Inpainting in Midjourney: A Deep Dive, Secret Commands & Tips for the Long Awaited Feature!

In this video, Tim dives into the newly released inpainting feature in Midjourney, a tool used for restoring or altering parts of images. He explains how inpainting works by highlighting an area and re-prompting within it, allowing users to seamlessly edit and restore images. Tim covers the limitations, such as the feature's incompatibility with panned images, and shares secrets and tricks he discovered. He demonstrates the inpainting process with various examples, showing how it can be used to change character details, emotions, and add props. Tim also discusses the 'slider method' for when inpainting doesn't respond as expected. He concludes by highlighting the tool's potential and invites viewers to share their own tips and experiences with inpainting.

Takeaways

• 😀 Inpainting in Midjourney allows users to restore or modify specific areas of an image by highlighting and reprompting within those areas.
• 🔍 The 'very region' button is crucial for inpainting in Midjourney, offering a way to reimagine certain parts of the canvas based on the surrounding image context.
• 🖌️ Tools like free draw and rectangle are available for selecting areas to inpaint, though the free draw tool can be finicky.
• 🔄 The inpainting feature works best with upscaled images and should cover 20% to 50% of the image for optimal results.
• 🎭 Inpainting can be combined with zooming for better control over the image's detail and to recontextualize subjects while maintaining the background.
• 👁️ Using inpainting, users can modify physical attributes, emotions, or add props to characters within the images.
• 🎨 The inpainting results are influenced by the specific prompts given, and it may take multiple attempts to achieve the desired outcome.
• 🚫 Certain limitations exist, such as the inability to inpaint after panning or the occasional non-responsiveness to complex prompts.
• 📚 Midjourney documentation suggests that other commands like 'chaos', 'stylize', and 'weird' can be used during inpainting to influence the results.
• 🆕 Users can expect future enhancements to inpainting in Midjourney, potentially including image prompting and weighting adjustments.

Q & A

What is inpainting in the context of the video?

Inpainting refers to the technique of restoring damaged or deteriorated parts of images or video. In the context of the video, it's a feature in mid-journey that allows users to highlight an area and re-prompt within that area, enabling the system to reimagine the scene without additional user input.

What are the limitations of inpainting mentioned in the video?

The video mentions that inpainting works best when covering a region of 20 to 50% of the image, although smaller areas can also yield good results. Additionally, inpainting only works with upscaled images and cannot be used after panning, which might be a temporary glitch.

How does the inpainting feature handle lighting in images?

The inpainting feature handles lighting quite impressively, as it manages to maintain the overall lighting and composition within the frame, even though it may not be completely accurate in terms of motivated light.

What is the 'slider method' mentioned in the video?

The slider method is a technique where a colon colon is used to repeat part of the phrase in the prompt. This method is used to split the prompt and can help achieve desired results in inpainting when the feature doesn't respond as expected to a single prompt.

How does inpainting work with different styles?

Inpainting can be combined with different styles by taking two photographers' styles and mashing them together, creating a hybrid of the two styles, which can yield interesting results.

What are some of the commands available for use with inpainting?

Some of the commands that work with inpainting include 'chaos', 'stylize', 'style', and 'weird'. However, 'image waiting' and 'version' do not work, and 'weird' should be used with caution as it can produce very unusual results.

How can inpainting be used to change a subject's ethnicity or emotions?

Inpainting can be used to change a subject's ethnicity or emotions by issuing a new prompt that specifies the desired characteristic, such as 'Korean man' or 'happy'. The system will then attempt to adjust the image accordingly.

What is the 'remix mode' and why is it important for inpainting?

Remix mode is a setting in mid-journey that, when turned on, allows users to issue specific prompts for certain areas of the canvas. It is important for inpainting as it enables the feature to work by providing the system with instructions on what to reimagine within the selected area.

What are the tools available for selecting areas in inpainting?

The tools available for selecting areas in inpainting include the 'free draw' rectangle tool and the 'lasso' tool, which can be used to highlight specific areas of the canvas for the system to reimagine.

How does inpainting handle adding props to a scene?

Inpainting can add props to a scene by including the desired prop in the prompt. For example, adding 'holding a red balloon' to the prompt resulted in the subject holding a red balloon in the inpainted image.

What is the significance of the 'undo' button in inpainting?

The 'undo' button is a crucial feature in inpainting that allows users to revert to the previous state of the image if they are not satisfied with the results of a particular prompt or if an error occurs.

How can the inpainting feature be used to fix common issues like extra limbs in generated images?

The inpainting feature can be used to fix issues like extra limbs by highlighting the area with the unwanted element and issuing a new prompt that excludes the undesired feature, allowing the system to generate a corrected image.

Outlines

00:00

🎨 Introduction to In-Painting Feature

This paragraph introduces the new in-painting feature, which is a technique used to restore or enhance parts of images or videos. The feature allows users to highlight an area and re-prompt within that area, making it easier to correct unwanted elements like extra limbs. The speaker shares their excitement about the feature's potential and mentions some limitations and secrets they've discovered. They also explain that in-painting works best with upscaled images and requires remix mode to be enabled for more specific prompts. The paragraph concludes with a simple demonstration of the feature using a prompt to generate an image of Steve Aoki's dad with a seductive third wife.

