Generation tabs

Nine pills set the category of space being generated. Selecting a type changes the available design style sub-pills below.
Interior - furnished interior rooms and spaces.
Architecture - building exteriors, façades, and site contexts.
Exterior - outdoor environments, landscapes, and streetscapes.
Product - individual product objects with full studio or lifestyle context.
Furniture - standalone furniture pieces.
Brand Packaging - product packaging and branding mock-ups.
Pattern - repeating surface patterns and textile designs.
Stickers - illustrated sticker-style artwork.
Logo - logo and wordmark concepts.

A second row of style pills appears below the space type and changes to match it. For Interior, options include Japandi, Scandinavian, Industrial, Mid-Century Modern, Art Deco, Contemporary, Bohemian, Minimalist, Coastal, Wabi-Sabi, and Rustic. For Architecture, styles include Modernist, Brutalist, Parametric, Neo-Classical, Bauhaus, Art Nouveau, Contemporary, Deconstructivist, and Biophilic. Other space types carry their own relevant style sets. Selecting a style auto-populates a starting prompt and influences the generation strongly.

Located in the right sidebar, this panel accepts one or more uploaded images that the AI uses as a visual reference for style, palette, and composition. Drag an image onto the panel or click to open the file picker.

Choose a base resolution (1K, 2K, 4K, Existing, or Custom) and an aspect ratio. Existing reads the resolution of the selected canvas image and matches it exactly. Custom unlocks independent width/height fields.

Converts a collage or moodboard image into a cohesive, photorealistic render. CreativeArc reads the spatial composition and surface references from the moodboard and synthesises a unified scene.

Produces a new variation of the selected image while preserving its overall composition and style. Use this to explore alternate colour schemes or material combinations within the same layout.

Applies the visual style of a reference image (uploaded via the reference panel) onto the selected canvas image. The structural composition of the source is retained; only the aesthetic is transferred.

Reformats the selected image to a different aspect ratio. When selected, a sub-strip appears:
DIRECTION - Vertical / Horizontal: choose whether to extend the canvas vertically (portrait) or horizontally (landscape). CreativeArc fills any new space with contextually appropriate content.

Produces a curated moodboard grid from the selected image - atmospheric photography, texture swatches, and colour references assembled into a single presentation layout.

Generates a full design board showing furniture selections, material finishes, and colour palette for the space. Formatted for client presentations.

Creates a focused material and finish reference board - surface textures, stone samples, fabric swatches, and paint chips in a clean layout. Useful for FF&E presentations.

Produces a brand-style board with logo usage, typography, colour system, and brand tone reference. Best used with product, packaging, and brand-identity projects.

Assembles a furniture selection board showing individual pieces, finishes, and dimensions in a presentation grid format.

Places a building (from a reference image or description) into a site photograph. CreativeArc reads the perspective and ground plane of the site image and composites the architecture into it realistically.

Generates an alternate façade treatment or material finish for a selected building image. Upload a reference to guide the new material language, or rely on the prompt alone.

Converts an aerial map, site plan, or satellite image into a photorealistic site render - showing landscaping, circulation, and context buildings in perspective.

Converts a perspective render into a top-down floor plan drawing. CreativeArc reads the spatial layout and traces it into a clean 2D plan with space outlines and object positions.

Transforms a floor plan or elevation image into a cut-section drawing - showing the vertical slice through the space with ceiling heights, openings, and spatial layers.

Renders the image with interior photography lighting - soft ambient fill, specular highlights on surfaces, and accurate depth of field. Best for furnished interior spaces.

Applies outdoor daylight rendering - natural sky illumination, shadow casting, and lens characteristics typical of architectural photography.

Renders in a neutral studio environment with controlled three-point lighting. Eliminates environmental distractions - ideal for product and furniture shots.

Converts the image into a hand-painted watercolour illustration with visible paper texture, wet-edge bleeding, and translucent washes.

Produces a clean architectural line illustration - outlines only, minimal tone, suitable for overlay or schematic use.

Creates a pencil-shaded sketch with hatching, tonal variation, and hand-drawn texture - the look of a designer's presentation drawing.

Produces a loose, gestural concept sketch - quick lines and rough form, ideal for early-stage design communication.

Converts the image into a flat-colour vector-style illustration with clean outlines and reduced detail - suitable for diagrams and explainer graphics.

Renders the space or object as a white clay model - flat diffuse surface, no textures - to communicate form and volume without material distractions.

