Merge pull request #83 from Runware/feature-videoOutput

Adding Runware Output Node

__init__.py

@@ -71,6 +71,7 @@ from .modules.safetyInputs import safetyInputs
 from .modules.settings import RunwareSettings
 from .modules.videoAdvancedFeatureInputs import videoAdvancedFeatureInputs
 from .modules.advancedFeatureSettings.wanAnimateAdvancedFeatureSettings import RunwareWanAnimateAdvancedFeatureSettings
+from .modules.videoOutputs import RunwareVideoInferenceOutputs
 
 RUNWARE_COMFYUI_VERSION = "1.4.0 Beta"
 
@@ -156,6 +157,7 @@ NODE_CLASS_MAPPINGS = {
     "Runware Settings": RunwareSettings,
     "Runware Video Advanced Feature Inputs": videoAdvancedFeatureInputs,
     "Runware Wan Animate Advanced Feature Settings": RunwareWanAnimateAdvancedFeatureSettings,
+    "Runware Video Inference Outputs": RunwareVideoInferenceOutputs,
 }
 
 NODE_DISPLAY_NAME_MAPPINGS = {

clientlibs/main.js

@@ -1,6 +1,6 @@
 import { app } from "../../scripts/app.js";
 import { api } from "../../scripts/api.js";
-import { promptEnhanceHandler, syncDimensionsNodeHandler, searchNodeHandler, APIKeyHandler, captionNodeHandler, mediaUUIDHandler, videoTranscriptionHandler, handleCustomErrors, videoInferenceDimensionsHandler, videoModelSearchFilterHandler, audioModelSearchFilterHandler, useParameterToggleHandler, imageInferenceToggleHandler, upscalerToggleHandler, videoUpscalerToggleHandler, audioInferenceToggleHandler, acceleratorOptionsToggleHandler, bytedanceProviderSettingsToggleHandler, openaiProviderSettingsToggleHandler, lightricksProviderSettingsToggleHandler, klingProviderSettingsToggleHandler, lumaProviderSettingsToggleHandler, briaProviderSettingsToggleHandler, pixverseProviderSettingsToggleHandler, alibabaProviderSettingsToggleHandler, mireloProviderSettingsToggleHandler, googleProviderSettingsToggleHandler, syncProviderSettingsToggleHandler, syncSegmentToggleHandler, settingsToggleHandler, audioInputToggleHandler, speechInputToggleHandler, briaProviderMaskToggleHandler, wanAnimateAdvancedFeatureSettingsToggleHandler, videoAdvancedFeatureInputsToggleHandler, audioInferenceInputsToggleHandler } from "./utils.js";
+import { promptEnhanceHandler, syncDimensionsNodeHandler, searchNodeHandler, APIKeyHandler, captionNodeHandler, mediaUUIDHandler, videoTranscriptionHandler, videoOutputsHandler, handleCustomErrors, videoInferenceDimensionsHandler, videoModelSearchFilterHandler, audioModelSearchFilterHandler, useParameterToggleHandler, imageInferenceToggleHandler, upscalerToggleHandler, videoUpscalerToggleHandler, audioInferenceToggleHandler, acceleratorOptionsToggleHandler, bytedanceProviderSettingsToggleHandler, openaiProviderSettingsToggleHandler, lightricksProviderSettingsToggleHandler, klingProviderSettingsToggleHandler, lumaProviderSettingsToggleHandler, briaProviderSettingsToggleHandler, pixverseProviderSettingsToggleHandler, alibabaProviderSettingsToggleHandler, mireloProviderSettingsToggleHandler, googleProviderSettingsToggleHandler, syncProviderSettingsToggleHandler, syncSegmentToggleHandler, settingsToggleHandler, audioInputToggleHandler, speechInputToggleHandler, briaProviderMaskToggleHandler, wanAnimateAdvancedFeatureSettingsToggleHandler, videoAdvancedFeatureInputsToggleHandler, audioInferenceInputsToggleHandler } from "./utils.js";
 import { RUNWARE_NODE_TYPES, RUNWARE_NODE_PROPS, SEARCH_TERMS } from "./types.js";
 
