Sergio is a specialist in generative AI and technology, and leads the Redraw development team.
How to Edit Specific Parts of an AI-Generated Image Without Losing What Was Already Approved
Text mode, area selection with layers, and filters: see how Redraw's image editor changes only what you asked for and keeps what the client already approved.
4 min
|
06.05.2026
Sergio is a specialist in generative AI and technology, and leads the Redraw development team.
[
Redraw
]
Create your free account
Start transforming your projects today using Redraw.
I want to register
Share this article
Most AI image generation tools work with the entire image. You describe, generate, and if something isn't right, you have to start over from scratch. There is no middle ground between accepting the result or losing everything that was already approved.
Redraw solves this problem in a way no other image generation platform for architecture and interior design does: you can edit the whole image with a text instruction, use a reference image to guide the result, or select exactly what you want to change. When the edit is made by area, the AI changes only what you asked for, while the rest of the image stays preserved.
What is Redraw's image editor?
The image editor is a feature that brings together three editing modes, one of them being layer-based editing. With it, you work more precisely: each selected area gets its own instruction, and everything is sent to the AI at once.
The feature has three main resources. Text mode lets you apply a general instruction to the entire image, ideal for broad adjustments of style, atmosphere, materials, decoration, or visual direction. Select area mode lets you outline specific regions of the image with a selection tool and edit each one with more control; this is the layer-based editing mode. Filters mode lets you adjust visual aspects of the image, such as light, texture, and preset filters, and save the result directly, without combining it with the Text or Select area modes.
In addition, Redraw lets you use a reference image in editing. This reference can be used in Text mode, when you want to guide the overall transformation of the image, and also in layers, when you need to indicate more clearly the style, material, color, object, or finish you want in a specific area.
How does Text mode work?
Text mode is recommended when you want to make a broader change to the image, without selecting a specific region. In this case, you write a general instruction for the AI, explaining what you want to change or improve.
You can ask, for example, to make the living room more sophisticated, turn the space into a minimalist style, change the color palette, improve the decoration, change the visual proposal of the space, or adapt the scene to a specific aesthetic reference.
You can also add a reference image in Text mode. This helps when you want the AI to follow a clearer visual direction, such as a furniture style, a material palette, a lighting atmosphere, or a finish pattern. Instead of relying only on the written description, the visual reference helps guide the result better.
Text mode is most useful when the change involves the image as a whole. For specific changes, such as swapping just a sofa, changing a lamp, or adjusting a specific wall, Select area mode offers more control.
How does layer-based editing work in practice?
When you enter Select area mode, you choose a selection tool to create your editing layers.
Magic selection uses AI to automatically detect the object under the cursor. You hover over the area and the platform highlights the outline in real time. One click confirms the selection.
The Lasso lets you draw freely around a region of the image, without depending on a rigid shape.
The Brush works like a freehand marking, ideal for irregular regions or ones that blend into the background.
The Rectangle creates a box over the desired area, useful for well-defined elements like furniture, windows, and architectural elements.
You can switch between selection tools while creating the layers, choosing the most suitable one for each area you need to adjust.
After selecting an area, an instruction field opens directly over the selection. You type what you want to change in that region and confirm. The layer is added to the side panel with the instruction and the individual cost in coins.
You can also use a reference image in a layer. This is useful when the text instruction alone isn't enough to explain the desired result. For example, you can select a sofa and add a reference image with the model, fabric, or style you want to apply. You can also select a wall and use a texture, color, or finish reference.
You can create up to four layers per edit. The panel shows the total in coins before you confirm. When you're ready, you send everything at once. The AI processes all instructions simultaneously and returns the image with the edits applied.
How does Filters mode work?
Filters mode lets you adjust visual aspects of the image and save the result directly from the panel, without creating a Text instruction or a Select area layer.
When you click Filters, the side panel opens with preset filters and light and color controls, such as exposure, highlights, shadows, gamma, and brightness. You apply the desired adjustment and save the image directly from the button in the panel.
Use Filters mode when you want to refine the visual reading of the image without changing its composition or elements. It is useful for balancing lighting, adjusting contrast, applying an overall color or texture, or testing atmosphere variations without generating a new image from scratch.
Why does this change the market standard?
AI image generation tools deliver fast results, but they charge a high price when the result isn't exact: you lose the entire image and start again. In projects with a client waiting for approval, this trial-and-error cycle consumes time, coins, and energy.
In Redraw, you are no longer generating images and hoping they come out right. You are editing with real control over specific parts of the composition.
This means an image approved by the client can be adjusted with much more confidence. The sofa changes, the lamp changes, the wall color changes. The composition, the lighting, and what was already approved stay preserved.
And when the change needs to follow a visual reference, you don't depend only on a prompt. The reference image helps communicate to the AI exactly the kind of result you're looking for, both in general edits and in layer changes.
When to use each editing resource?
Use Text mode when you want to change the entire image or steer the overall aesthetic of the scene. It is ideal for changes of style, atmosphere, decoration, color palette, general lighting, or visual proposal.
Use Select area mode when the change is specific and you don't want to risk the rest of the image. This mode is especially useful when the client approves the overall composition but asks for specific adjustments. It also helps when you want to test variations of an element without redoing the whole scene, when the image has multiple adjustments in different regions and you want to solve everything in a single generation, or when the general prompt isn't precise enough for the result you need.
Use Filters mode when you want to adjust visual aspects of the image, such as light, texture, or preset filters, and save the result directly. It works independently, without needing to combine with the Text or Select area modes.
Tips to get the most out of layer-based editing
Plan the layers before you start selecting. Identify all the regions that need changes, create the layers in sequence, and review the panel before confirming. This avoids generating more than once for the same adjustment.
Before creating the layers, choose the selection tool best suited to the type of change you need to make. Magic selection works best for well-defined objects. For regions with irregular edges or that blend into the background, the Lasso or the Brush can offer more control.
Use a reference image when the result depends on a visual characteristic that is hard to explain in text alone, such as texture, finish, furniture style, fabric type, wood pattern, specific color, or aesthetic language.
Check the total cost in the panel before confirming.
Frequently asked questions
Can I edit the whole image without selecting an area?
Yes. For that, use Text mode. It lets you apply a general instruction to the image, without creating a specific selection.
Can I use a reference image in Text mode?
Yes. The reference image can be used in Text mode to guide the overall transformation of the image. This helps when you want to follow a specific aesthetic, materiality, style, or atmosphere.
Can I use a reference image in a layer?
Yes. You can use a reference image in a layer to guide the editing of a specific area. This is useful for swapping furniture, applying materials, changing finishes, or bringing the result closer to a visual reference.
Does Filters mode work independently?
Yes. In Filters mode, you apply preset filters or adjust visual aspects like light and texture, and save the image directly from the panel, without combining with the Text or Select area modes.
What happens to the parts of the image that weren't selected?
They stay preserved. The AI applies the edits only to the regions and instructions you defined.
Can I use more than one selection tool in the same edit?
Yes. In a single layer-based edit, you can use different selection tools to create the areas of the operation, choosing the most suitable one for each adjustment.
Does the four-layer limit apply to all modes?
The limit is specific to Select area mode. The Text and Filters modes work independently.
Is the Image editor available on all plans?
Yes. The feature is available on the Basic, Expert, and Business Elite plans. Coin consumption is the same on all plans.
Share this article
Summary of the content
[
Blog
]
Related articles
Access Blog
