ImageRAG
Fix AI Art Mistakes and Create Cohesive Projects
Watch “How to Make Highly Specific Revisions in Midjourney” for a great overview and tutorial of ImageRAG.
What is ImageRAG?
ImageRAG is a software, sold as a subscription service, that works like a an online workspace to manage and control AI Art. The tool allows creators to “stitch” and “unstitch” image collections. This enables users to refine small regions using any AI image model. This can be done for one image at a time or for multiple images at once. The level of control this offers unlocks consistency across an entire project, without sacrificing the ease that comes from using AI Tools.
What does ImageRAG solve for you?
It enables the organization of your AI Art characters, backgrounds, and scenes to help you manage workflows within your projects. Whether that is illustrating a story, building starting frames for a short film, or developing creative for a real-life product— ImageRAG is the workspace to give you the control you need.
The idea being that, with practice, you’ll be able to build coherent projects with consistent characters and objects, in exactly the style you want, while you control every detail you care about. To the extent possible, the aim to remove the guesswork and randomness from AI generations.
1
ImageRAG gives the ability to manage what an image model sees and work on. This lets you control prominence to make higher quality edits.
2
By building collections, you are able to create more a canvas for AI generation that can contains more context for the model to understand your goal.
3
The edits you make to the cropped portions of your images fit back into their originals, to create polished final results.
AI Image models prioritize prominent details. However, the best creative art has specific details that are important to the scene all throughout your canvas. Think of a painting in a museum, it’s appeal is not just about a few prominent features, but all of the details that build on each other and make a cohesive statement. ImageRAG allows you to focus on small areas of your scene and feed them to an AI model so those small details become the prominent focus of the generation. Once you are happy with the details in the small part of the scene, ImageRAG seamlessly puts it back into your scene/project so that you can focus on making your whole project just how you want it.
AI Image models don’t solve a problem the same way twice. This makes it difficult for you to keep details of people and objects consistent across your project. However, it is much better at solving consistency within the same image. By stitching the same character or object from across your project into one image and asking the AI Image generator to update for consistency across your stitched feature, you can create the consistency you need across your story arc. ImageRAG helps you to create these stitches and then redistribute the pieces back where they belong without your worrying about keeping track of alignment issues which can get quite finicky and tedious.
You can organize your scenes and assets so you can concentrate on your role as art director and create whatever you can imagine.
Finish Projects and Create More
On Glibatree Tools, I am sharing exclusive access to ImageRAG and all the custom tools I make, and use, to help you get more from AI Art generations.
Key Features
- Stitch and unstitch images to edit details for consistency across you creative projects
- Download and Upload images between ImageRAG and your favorite AI image Generator and editing software
- Organize your projects and easily access your assets as you build your creative visions
Work Flows
Steps for using ImageRAG to fix details on your canvas by making them prominent.
Click on + New Project to create a new project. Double Click on the Project Name to rename it to something meaningful.
Click Upload Images(s) to upload an image. You can double click on the description to change the name to something meaningful. You can also drag an image into the canvas and it will automatically upload it too.
You can hold the left button and drag your mouse to select a region, hover over a selection handle in the canvas and hold the left button and drag the mouse to resize or move the selection or hover over the handle in the selection area, hold the left button and drag to either move or resize the selection.
Click Create Collection in the Options Section to create a collection. Collections can have one item or multiple items. For this work flow, you will just have one item in your collection.
With your collection highlighted, click Download & Lock from the options section. Bring the downloaded file into you AI Image generator for revising. This will allow you to fix the details because they will be prominent.
The stitch process will work well even if your choose to upscale.
Be sure to not change the border of your selection, so that it fits back into your image without introduced artifacts or sharp transitions.
Drag the revised image into the Check In Image Section, or click Select From Computer to upload the revised image.
You can also click “Unlock Manually” to restart the process and change your selection.
You can also click “Download Stitched Image” if you need to re-download the image.
Once the revised image is uploaded, you can review the changes.
If you click on the old box in Review Changes, you will see the original selection.
If you click on the new box in Review Changes, you will see the new image.
If you hold “shift” while you click you will see both the old and new images stacked vertically.
If you click on your item inside the collection, you will see the entire image. If new is clicked you will see the revised image “stitched” into the entire image. If both old and new are highlighted, you both the original image and the stiched versions side by side.
Clicking Accept All Changes will incorporate your changes into you image.
Clicking Reject All Changes will allow you to upload a different revised image.
Clicking Unlock Manually will restart the process, and allow you to change your selection or pick anew image.
Download Stitched Image downloads another copy of your selection to use in your AI Image Generator.
For the full, proven, consistent character workflow: click here
Below are steps to utilize ImageRAG to fix character details on your canvas by making them prominent and adding additional characters in context.
Click on + New Project to create a new project. Double Click on the Project Name to rename it to something meaningful.
Click Upload Images(s) to upload your images. You can double click on the description of each image to change the names to something meaningful. You can also drag images into the canvas and it will automatically upload it too.
Remember that the revision process will change the selected parts of each image. If you want to change objects in the same image, it may make sense to do this as a multi-step process so that your final result incorporates each change.
Highlight all the images that have characters/object that your would like to work on together and click Create Collection in the Options Pane to create a colection.
It is wise to put your best resolution selection and likely reference image first. It sets the dimensions of the other image selections.
3 to 5 items in a collection is typically recommended. It gives enough context to create consistency and is of a good size for the AI Image generator to work with.
Click on each item in your collection and select the area of interest.
You can hold the left button and drag your mouse to select a region, hover over a selection handle in the canvas and hold the left button and drag the mouse to resize or move the selection or hover over the handle in the selection area, hold the left button and drag to either move or resize the selection.
With your collection highlighted, click Download & Lock from the options section. Bring the downloaded file into you AI Image generator for revising. This will allow you to fix the details because they will be prominent.
The stitch process will work well even if your choose to upscale.
Be sure to not change the borders around any of your selections, so that it fits back into your images without introduced artifacts or sharp transitions.
Drag the revised image into the Check In Image Section, or click Select From Computer to upload the revised image.
You can also click “Unlock Manually” to restart the process and change your selection.
You can also click “Download Stitched Image” if you need to re-download the image.
Once the revised image is uploaded, you can review the changes.
If you click on the old box in Review Changes, you will see the original selection.
If you click on the new box in Review Changes, you will see the new image.
If you hold “shift” while you click you will see both the old and new images stacked vertically.
If you click on any of yours item inside the collection, you will see that entire image. If new is clicked you will see the revised image “stitched” into the entire image. If both old and new are highlighted, you both the original image and the stitched versions side by side.
Clicking Accept All Changes will incorporate your changes into you image.
Clicking Reject All Changes will allow you to upload a different revised image.
Clicking Unlock Manually will restart the process, and allow you to change your selection or pick anew image.
Download Stitched Image downloads another copy of your selection to use in your AI Image Generator.
Sign up / Sign In today
If you are using ImageRag and run into any issues or bugs, please feel free to email support@glibatree.com.