Applies the design onto a product or surface mockup (packaging, wall, signage, etc.) in a realistic context, preserving the template's lighting and perspective.

Multi-select pills that refine photographic quality without changing the design, layout, or geometry of the image:
Lighting - improves exposure, tonal balance, and dynamic range.
Materials - sharpens material and surface realism.
Fabrics - adds texture and weave detail to soft furnishings.
Reflections - refines reflective surfaces (glass, metal, polished floors).
People - improves the realism of people already in the scene.
Warmth - adjusts colour temperature for a warmer, richer feel.
Photo Finish - applies a high-end architectural photography grade.
Depth - adds subtle depth-based grading to separate foreground, midground, and background.

Sets the primary light source and sky condition:
Day - bright overhead natural light, clear sky.
Golden Hour - warm, raking side-light just after sunrise or before sunset.
Evening - dim ambient light with artificial interior lighting dominant.
Night - dark exterior with lit windows and artificial sources only.
Overcast - flat, diffuse cloud-filtered light with no hard shadows.

Layered atmospheric effects (multi-select):
Rain - wet surfaces, reflections, and rain-streaked glass.
Fog - depth-fade mist softening distant elements.
Autumn - warm leaf-fall colour grading and golden foliage.
Winter - snow, cold blue ambient light, and frost detail.
Volumetric - visible light shafts and god rays in the scene.
Make brighter - globally lifts exposure and reduces shadow depth.

Shifts the overall colour cast of the light:
Warm - amber candlelight range, 2400–2800 K.
Neutral - clean white light, 3000–3500 K.
Cool - blue-daylight tone, 4000–5000 K.

Controls the tonal range between shadows and highlights:
Soft - compressed dynamic range, gentle gradients, film-still quality.
Balanced - standard photographic contrast.
Dramatic - deep shadows, bright highlights, editorial punch.

Replaces one colour in the image with another. When selected, a sub-strip appears with two controls:
FROM - an eyedropper tool. Click the eyedropper, then click a colour on the canvas image to sample the source colour that will be replaced.
TO - a colour picker swatch. Click it to open CreativeArc's colour picker - a shade area and hue slider, a live preview, a hex field, and an eyedropper for sampling any colour on screen (supported browsers). A clear (×) button resets the selection.

Replaces a material or surface texture on a product or object. Upload one or more texture reference images via the texture panel, then for each uploaded texture a text input appears labelled TEXTURE → PART. Type the name of the part or surface you want to apply that texture to (e.g. "cabinet door", "seat cushion"). CreativeArc replaces only the specified surface with the new material.

Works like Replace Material but is tuned for interior surfaces. Upload texture references and, for each one, enter the interior element to re-surface in the TEXTURE → OBJECT field (e.g. "floor", "wall", "countertop"). Multiple textures can be mapped simultaneously to different objects in a single generation.

Removes all furniture and loose objects from the selected image, leaving only the permanent architectural shell - walls, floor, ceiling, windows, and fixed elements. Useful as a staging base.

Extracts the main furniture piece or object from the image onto a transparent background. The output is a clean PNG cutout, ready for compositing or catalogue use.

Places the selected furniture piece (or the object described in the prompt) into a styled lifestyle environment scene, adding context, lighting, and background.

Scans the selected image and identifies all detectable furniture items. When selected, a sub-strip appears containing a Scan image button. Clicking it analyses the image and returns a list of detected objects. Select any item from the list to generate a cutout of just that piece.

Renders the selected furniture piece as a clean product studio shot. When selected, a sub-strip with five control groups appears:
VIEW - Assembled (complete product as-is) · Exploded (components separated and floating to reveal construction).
BG - a colour picker swatch sets the studio background colour; defaults to white. Clicking it opens CreativeArc's colour picker (shade area, hue slider, hex field, eyedropper).
SHADOW - Off (no shadow) · Soft (diffuse ground shadow) · Medium (defined contact shadow).
FRAMING - Original (keep source proportions) · Centered (centre the object with equal padding) · Tight (crop closely around the piece).
LIGHTING - Neutral · Soft daylight · Slightly dramatic · Warm studio · Cool studio.

Generates clean technical line drawings of a furniture piece - front, side, and top elevations - from a photograph or render. Output uses black linework on a white background.

