Image nodes

Image nodes are the creative core of Spaces. They split into two groups: generators that create images from scratch, and processors that transform existing images. You can chain them together to build complete visual pipelines.

In this article

• Image Generator
• Image Upscaler
• Variations
• Image Editor
• Crop
• Expand Image
• Remove Background
• Adjust
• Change Camera Angle
• Experimental nodes
• Common workflows
• Tips and best practices

Image Generator

The Image Generator creates original images from text prompts using AI. Choose from dozens of models, each specializing in different styles: photorealism, illustration, anime, 3D rendering, and more.

How to use it

1. Add the node

   Search for Image Generator in Spotlight. You can also search for a specific model like Flux Ultra to add a pre-configured node.

2. Write your prompt

   Describe what you want to generate. You can type directly on the card or connect a Text or Assistant node to the Prompt input.

3. Choose a model

   Click the model selector to browse available AI models. Each model has different strengths, features, and credit costs.

4. Set your options

   Pick an aspect ratio, resolution (1K, 2K, or 4K), and how many images to generate (1 to 10).

5. Add references (optional)

   Connect reference images to the Reference input to guide style or composition.

6. Run the node

   Click Run. Results appear on the node card. Use the arrow buttons to browse multiple generations.

Settings

Depending on the model selected, additional settings may be available in the node inspector: Seed (for reproducible results), Smart Prompt (AI-enhanced prompt optimization), Modifiers (style keywords), Color Palette (color guidance), Image Prompt Strength (how much the reference influences output), Guidance Scale (prompt adherence), and Unlimited Mode (bypass credit limits on supported plans).

Input and output

Use cases

• Text-to-image. Type a prompt, get an image. The simplest and most common workflow.
• Style transfer. Connect a reference image and describe what you want. The AI uses the reference as a visual anchor.
• Batch generation. Connect a List of prompts to generate one image per prompt automatically.
• Production pipeline. Image Generator at 1K for speed, then Image Upscaler to 4K for the final output.
• Video from image. Connect the output to a Video Generator to animate your creation.

Image Upscaler

The Image Upscaler enhances resolution and quality using AI. It takes a low-resolution image and outputs a sharper, larger version. Two modes cover different needs.

How to use it

1. Add the node

   Search for Image Upscaler in Spotlight. You can also search for a specific mode like Image Upscaler Creative.

2. Connect an image

   Connect an Image Generator output, an Upload node, or any image source to the Input Image port.

3. Choose a mode

   Creative for AI-enhanced detail, Precision for faithful scaling.

4. Set the scale

   2x, 4x, 8x, or 16x depending on the mode.

5. Add a prompt (Creative mode)

   Optionally describe what the image contains to help the AI make better enhancement decisions.

6. Run the node

   The upscaled image appears on the card.

Creative mode highlights

Creative mode offers three AI engines (Illusio for illustrations, Sharpy for sharp detail, Sparkle for photos) and four presets (Subtle, Vivid, Wild, Custom). Four sliders give you precise control: Creativity, Resemblance, HDR, and Fractality. You can also select a content category (Portrait, Nature, 3D, Film, etc.) to optimize the AI's approach.

Precision mode highlights

Precision mode offers four models (Ultra Sublime, Ultra Photo, Ultra Denoiser, Ultra) and three presets (Balanced, Portraits, Grainy Analog). Sliders for Sharpness, Grain, and Ultra Detail let you fine-tune the output.

Use cases

• High-res output. Generate at 1K for speed, upscale the winners to 4K.
• Photo restoration. Upload an old or low-quality photo, use Creative mode to enhance it with AI-added detail.
• Print preparation. Precision mode at 8x or 16x for print-ready resolution.
• Artistic enhancement. Creative mode with the Wild preset for richly detailed textures and depth.

Variations

The Variations node generates multiple image variations from a single source. Choose a mode, and the AI creates a grid of alternatives that you can split into individual images for downstream processing.

This node is being rolled out gradually and may not be available in your account yet.

How to use it

1. Add the node

   Search for Variations in Spotlight.

2. Connect a source image

   Connect any image output to the Input Image port.

3. Choose a mode

   Select the type of variation you want (see table below).

4. Set the grid size

   Choose how many variations to generate (2x2, 3x3, 4x4).

5. Enable Split Output

   Turn this on to use each variation independently in downstream nodes.

6. Run the node

   The variation grid appears on the card.

Variation modes

Use cases

• Product photography. Angles mode for multiple views without re-shooting.
• Inclusive content. Demographics mode for diverse representation in marketing materials.
• Character design. Expressions mode for emotion sheets.
• Storyboarding. Storyboard mode with a narrative prompt for visual sequences.
• Creative exploration. Custom mode to discover unexpected directions.

