Vidu-style video models can load parameter schemas that change the Generate Video node height and trigger React Flow dimension updates. The fix keeps schema-driven resize notifications stable, skips no-op node height writes, and ignores duplicate React Flow node changes before they enter the workflow store.
It also preserves completed media outputs when a stopped or reloaded workflow still has transient loading/skipped state, so existing video results do not appear stuck after recovery.
Constraint: The crash was reported from a browser state where choosing Vidu hit React's maximum update depth in onNodesChange.
Rejected: Clear browser storage as the fix | it would hide persisted workflow loops without fixing the update path.
Rejected: Remove dynamic video handles | visible media/text inputs are product behavior and required for model-specific workflows.
Confidence: high
Scope-risk: moderate
Directive: Keep React Flow dimension and schema resize writes idempotent; new node resize paths should no-op when dimensions are already applied.
Tested: npm test -- --run src/components/__tests__/GenerateVideoNode.test.tsx -t 'should not resize repeatedly'
Tested: npm test -- --run src/components/__tests__/GenerateVideoNode.test.tsx src/components/__tests__/ModelParameters.test.tsx src/store/__tests__/workflowStore.integration.test.ts src/store/__tests__/skipPropagation.test.ts src/store/execution/__tests__/generateVideoExecutor.test.ts
Tested: npm run build
Tested: npm run test:run
The accumulated fixes move the test build toward the current product shape: Popi models are the default provider surface, remote community workflow APIs are removed from the local template flow, media nodes expose accurate preview/preview-modal behavior, and Generate Video owns first-class duration, resolution, and sound defaults instead of burying those values in provider schemas.
The test suite was brought back in line with the product mode split: Popi provider mode is the default runtime, while multi-provider model search remains covered explicitly under multi mode.
Constraint: Product direction is canvas-owned workflow structure with Popi Models as the tested default provider path.
Rejected: Keep legacy community workflow API surface | user requested removing remote data paths and retaining local templates.
Rejected: Treat video duration/resolution/sound as hidden provider params | these are user-facing Generate Video controls.
Confidence: high
Scope-risk: broad
Directive: Do not reintroduce remote community template loading without a product decision on hosted templates and i18n ownership.
Tested: npm test -- --run src/store/utils/__tests__/nodeDefaults.test.ts src/store/execution/__tests__/generateVideoExecutor.test.ts src/components/__tests__/GenerateVideoNode.test.tsx src/components/__tests__/GenerationComposer.test.tsx
Tested: npm run build
Tested: npm test -- --run src/components/__tests__/ModelSearchDialog.test.tsx src/store/utils/__tests__/localStorage.test.ts
Tested: npm run test:run
Not-tested: Live provider generation calls against paid external APIs.
Resolve conflicts in 4 executor files by combining develop's
runWithFallback structure with feature branch's client-side polling.
Both capabilities are now integrated: executors use runWithFallback
for model redundancy, and pollGenerateTask for long-running Kie tasks.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Gemini image models (nano-banana, nano-banana-pro, nano-banana-2) were
not showing parameters when set as fallback models. Three root causes:
1. Schema API returned empty for Gemini image models — added
getGeminiImageSchema() with aspect ratio, resolution, and search
grounding parameters per model
2. ModelParameters had an early-return guard skipping all non-Veo Gemini
models — removed to make the component provider-agnostic
3. nanoBananaExecutor ignored parametersOverride for Gemini-specific
fields (aspectRatio, resolution, useGoogleSearch, useImageSearch) —
now reads from override when available
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
When a model schema loads, enum dropdowns showed "Default" but never
wrote the actual default value into node.data.parameters. This caused
the Kie API to receive missing parameter values, resulting in errors
like "mode is not within the range of allowed options."
Now a useEffect watches for schema changes and writes any schema-defined
defaults into parameters when the corresponding key is undefined.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
The render guard at line 222 blocked all Gemini-provider models from
rendering parameters, including Veo models. Hoisted the isVeoModel
check to component scope and added the same exception used in the
fetch guard so Veo parameters (aspect ratio, duration, resolution)
display correctly.
Fixes#106
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
ModelParameters had an early return for all Gemini provider models,
skipping the API schema fetch. This was correct for Gemini image
models (which have built-in parameters) but prevented Veo video
models from loading their aspect ratio, duration, and resolution
parameters from the /api/models/[modelId] endpoint.
Closes#106
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
On 144Hz+ displays, requestAnimationFrame fires ~144 times/sec. This
adds a throttledRAF utility that limits callback execution to ~60fps
regardless of display refresh rate, normalizing performance across
60Hz, 120Hz, and 144Hz monitors.