Eight pills set the camera perspective of the generated image:
Front view - straight-on elevation facing the primary wall.
Side view - 90° rotated elevation showing the side wall or profile.
Top view - directly overhead plan view of the space.
Closeup - tight shot on a surface detail, object, or material feature.
Object closeup - macro-level focus on a single furniture piece or decorative item.
Axonometric - parallel-projection 3D view showing three faces simultaneously without perspective distortion.
Drone view - elevated exterior oblique angle, as seen from a drone or bird's-eye perspective.
Multiple views - produces a composite image showing the space from several angles simultaneously in a grid layout.

Adds people to the scene. Five control groups appear:
DENSITY - Low · Some · Medium · Busy · Packed.
AGE - multi-select: Children · Teenagers · Young Adults · Adults · Seniors.
ACTIVITY - multi-select: Standing · Walking · Seated · Working.
STYLE - multi-select: Casual · Business · Formal · Sport.
FOCUS - Sharp (people in clear focus) · Blurred (people soft and out of focus).

Adds vehicles to the scene.
TYPE - multi-select: Cars · Bicycles · Motorcycles · Buses.
FOCUS - Sharp · Blurred.

Adds vegetation to the scene.
TREES - None · Few · Dense.
PLANTS - multi-select: Flowers · Plants · Grass.
STYLE - multi-select: Urban · Tropical · Mediterranean · Wild.

Adds birds in flight to the scene.
AMOUNT - None · Few · Flock.
FOCUS - Sharp · Blurred.

A toggle sets whether you're upscaling an image or a video node-the model list and controls below change accordingly.

Choose the upscale engine. Clarity Pro Upscaler offers 2×, 4×, 8×, and 16× with a creativity slider. Topaz Image Upscale offers 2×, 4×, and 6× with five processing modes and optional face enhancement. P-Image Upscale offers 2×, 4×, 6×, and 8× as a lower-cost alternative. Flux Deblur sharpens and removes blur without changing resolution (no scale control). Magnific offers Creative mode (prompt-guided, HDR/resemblance/fractality sliders, engine and flavor selection) and Precision mode (a curated set of Magnific v1/v2 flavors with sharpen/grain/detail sliders).

Scale multiplier buttons. Available options change based on the selected model-Clarity Pro: 2× · 4× · 8× · 16×; Topaz: 2× · 4× · 6×; P-Image: 2× · 4× · 6× · 8×. The output pixel dimensions are shown live next to the selected scale.

Slider from −10 to +10 controlling how much the AI can reinterpret texture and detail. Lower values are more faithful to the source; higher values add enhancement. Labels: Strict · Faithful · Balanced · Enhanced · Creative.

Five processing modes optimised for different source types: Standard V2-general-purpose; Low Resolution V2-heavily pixelated or very small sources; CGI-computer-generated renders; High Fidelity V2-preserves fine detail; Text Refine-images with text or sharp graphic elements.

Toggle to enable face-specific enhancement. When on, exposes two sub-sliders: Creativity (0–1) controls how much detail the AI adds to faces; Strength (0–1) controls the intensity of the enhancement blended into the result.

Choose between Flux Video Upscale and Topaz Video Upscale-Topaz additionally lets you set a target resolution (720p / 1080p / 4K) and frame rate.

A dedicated Upscale action button replaces the standard Generate button when this tab is active. Drop your source image or video onto the upscale drop zone or select a canvas node, then click Upscale to process.

Choose the video engine-each has its own aspect ratios and generation modes:
Veo 3.1 - Google's model; supports reference-image and native clip-extension modes.
Gemini Omni Flash - Google's video-to-video editor; applies a prompted edit to an uploaded source clip rather than animating a still image.
Kling V3 - cinematic motion from a still image, with a first/last-frame mode for precise start-and-end control.
Flux 3 - Black Forest Labs' model, with a reference-image mode and a wide range of aspect ratios.
Seedance 2.5 - ByteDance's model, with a reference-image mode and adaptive aspect ratio.

Pill tab for the camera's viewpoint: Eye Level · Low Angle · High Angle · Ground Level · Worm's Eye · Bird's Eye / Top Down · Aerial · POV · Over the Shoulder · Profile · Dutch Angle · Tableau · Object POV · Voyeur · Fourth Wall.

Pill tab for how the camera moves during the shot: Pan Left/Right · Tilt Up/Down · Zoom In/Out · Dolly In/Out · Tracking Shot · Orbit · Rotate · Arc Shot · Crane Up/Down · Pedestal Up/Down · Trucking Left/Right · Handheld · Steadicam Follow · Whip Pan · Dolly Zoom · Push Through · Camera Roll · Locked-On · Lazy Susan · Snorricam.

