for viewers to download and use with the AI tool.

Krita's AI diffusion plugin allows users to select different models and utilize a live preview feature.

The presenter prefers Krita's AI diffusion over Adobe due to the ability to select models and the open-source nature of the tool.

The presenter demonstrates how to adjust the resolution of the image to accommodate older hardware.

The AI tool struggles with background object removal when the object is selected and painted over.

A new method is introduced using a static diffusion pattern fill to trick the AI into removing background objects.

The presenter explains the importance of adjusting settings such as padding and blending for better AI-generated results.

The AI diffusion tool can create realistic images based on the static pattern provided, even without specific object selection.

The presenter suggests using different models in Krita depending on the style of artwork being worked on.

The video highlights the advantage of using free, open-source tools like Krita's AI diffusion over paid alternatives.

The presenter encourages viewers to support the developers of the AI diffusion plugin and open-source community.

The video concludes with a call to action for viewers to share the video and support the use of open-source AI tools.