DRAW THINGS: New in 2024
TLDRThe video from January 2024 introduces the new features of 'Draw Things' for the Year of the Dragon. It focuses on the revamped user interface, catering to new and casual users, and promises in-depth tutorials for advanced features in future videos. The update includes new add-ons such as low-step models, an IP adapter, a QR code generator, and stable video diffusion. The project manager has been overhauled, creating a database for each project that stores history, prompts, and images. The interface now allows for easy project management and deletion of old files. Users can also export projects to external drives or transfer them between devices. The video also covers the new thumbnail sidebar, advanced settings, and the use of LoRA, ControlNets, and style templates. It concludes with tips on using the preview feature to gauge the progress and quality of image generation, advising viewers on when to abort or adjust the rendering process for optimal results.
Takeaways
- ๐จ **New User Interface**: Draw Things has a refreshed user interface in 2024, designed to be more intuitive for new and casual users.
- ๐ **Automatic Optimization**: The AI chip icon in the machine settings allows Draw Things to automatically optimize for your system, though manual overrides are available.
- ๐ **Project Manager Enhancement**: A new project manager feature creates a database for each project, storing history and images, addressing previous issues with root folder accessibility.
- ๐๏ธ **Storage Management**: Users can now more efficiently manage storage by deleting old images and projects, and can perform a deep clean to vacuum the database.
- ๐ **Exporting and Transfer**: Projects can be exported to external drives, moved between devices, and even transferred to an iPad.
- ๐ผ๏ธ **Thumbnail Sidebar**: A new thumbnail scrolling sidebar provides a visual history of a project, allowing for easy deletion or export of images.
- ๐ **Prompts and Tokens**: The script introduces new ways to manage prompts with linebreaks and tokens that reflect grouping, offering more control over the generation process.
- ๐ญ **Style Templates**: Users can apply style templates to their prompts for a more streamlined creative process, with the option to customize and save their own styles.
- ๐ **Image Interrogater**: An image interrogator tool is introduced for basic analysis, with potential for a more advanced version in the future.
- ๐ **Menu Rail Customization**: The menu rail has been updated with new tools and the potential for further evolution, including the training for custom LoRA and textual inversions.
- โ๏ธ **Parameter Settings**: The parameter sidebar has been expanded with new settings and options for model management, LoRA, control net, and other advanced features.
Q & A
What is the significance of the new user interface in Draw Things for the Year of the Dragon in 2024?
-The new user interface in Draw Things for 2024 is designed to be more user-friendly, particularly for new and casual users. It includes a refreshed look and automatic optimization for the user's system, making it easier to navigate and manage projects.
What are some of the new add-ons introduced to Draw Things in the last couple of months?
-Some of the new add-ons introduced to Draw Things include low-step models, IP adapter, QR code generator, and stable video diffusion. These additions are set to receive dedicated feature and workflow videos.
How does the Project Manager feature in Draw Things work?
-The Project Manager in Draw Things creates a database for each project, storing the entire history of prompts and images in the user's root drives Library folder. It allows users to manage their projects more effectively, including deleting old projects and images to free up storage.
What is the purpose of the new thumbnail scrolling sidebar in Draw Things?
-The new thumbnail scrolling sidebar in Draw Things provides a more detailed view of the project's history. It allows users to right-click to delete or export images and offers options to sort the thumbnails chronologically, by link, or to hide working images.
How can users customize their prompts in Draw Things?
-Users can customize their prompts in Draw Things by partitioning them with linebreaks, using round and square brackets to adjust the weight of words, and utilizing style templates. They can also attach textual inversions and organize prompts by themes.
What is the role of the LoRA training in Draw Things?
-LoRA (Low-Rank Adaptation) training in Draw Things allows users to train custom LoRA models for more personalized and specific image generation. It is a separate feature that has its own video walkthrough for detailed instructions.
How does the 'moodboard' tool in Draw Things function?
-The 'moodboard' tool in Draw Things is a container for multiple images. Users can load images into the moodboard and then adjust the individual influence weight of each image to mix their contributions to the final image.
What are the different ways to save and export images in Draw Things?
-In Draw Things, users can save and export images through the export and share buttons at the top right. They can choose detailed save options, including copying to the clipboard, saving just the visible canvas, and deleting models or compressing new models to 8-bit.
How does the parameter settings sidebar in Draw Things help users control the image generation process?
-The parameter settings sidebar in Draw Things provides controls for stable diffusion, allowing users to save and manage configurations, select models, adjust LoRAs, control net parameters, and access basic and advanced settings. It also includes options for seed, image size, text guidance, and batch size.
What is the significance of the preview image during the image generation process in Draw Things?
-The preview image during the image generation process in Draw Things is not just a preview; it provides important information about the step count and whether the image generation is overcooking or undercooking. Users can abort the render if it's not developing as desired based on the preview.
