/
Use every Flowgen input, transform, generation, editing, and output node.
Estimate scene depth from one image locally with Depth Anything V2, then output grayscale, inverted, or Turbo-colored depth at source, 1K, or 2K size.
Apply one of 18 generative visual treatments—or a custom instruction—to a connected image with Seedream 5 Pro Edit.
Bind a site collection, map typed fields, resolve stable entry identity, save a draft or published record, mirror images, and preserve concurrent versions.
Use a paired source Items List and Results List for every new repeated row workflow. Group Repeater, loop groups, and Group Batch mode are deprecated and remain available only so existing canvases can
Use Flowgen’s Object Select node to click one object, Shift-click additional objects, Alt-click objects to remove, drag a box around several, choose a tighter/looser mask, and emit both cutout and mas
Crop one connected image with normalized Left, Top, Width, and Height controls or lock the result to Square, Wide, Portrait, or Story proportions.
Upload or select an HTML source and expose it through a typed HTML output for joining, validation, authoring, or rendering workflows.
Create distance-aware background or foreground blur from an optional depth map, or estimate depth locally from the source image.
Choose from the current live model/workflow picker, including local-first segmentation and Seedream 5 Pro Gen Effect; GPT Image 1.5 and Imagen 4 are retired from new-node pickers and remain only for s
Turn image transparency, brightness, or a locally segmented foreground subject into a reusable grayscale mask.
Convert part of a connected video into a browser-generated 256-color GIF with controlled frame rate, width, duration, and looping.
Read entries from a bound site collection as a typed dataset, filter draft/live state, order and limit results, or check whether one source URL already exists.
Upload or select a 3D model and expose it as a typed model result for preview, optimization, or a 3D scene.
Use the Document Editor card in video project mode to open an editable native f0 video document from connected media.
Create typed records, fill them with a selected AI model and rich connected context, review them as a list/grid/carousel, or pair the List with a Results List to run and collect a complete row workflo
Render a local depth-blurred video, pick static focus from the opening frame, or pass the effect into a downstream composition.
Mirror a connected image horizontally or vertically and rotate it by exact quarter turns without sending the image to a generation model.
Build Route & Switch branches whose winning text or media value stays current through per-case returns, Result, List iterations, exposed App switches, and downstream Extract / Merge nodes.
Use a live, pure Transform to take apart one input or bring several ordered inputs together as text, fields, or a list; add safe visual steps, inspect actual values, follow Route winners, and expose n
Apply one of 116 deterministic GPU effects to an image or video with a visual thumbnail picker, effect-specific controls, position picking, live previews, and free local rendering.
Split a connected image into four grayscale channel images or choose one channel and publish that same selected channel on every output socket.
Use the Agentic Web card with the interactive preset to create a code-backed interactive section without enabling the motion or 3D capability flags by default.
Extract 2–16 dominant colors from image pixels, returning both a visual swatch strip and a prevalence-ordered JSON color list.
Understand exactly which legacy HTML nodes were retired, why existing flows still work, what replaces each node, and how to migrate without confusing the current Unified Section/Agentic Web node with
Pull the image, SVG or video you already placed in the page editor straight into flowGen as a ready-made input node — no re-uploading.
Choose among 20 focused nodes for deterministic local image work, prompted masks, video mattes/GIFs, depth workflows, before/after review, and a paid generative effect.
Upload audio, use a URL, choose a project asset, or replace the current source and expose it to voice, lip-sync, composition, and audio-render nodes.
Apply an even Gaussian blur or a lens-style bokeh pass with aperture shape, rotation, highlight bloom, and threshold controls.
Set master input black/white points, midtone gamma, and output black/white points on a connected image.
Apply a connected grayscale mask to an image as its alpha channel, with optional inversion and feathering.
Use the Grid contact sheet to scan media-heavy records, resize tiles, and open any complete item in a fullscreen carousel while preserving the same typed rows, field order, locks, and batch behavior.
Upload an image, paste a URL, choose one from the project Asset Browser, or replace the current source while preserving a typed image output.
Correct exposure, brightness, contrast, saturation, vibrance, temperature, and tint in one deterministic image step.
The Custom Code node lets you author your own flowGen node in TSX. You declare its input/output sockets, an optional run() transform executed when you press Run, and an optional live View rendered in
Detect or mark the changeable parts of one HTML page, expose stable typed sockets, fill one record at a time, and optionally register the template as a site collection.
Evaluate ordered cases live, force a branch manually or through Pick by name, return case-specific typed values, merge the winner on Result, or rerun an earlier stage within a strict bound.
Use the shared video-generation card in extension mode so a connected source clip continues with a compatible prompt, model, duration, and model-specific controls.
Inspect two image, mask, or video layers with an interactive before/after wipe without running a generation or flattening the comparison.
Composite a second image over a base with 15 blend modes, opacity, and an optional invertible/feathered mask.
Invert either the visible RGB colors or the alpha channel of one image, making hidden pixels visible and visible pixels transparent when Alpha is selected.
Create a full matte video or a keyed subject video from chroma-key footage or local neural foreground segmentation.
Video Mask by Text is retired because it could not reliably track a described subject through a clip; saved nodes remain visible as removal tombstones.
Understand the twelve visible Brain/Brand/Business/Persona/Dataset/Document Add actions and the two underlying reader/writer card contracts they configure.
Apply one semantic, shot-local motion revision to an upstream shot: source windows, timing, reframe, element/camera tracks, cue markers, and entrance-hold-exit choreography.
Upload or select a PDF and expose it as a typed document source for research, context, extraction, or another compatible node.
Feather, smooth, tighten, expand, and increase the contrast of a connected matte before compositing.
Upload a video, use a URL, choose project media, or replace the current source while keeping a typed video output in the graph.
Run the same site publishing pipeline from Flowgen, using a connected Website checkpoint and an explicit Preview or Live target.
Combine Read a Page, Page Template, Save a Page, Find Pages, Items, Route, Approval, Website Editor, and Publish Site without duplicating content or confusing HTML generation with CMS storage.
Analyze one video into a reconciled millisecond timeline, searchable Markdown, and structured JSON with optional visual/audio descriptions, transcription, subject tracks, and removable dead-part sugge
Estimate depth frame by frame in the open browser, or carry the depth operation into a downstream Video Render composition.
Map an arriving JSON record or dataset onto a website post type, the marked fields of a Page Template, or your own target field list—with direct key mappings, text templates, fixed values, and reviewa
The Text Prompt node holds the words you wire into image, video, voice and stock nodes — type your own, or let Generate Idea / Add Details draft and expand it.
Source nodes let you bring your own files into a flow: drop a file into the dashed area and it uploads to storage, then exposes a typed output handle to wire into editors and generators.
Search Brain records by a query, optional built-in/custom item-type filters, and a bounded result count, then emit both structured chunks and formatted context.
Create an origin shot or add a non-destructive design revision with editable elements, source choices, layout variants, render profiles, and a reusable poster output.
Set exact output dimensions up to 8192 pixels, either scaling the image itself or changing only its canvas, with optional proportion locking.
Enlarge one image by 1x, 2x, 4x, or 8x with an instant browser resize, free Swin2SR super-resolution on this device, or paid Topaz/Recraft processing.
Describe an object in a still image and receive a mask, using on-device OWL-ViT plus SAM2 first and a charged cloud fallback only when necessary.
Read one public URL through the server-side page reader and return clean Markdown plus structured images, colors, fonts, links, title, and description.