05:01

🖌️ In-Painting Techniques and Tools

The second paragraph delves into the technical aspects of in-painting, discussing how it handles lighting and composition. It covers the use of different tools like the rectangle and lasso tools for selecting areas to be inpainted. The speaker demonstrates changing a subject's pose, expression, and adding props like a red balloon to alter the narrative of an image. They also mention the challenges of using the lasso tool and the importance of the undo button for trial and error. The paragraph includes examples of successful in-painting, such as changing a character's ethnicity or emotions, and the limitations when in-painting ignores certain prompts. It concludes with a discussion on the slider method as a workaround for prompts that aren't initially successful.

10:01

🔄 Advanced In-Painting Strategies and Commands

The final paragraph explores advanced strategies for in-painting, including the slider method for adding elements that initially fail to appear. It also discusses the use of various commands like 'chaos', 'stylize', 'style', and 'weird' to achieve different effects, with a caution that 'weird' can produce very unusual results. The speaker provides examples of how these commands can be used, such as adding a clown and an alien to a scene with significant changes to the background elements. They highlight the impressive ability of in-painting to adjust background details to accommodate new elements. The paragraph concludes with an invitation for viewers to share their thoughts and any tips or tricks they've discovered regarding the in-painting feature.

Keywords

Inpainting

Inpainting is a technique used to restore damaged or deteriorated parts of images or videos. In the context of the video, it refers to a feature in mid-journey that allows users to highlight an area of an image and re-prompt within that area, enabling the system to reimagine the selected region based on contextual clues from the surrounding image.

Mid-journey

Mid-journey is the software or platform being discussed in the video, which offers an inpainting feature. It is used to create or modify images based on user prompts. The video dives into how to use this feature effectively, its limitations, and the fun that can be had with it.

Remix Mode

Remix mode is a setting within mid-journey that, when turned on, allows users to make specific changes to parts of an image using the inpainting feature. It is essential for accessing more targeted modifications without affecting the entire image.

Upscaled Images

Upscaled images are those that have been enlarged or 'scaled up' from their original size, often using algorithms to maintain or improve quality. The video mentions that inpainting in mid-journey works best with upscaled images, indicating the importance of resolution for the feature's effectiveness.

Free Draw Tool

The free draw tool is a feature within the inpainting module of mid-journey that allows users to manually select areas of the image to be reimagined. It is described as being a bit finicky, suggesting it may require practice to use effectively.

Lasso Tool

The lasso tool is another selection tool used in the inpainting process. It allows for more precise, freeform selection of areas in the image. The video demonstrates its use in changing specific parts of an image, such as adding a red balloon to a character's hand.

Zooming Out

Zooming out is a technique used in conjunction with inpainting to provide more context for the AI to work with. It can help in achieving better results by giving the system a broader view of the image, which can improve the accuracy of the inpainting.

Chaos and Weird

Chaos and weird are commands that can be used in mid-journey to introduce more randomness or unusual elements into the inpainting process. These commands can lead to unexpected and creative results, although they may require careful use to achieve the desired outcome.

Niji

Niji refers to a specific style or feature within mid-journey that allows for the creation of images with a particular aesthetic. The video shows how inpainting can be used in conjunction with Niji to modify elements of an image, such as changing a character's pose or adding new elements like ninjas.

Slider Method

The slider method is a technique mentioned in the video for dealing with the limitations of the inpainting feature. It involves repeating parts of the prompt with additional specifications to guide the AI more effectively towards the desired outcome, such as adding a white wolf walking next to a character.

Post-Processing

Post-processing refers to the editing and adjustments made to an image after it has been generated by mid-journey. The video suggests that users may need to use an image editor to clean up or blend parts of the image that did not inpainting correctly, such as removing smudged background details.

Highlights

Inpainting, a long-awaited feature, has arrived in Midjourney, allowing users to restore or alter parts of images.

Inpainting can fix issues like extra limbs in generated images, making it easier to edit and perfect the final output.

The 'Very Region' button is the primary tool for inpaintings, allowing users to select areas for Midjourney to reimagine.

Remix mode must be enabled for specific inpaintings, which can be accessed through the forward slash settings.

Inpainting works best with upscaled images and can cover regions as small as 20 to 50 percent of the image.

The feature can be combined with zooming out for more precise control, although it cannot be used after panning.

Inpainting can change character details, such as ethnicity and emotions, and add props to the scene.

The tool can handle complex tasks like adjusting lighting and composition to match the original image.

Inpainting can be finicky, especially with the free draw and lasso tools, which may require practice to master.

The slider method can be used as a workaround to achieve desired outcomes when inpaintings fail to respond.

Commands like 'chaos', 'stylize', 'style', and 'weird' can be used to influence the outcome of inpaintings.

Inpainting is powerful for recontextualizing subjects while maintaining the original background.

The feature can also work with niji, allowing for creative modifications in different styles.

Inpainting can sometimes ignore user prompts, but persistence and experimentation can yield desired results.

The tool is capable of impressively adjusting reflections and details to accommodate new elements in the image.

Inpainting is a significant advancement in image editing, offering a range of creative possibilities.

The presenter, Tim, encourages viewers to share their tips and tricks for using the inpaintings feature effectively.