Image Effects Nodes

Analyze, transform, and apply effects to images.

Image Background

Remove or generate backgrounds for images

Process image backgrounds by removing them entirely or generating new backgrounds based on text prompts.

Property	Description
Inputs	image (image), input (text, optional -- for background generation prompt)
Outputs	image (image)

Configuration:

Setting	Description
Model	AI model for processing (see table below)

Available Models:

Model	Mode	Description
wavespeed-ai/image-background-remover	Remove	Fast background removal (default)
bria/remove-background	Remove	High-quality edge detection
bria/generate-background	Generate	AI-generated new background from text prompt

When using bria/generate-background, connect a text source to the input handle to provide a prompt describing the desired background. The handler also falls back to the flow context's promptText if no direct text input is connected.

How it works:

1. The image is sent to the generate-image edge function with the selected model
2. For async jobs, the node polls poll-wavespeed-job for up to 2 minutes with progress updates
3. Credits are refunded automatically if the job fails or times out

Use cases:

• Product photography cleanup
• Portrait backgrounds
• Composite creation
• E-commerce product images

---

Angles

Generate consistent camera angle packs from a hero image

Creates multiple camera angle variations from a single hero image using edit models. Uses progressive reference anchoring for consistency across generated angles.

Property	Description
Inputs	images_in (image), ctx_in (flow, optional context)
Outputs	images_out (image, all angles), anchors_out (image, best reference anchors), json_out (any, metadata), log_out (text, generation log)

Configuration:

Setting	Options	Description
Angle Pack	action, starter, cinematic, creator, detail, realworld, custom	Pre-configured angle sets
Model	bytedance/seedream-v4/edit, google/nano-banana-pro/edit	AI edit model
Style Lock	strict, balanced, creative	Visual style consistency
Lighting Lock	strict, balanced, loose	Lighting consistency
Background Stability	keep, simplify, replace	Background handling
Outfit Lock	true/false	Preserve outfit
Upscale 4K	true/false	Output in 4K via ultimate-image-upscaler
Refiner	none, photorealistic, cinematic, identity_lock	Photo refinement (Nano Banana)

Angle Packs:

Pack	Angles	Description
Action Pack (Default)	11 angles	High-quality action angles
Starter Pack	8 angles	Essential angles for beginners
Cinematic Pack	12 angles	Cinematic composition angles
Creator Pack	16 angles	Comprehensive creator angles
Detail Pack	10 angles	Detail-focused close-ups
Real-World Perspectives	5 angles	Extreme aerial/bottom-up (Nano Banana only)
Custom	Variable	Select specific angles

Available Angles (27 total):

Distance/Framing (7):

• Wide Establishing, Full Body, Three Quarter, Medium Shot, Close-up Face, Extreme Close-up Eye, Macro Hand/Prop

Vertical/Camera Angle (4):

• Low Angle Hero, High Angle Vulnerable, Bird's Eye Top-Down, Dutch Angle

Rotation/Viewpoint (11):

• Profile Left, Profile Right, 3/4 Left, 3/4 Right, Back View, Over-the-Shoulder Left, Over-the-Shoulder Right, POV Subject, Off-center Left, Off-center Right, Silhouette Long

Real-World Perspectives (5) -- Nano Banana only:

• Aerial Top-Down, Top-Down Environment, Worm's Eye Dramatic, Bottom-Up Heroic, Bottom-Up Portrait

How it works:

1. Subject Analysis -- The hero image is analyzed using the openrouter-chat edge function to detect subject type (person, animal, product, vehicle, etc.). Results are cached in node state.
2. Angle Plan -- An ordered list of angles is built from the selected pack, respecting prerequisites.
3. Lock Block -- Identity, style, and lighting lock instructions are generated based on config and subject analysis.
4. Progressive Generation -- Each angle is generated sequentially using edit models. As angles complete, they update an anchor registry (base, fullBody, face, env, threeQuarter) that subsequent angles reference for consistency.
5. Optional Upscale -- If Upscale 4K is enabled, each generated angle is upscaled via wavespeed-ai/ultimate-image-upscaler.