Image Editor

The Image Editor is an umbrella node that gives you five editing tools for images, all without leaving Spaces. When you add it, you choose which tool to use. You can also search for a specific tool in Spotlight (for example, "Crop" or "Remove Background") to skip the selection step.

Image Editor is a terminal node — it has input ports but no output ports. Edited images are saved to your library. Use it as the final step in your image workflow.

Crop

Crop, rotate, and resize images with manual controls or AI-assisted framing.

• Aspect ratio presets. 1:1, 4:3, 16:9, 3:2, and their vertical counterparts. Select Free for unconstrained cropping.
• Custom dimensions. Enter exact width and height in pixels.
• Rotation. Straighten horizons or add a creative tilt (-45 to +45 degrees).
• AI auto-crop. Let the AI decide the best crop based on image content. Connect a prompt like 'focus on the person's face' for guided framing.

Expand Image

Extend the canvas in any direction with AI-generated content that blends seamlessly with the original.

• Choose a target aspect ratio or set custom dimensions. The AI fills the new canvas area naturally.
• Add a prompt to guide what the AI generates in the expanded area. For example, 'continue the forest landscape' or 'add a sunset sky'.
• When left empty, the AI infers from the existing image content.
• Select a model and set a seed for reproducible results.

Remove Background

Remove, replace, or recolor the background of any image. Three modes cover different needs.

Enable Preserve Details for subjects with hair, fur, lace, or glass. It makes a noticeable difference on tricky edges. In Replace mode, you can generate up to 4 background alternatives in a single run.

Additional settings include Preserve Details (toggle for fine edge refinement), Number of Images, and Seed for reproducible results.

Adjust

Apply precise color and exposure adjustments with eight sliders.

Change Camera Angle

Modify the camera perspective of an image. The AI fills in the unseen parts of the scene.

• Horizontal rotation. Rotate the virtual camera left or right. Range is -90 to +90 degrees with a single reference, or -180 to +180 degrees with multiple references.
• Vertical rotation. Tilt the camera up or down (-90 to +90 degrees).
• Close-up / Zoom. Push the camera closer to the subject (0 to 10, single reference only).
• Multiple references. Connect several images of the same subject for dramatically better results at extreme angles.

Experimental nodes

Spaces includes a few experimental image nodes that are still in development. These may change or be removed in the future.

• Image to SVG. Converts raster images (PNG, JPG) into scalable vector graphics. Useful for logos, icons, and illustrations.
• SVG Generator. Creates vector graphics from text descriptions.
• SVG Animation. Adds CSS/SMIL animations to static SVG files.
• Image to 3D. Converts a 2D image into a 3D model (GLB format) that you can view and rotate on the canvas.

Common workflows

Image nodes connect naturally with each other and with the rest of Spaces. Here are some typical pipelines.

1. Text to Image. Text node with your prompt, connected to Image Generator. The simplest starting point.
2. Generate and upscale. Image Generator at 1K for fast iteration, then Image Upscaler to 4K for the final output.
3. Generate and explore. Image Generator connected to Variations. Explore different angles, expressions, or styles from a single generation.
4. Edit and polish. Any image connected to Image Editor for cropping, background removal, or color grading.
5. Full production pipeline. Text to Image Generator to Upscaler to Designer. End-to-end from idea to finished graphic.

Tips and best practices

Connect an Assistant to the prompt port. The Assistant can write optimized prompts that get better results from each model.

Use @mentions in prompts. Reference other nodes' content dynamically. The prompt updates automatically when upstream nodes change.

Start at 1K, finish at 4K. Generate fast at low resolution, pick the best result, then upscale. This saves both time and credits.

Set a seed when iterating. Lock the seed so only your prompt changes affect the output. This makes it much easier to compare variations.

Reference images plus a clear prompt. When you connect references, the AI uses them as a visual guide, not a strict template. A clear prompt keeps the output on track.

Search for specific tools in Spotlight. Instead of adding a generic Image Editor and then choosing a tool, search directly for "Crop" or "Remove Background" to skip a step.

Adjust Exposure first, then Brightness. They affect different tonal ranges. Exposure is the coarse control, Brightness is the fine-tuning.

A tiny amount of Grain makes AI images feel more organic. Try values around 0.02 to 0.04 in the Adjust tool.

Enable Preserve Details for tricky edges. Hair, fur, lace, and glass all benefit from this toggle in Remove Background.

Enable Split Output on Variations. This lets you route each variation independently to downstream nodes like Upscaler or Designer.