What are some of the advanced settings and parameters available in Draw Things for specific models?
-Advanced settings in Draw Things include Samplers designed for new Fast models, specialized parameters for specific models like SDXL and stable video diffusion, and options for face restorer, high-resolution fix, upscaler models, and mask feathering.
How does the HTTP API server feature in Draw Things work?
-The HTTP API server in Draw Things can be toggled on and off from the interface. While the script does not provide specific details on its functionality, it suggests that it may be used for remote or automated interactions with the Draw Things application.
Outlines
๐จ Introduction to the New User Interface
The video script introduces the audience to the new user interface for 'Draw Things' in the Year of the Dragon, 2024. It is aimed at new and casual users who might need a refresher. The presenter plans to cover the interface in detail over the next 25 minutes and mentions new add-ons like low-step models, IP adapter, QR code generator, and stable video diffusion. The video will also touch upon the machine settings, UI settings, and the importance of managing projects and cleaning the database.
๐ Project Manager and Interface Features
The script explains the new project manager feature, which creates a database for each project and stores history, prompts, and images. It discusses the issue with hidden root folders on OS 10 and how the project manager changes the workflow. The presenter also talks about exporting and moving projects, deleting empty models, and the context menu changes based on the selected project. The video continues with the thumbnail scrolling sidebar, which shows the project's history, and the ability to delete or export images. It also covers the image interrogator, basic and advanced settings menus, and the Javascripts menu.
๐๏ธ Canvas Tools and In-Painting Techniques
The paragraph delves into the various tools available around the canvas, such as the magic wand, eraser, and paintbrush, which are used for in-painting and other image editing tasks. It also discusses the parameter settings in the sidebar, which control stable diffusion, and the ability to save and manage configurations. The script mentions the process of pasting from the clipboard, loading images, and using canvas layers like scribble, pose, color palette, and mood board with controlNet models. The moodboard tool is highlighted for its ability to mix multiple images and adjust their influence on the final render.
๐ Model Management and Parameter Customization
The script provides an overview of how to manage models in 'Draw Things,' including downloading, compressing, and mixing models. It also covers the LoRA (Latent Diffusion Models) and control net parameters, which allow users to stack and adjust the influence of different models on the final image. The paragraph explains the process of toggling between text-to-image and image-to-image modes and the importance of text guidance strength. It concludes with a discussion on basic settings like seed, image size, and text guidance, as well as advanced settings like samplers and model-specific parameters.
๐ Batch Generation and Advanced Interface Features
The final paragraph focuses on the generation process, including the power button to generate images and the ability to generate up to 100 images at a time with scripting. It discusses the advanced menu, which houses specialized parameters for different models, and the various settings that can be toggled on and off, such as face restorer, high-resolution fix, and upscaler models. The script also explains the importance of monitoring the preview image during generation to gauge the step count and make adjustments to avoid overcooking the image. The video ends with a teaser about a possible 'Draw Things 2.0' release and an invitation for viewers to subscribe for updates.
Mindmap
Keywords
๐กUser Interface
๐กProject Manager
๐กLoRA
๐กControlNets
๐กStable Diffusion
๐กIP Adapter
๐กIn-Painting
๐กTextual Inversions
๐กStyle Templates
๐กJavaScripts Menu
๐กParameter Efficient Fine-Tuning (PEFT)
Highlights
Introduction to the new user interface for Draw Things in 2024, tailored for new and casual users.
Automatic optimization of Draw Things for the user's system with the option to manually override settings.
The importance of the new UI setting for font size and the ability to close hint text by clicking the header.
The project manager's role in creating a database for each project, storing history, prompts, and images.
The challenge of accessing root folders on OS 10 and how the project manager changes the workflow.
The capability to export projects and store them on an external hard drive or move them to another device.
New thumbnail scrolling sidebar displaying more of the project's history with options to delete or export images.
The introduction of style templates for text-based styling and the ability to add custom styles.
The functionality of the prompts and token windows, including partitioning prompts and token weight adjustments.
The ability to rename, export, or start a new project from the project name menu.
Export and share options with detailed save features, including copying to clipboard and saving just the visible canvas.
The inclusion of canvas layers for additional image input, such as scribble, pose, color palette, and mood board.
The use of the magic wand tool for AI segmentation and the eraser tool for image editing with mask parameter adjustments.
The parameter settings sidebar for controlling stable diffusion, including saving configurations and managing settings.
The model directory's new interface allowing users to search and manage models, including downloading and compressing models.
Exploring LoRA and control net parameters, including stacking and weight adjustments for fine-tuning image generation.
The advanced settings menu housing specialized parameters for specific models and the need for Draw Things to support new AI models.
Tips on using the preview image during rendering to gauge step count and image development, avoiding overcooking steps.
The potential evolution of the menu rail with more tools and the possible migration of control net image layers.
The possibility of Draw Things 2.0 and the continuous improvement of the platform.