Macro · Extreme Close-up · Close-up · Medium Shot · Cowboy Shot · Full Shot · Wide Shot · Establishing Shot · Two Shot.

Shallow Depth of Field · Deep Focus · Rack Focus · Reverse Rack Focus · Split Diopter · Focus Reveal · Tilt Shift.

24mm Wide Angle · Telephoto · Fisheye · Anamorphic · Macro Lens · Probe Lens · Halation · Parallax · Vignette.

Slow Motion · Fast Motion · Speed Ramp · Timelapse · Long Exposure · Bullet Time · Freeze Frame · Boomerang · Step Printing · Cinemagraph.

Leading Lines · Center Framing · Rule of Thirds · Symmetry · Negative Space · Frame Within Frame · Foreground Occlusion · Silhouette · Reflection Framing.

Rim Light · Chiaroscuro · Golden Hour · Neon Practicals · Volumetric Light · Hard Light · Haze · Spotlight · Light Flash · Projections · Underwater Light.

Match Cut · Whip Transition · Foreground Wipe · Jump Cut · Object Portal · Pass-through · Quick Cuts · Screen-in-screen.

Drone FPV · Bodycam · Dashcam · Mirror POV.

Cinemascope (21:9) · Vertical (9:16) · Vintage (4:3) · Split Screen.

Double Exposure · Datamosh · Kaleidoscope · Morphing · Slit Scan · X-ray · Levitation.

Dreamcore · Dystopian · Magical Realism · Maximalism · Diorama.

Stop Motion · Pixel Art · Zoetrope · Kinetic Typography.

Adds a few people moving naturally through the space with subtle camera drift for a cinematic, lived-in feel.

Available on models that support it (e.g. Kling V3). When expanded, lets you upload a specific start frame and end frame image. CreativeArc generates video that transitions from the first frame to the last, giving precise control over the motion arc.

Uses the last frame of the current generated video as the first frame of a new generation, chaining the motion seamlessly. For Veo results, an Extend natively (Veo) option is also available, continuing the exact clip as one continuous take. Extensions can be repeated to build longer sequences.

Rewrites the video prompt using cinematography guidelines - subject, action, camera style, composition, focus, ambiance, and audio descriptors. Replaces the current prompt text.

Sends the prompt to the selected model to generate the video. Generation typically takes 1–3 minutes. The result is added to the canvas as a video node.

Vector is a separate mode from Image, Video, and Upscale-select it from the mode pill above the sidebar. It always works from Recraft, and can generate from a text prompt or act on the Selected Image.

Choose Recraft V4.1 or Recraft V3 as the base model, plus a variant: Standard (raster), Vector (true SVG), Pro (higher resolution and detail), Pro Vector, Utility (simpler, more predictable output-good for mockups and product listing photos), Utility Vector, Utility Pro, and Utility Pro Vector. Recraft V3 only offers Standard and Vector.

When a Selected Image is present, the model pill also offers two one-click actions in place of a prompt-driven generation: Vectorize converts the selected image into a true SVG using Recraft's AI vectorizer; Remove BG strips its background using Recraft's AI background remover.

Recraft V3 and V3 Vector support named style presets (e.g. Photorealism, Studio photo, Line art, Flat illustration, and many more, grouped by category) that steer the overall look. V4.1 models don't use style presets.

A dedicated chip-based prompt builder for icon sets, available only in Vector mode. Set Type (Outline, Pictogram, Filled), Stroke (Thin, Medium, Bold), Corners (Sharp, Soft, Rounded), Detail (Minimal, Simple, Detailed), Proportion (Compact, Balanced, Wide, Tall), Style (Geometric, Organic, Technical, Friendly), Colour (Black, White, Brand, Custom), Background (Transparent, White, Custom), and Variation (Strict, Balanced, Creative). CreativeArc assembles these into a structured icon prompt automatically.

Analyses the selected image and generates a colour-coded segmentation mask where each unique material or surface is assigned a distinct flat colour. Use the mask as a reference to identify which areas to target with Replace Color or Interior Material.

Detects discrete objects in the image and generates a per-object segmentation mask. Each detected object receives a unique colour. Useful for selecting specific pieces to replace or remove.

Saves the current prompt text as a named custom prompt. Clicking the button opens a name input field with a Save confirmation and an × cancel button. A prompt must be present in the prompt box before saving.

Lists all previously saved custom prompts by name. Each entry has two actions: clicking the name loads the prompt into the prompt box and triggers generation immediately; the × button permanently deletes the saved prompt.