Use cases:

• Product photography (multiple views)
• Character design sheets
• Fashion lookbooks
• 3D asset reference creation
• Storyboard development

---

Analyzer

Analyze image content using AI vision

Performs AI-powered analysis of images with four distinct modes, using the OpenRouter chat API with vision capabilities.

Property	Description
Inputs	image (image)
Outputs	description (text)

Analysis Modes:

Mode	Description
describe	General description of image content, composition, colors, lighting, and style
tags	Comma-separated keyword extraction (subjects, style, mood, technical aspects)
recreate	Detailed prompt generation that could have produced the image
qc	Quality critique -- composition, lighting, exposure, sharpness, artistic merit

How it works:

• Sends the image to the openrouter-chat edge function in analyze mode
• Analysis result is stored in node state (analysisResult, analysisMode) for persistence
• The analysis text replaces promptText in the flow context, making it available to downstream nodes

Use cases:

• Automatic tagging and content description
• Quality assessment before publishing
• Prompt generation from reference images (use recreate mode)
• Technical quality critique

---

Veo Effects

Compile cinematographic prompts for Veo video generation

Deterministically compiles cinematographic prompts with camera effects, lighting, styling, and stability settings. This is a local compilation node -- no API calls, no credits consumed.

Property	Description
Inputs	text_in (text, prompt text), refs_in (image, optional reference images)
Outputs	veo_out (text, compiled Veo-ready prompt), refs_out (image, pass-through references), log_out (text, warnings)

Configuration:

Setting	Description
Camera	Camera movement effect (13 options)
Lighting	Lighting style (10 options)
Style	Visual style preset (8 options)
Camera/Lighting/Style Intensity	Effect strength: subtle, medium, dramatic
Stable	Reduce drift (boolean)
Lock Framing	Lock subject framing (boolean)
Simplify Motion	Simplify motion (boolean)
Reuse Refs	Reuse reference images (boolean)
Aspect	Output aspect ratio
Duration	Video duration in seconds
No Subtitles	Suppress subtitles (boolean)

Camera Effects (13):

• Gentle Dolly-In, Slow Pan, Parallel Tracking, Crane Rise, Handheld Urgency, Static Lock-Off, Whip-Pan, Overhead Top-Down, POV Reality, Slow Arc Orbit, Ultra-Wide Sweep, Telephoto Glide, Stutter Action

Lighting Presets (10):

• Soft Window Key, Noir Low-Key, High-Key Product, Sunset Backlight, Practical + Negative Fill, Neon Rim Blast, Dream Diffusion, Hard Silhouette, Studio Tri-Light, Fireplace Warmth

Style Presets (8):

• Cinematic Realism, Clean Commercial, Documentary Natural, and more

Use cases:

• Prepare images for Image to Video with Veo models
• Create motion-ready assets with precise cinematographic control
• Build consistent visual style across video shots

---

Combining Effects

Multi-Angle Product Pipeline

Generate complete product views:

Upload --> Image Background (remove) --> Angles --> Library

Video Preparation

Prepare images for optimal video generation:

Text to Image --> Veo Effects --> Image to Video

Professional Photography Workflow

Create complete product catalogs:

Upload --> Image Background (remove) --> Angles (Action Pack) --> Library

---

Best Practices

For Angles

1. Start with a clear, well-lit hero image
2. Remove backgrounds first for best results
3. Use "Action Pack" for most use cases
4. Enable "Outfit Lock" for consistent clothing
5. Use "Real-World Perspectives" pack for dramatic shots (requires Nano Banana model)

For Background Removal

1. Use high-contrast images (subject vs background)
2. wavespeed-ai/image-background-remover for speed
3. bria/remove-background for high-quality edges
4. bria/generate-background for creative new backgrounds

For Analysis

1. Use high-quality input images
2. Choose the appropriate mode for your need: describe for downstream prompts, recreate for reproducing images, qc for quality checks
3. Connect Analyzer output to Brain node for further processing