Applied to:
- workflowStore hover debounce (event-driven throttle)
- FloatingNodeHeader position tracking (continuous loop throttle)
- ModelParameters ResizeObserver (event-driven throttle)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The displaySchema useMemo was placed after conditional early returns,
violating React's Rules of Hooks and causing hook order mismatch errors
when the component toggled between rendering null and its content.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Wrap ModelParameters and ParameterInput in React.memo
- Add useMemo on sortedSchema and displaySchema computations
- RAF-debounce ResizeObserver with redundancy guard
- Stabilize onChange callback (pass name as prop, not closure)
- Eliminates re-renders during canvas pan/zoom/drag
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Change outer wrapper bg from #2a2a2a to bg-neutral-800 to match node body,
eliminating visible color mismatch at rounded corners
- Remove overflow-hidden from InlineParameterPanel (was clipping fields)
- Use min(180px, 100%) in grid minmax so columns shrink to fit narrow nodes
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add bg-[#2a2a2a] rounded-lg to outer wrapper when settings expanded to
fill gap behind rounded corners at node/panel junction
- Reduce grid min column width from 240px to 180px so fields fit in narrow nodes
- Add overflow-hidden to InlineParameterPanel content div as safety net
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Settings area uses lighter grey (#2a2a2a), field inputs use darker
grey (#1a1a1a) with no borders, matching the reference design.
Applied across InlineParameterPanel, GenerateImageNode, LLMGenerateNode,
and ModelParameters.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Solid dark background attached to image preview, labels on same line
as fields, slightly larger text (11px), darker rounded inputs matching
the reference design. Applied consistently across Gemini controls,
LLM controls, and ModelParameters (external provider fields).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Always display parameters inline instead of behind a collapsible
"Parameters" header. The toggle added unnecessary interaction cost.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use local state during editing and sync to store on blur, preventing
React Flow re-renders from resetting cursor position on every keystroke.
Also add click-to-seek on GenerateAudioNode waveform.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds 48-hour client-side caching to persist model data across dev server
restarts and page refreshes, eliminating repeated API calls during hot
reloading in development mode.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Split AudioInputNode waveform effect to avoid ResizeObserver churn on every animation tick
- Add missing kieApiKey/wavespeedApiKey deps to ModelParameters useEffect
- Clean up VideoSampleSource/Output on error in useApplySpeedCurve
- Fix off-by-one in useStitchVideos duplicate-frame skip (<= to <) and add try/finally for resource cleanup
- Replace leftover currentNodeId: null with currentNodeIds: [] in workflowStore
- Store full response metadata (status, headers, body) in deduplicatedFetch cache
- Add in-flight guard to imageStorage to prevent duplicate concurrent saves
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Multiple node instances (GenerateImageNode, GenerateVideoNode, etc.) were
each making their own /api/models requests when mounted, causing N duplicate
network requests when N nodes existed on the canvas.
This fix:
- Adds deduplicatedFetch utility that shares in-flight requests to the same URL
- Adds useProviderApiKeys() hook with shallow equality for stable API key refs
- Updates all components to use primitive dependencies instead of object refs
- Uses deduplicatedFetch for all model-fetching operations
Expected result: 1 request instead of N when multiple nodes need the same data.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Fixes issues with external provider models (e.g., seedream-4.5) where:
- Boolean/integer params with "image" in name were misclassified as image inputs
- Array params without items.type specification were incorrectly rejected
- Handle ID mismatch between node components (image-0) and store (image)
- dynamicInputs not passed to API in node regeneration code path
Changes:
- Update isImageInput() to check schema types, not just names
- Map both indexed (image-0) and legacy (image) handle IDs to schema names
- Add dynamicInputs to nanoBanana regeneration request body
- Add hidden backward-compat handles for edge routing
- Add comprehensive tests for schema classification and handle mapping
- Fix Toast test to use specific selector for close button
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Introduced a model search dialog for selecting external models in GenerateImageNode and GenerateVideoNode.
- Updated node titles to reflect selected models dynamically.
- Improved parameter handling by expanding the ModelParameters component to notify parent nodes on schema load.
- Adjusted handle positions and styles for better alignment and visual consistency.
- Enhanced input validation and processing for dynamic inputs across the workflow.
- Add ModelInputDef type to node data interfaces
- Update ModelParameters to pass inputs via onInputsLoaded callback
- GenerateImageNode and GenerateVideoNode store input schema
This sets up the foundation for dynamic handle rendering.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add onExpandChange callback to ModelParameters component
- Remove max-height constraint from expanded parameters
- GenerateImageNode/GenerateVideoNode calculate new height based on parameter count
- Node height grows when expanded, shrinks when collapsed
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add inline validation for number/integer parameters with min/max bounds
- Show red border and error text for out-of-range values
- Clear parameters when switching provider or model (schemas differ)
- Validation shows specific constraints (Min/Max values, integer requirement)
- Create ModelParameters component with collapsible UI
- Fetches schema from /api/models/{modelId} endpoint
- Renders appropriate inputs based on parameter type (string, number, boolean, enum)
- Add parameters field to NanoBananaNodeData and GenerateVideoNodeData
- Integrate into GenerateImageNode and GenerateVideoNode
TianYun
weige