 const nodeInitList = [];
@@ -11,6 +11,7 @@ app.registerExtension({
         api.addEventListener('runwareImageCaption', captionNodeHandler);
         api.addEventListener('runwareMediaUUID', mediaUUIDHandler);
         api.addEventListener('runwareVideoTranscription', videoTranscriptionHandler);
+        api.addEventListener('runwareVideoOutputs', videoOutputsHandler);
     },
 
     async nodeCreated(node) {
@@ -39,6 +40,18 @@ app.registerExtension({
                 mediaUUIDWidget.inputEl.style.fontSize = "11px";
                 mediaUUIDWidget.inputEl.style.fontFamily = "monospace";
             }
+        } else if(nodeClass === RUNWARE_NODE_TYPES.VIDEOINFERENCEOUTPUTS) {
+            // Style draftId and videoId widgets to show full IDs (like Media Upload)
+            const draftIdWidget = node.widgets.find(widget => widget.name === "draftId");
+            const videoIdWidget = node.widgets.find(widget => widget.name === "videoId");
+            for(const w of [draftIdWidget, videoIdWidget]) {
+                if(w && w.inputEl) {
+                    w.inputEl.style.width = "100%";
+                    w.inputEl.style.minWidth = "300px";
+                    w.inputEl.style.fontSize = "11px";
+                    w.inputEl.style.fontFamily = "monospace";
+                }
+            }
         }
 
         const nodeWidgets = node.widgets;

clientlibs/types.js

@@ -121,6 +121,7 @@ const RUNWARE_NODE_TYPES = {
     SETTINGS: "Runware Settings",
     VIDEOADVANCEDFEATUREINPUTS: "Runware Video Advanced Feature Inputs",
     WANANIMATEADVANCEDFEATURESETTINGS: "Runware Wan Animate Advanced Feature Settings",
+    VIDEOINFERENCEOUTPUTS: "Runware Video Inference Outputs",
 };
 
 const RUNWARE_NODE_PROPS = {
@@ -420,6 +421,10 @@ const RUNWARE_NODE_PROPS = {
         bgColor: DEFAULT_BGCOLOR,
         colorModeOnly: true,
     },
+    [RUNWARE_NODE_TYPES.VIDEOINFERENCEOUTPUTS]: {
+        bgColor: DEFAULT_BGCOLOR,
+        colorModeOnly: true,
+    },
 };
 
 export {

clientlibs/utils.js

@@ -106,6 +106,44 @@ function videoTranscriptionHandler(msgEvent) {
     return false;
 }
 
+function videoOutputsHandler(msgEvent) {
+    const outputData = msgEvent.detail;
+    const draftId = outputData.draftId || "";
+    const videoId = outputData.videoId || "";
+    const nodeID = outputData.nodeID;
+    
+    if(outputData.success && nodeID !== undefined && nodeID !== null) {
+        const nodeIdInt = typeof nodeID === 'string' ? parseInt(nodeID) : nodeID;
+        const outputNode = app.graph.getNodeById(nodeIdInt);
+        
+        if(outputNode !== null && outputNode !== undefined) {
+            const draftIdWidget = outputNode.widgets.find(widget => widget.name === "draftId");
+            const videoIdWidget = outputNode.widgets.find(widget => widget.name === "videoId");
+            
+            if(draftIdWidget) {
+                if(draftIdWidget.value !== undefined) draftIdWidget.value = draftId;
+                if(draftIdWidget.inputEl) {
+                    draftIdWidget.inputEl.value = draftId;
+                    draftIdWidget.inputEl.dispatchEvent(new Event('input', { bubbles: true }));
+                    draftIdWidget.inputEl.dispatchEvent(new Event('change', { bubbles: true }));
+                }
+            }
+            
+            if(videoIdWidget) {
+                if(videoIdWidget.value !== undefined) videoIdWidget.value = videoId;
+                if(videoIdWidget.inputEl) {
+                    videoIdWidget.inputEl.value = videoId;
+                    videoIdWidget.inputEl.dispatchEvent(new Event('input', { bubbles: true }));
+                    videoIdWidget.inputEl.dispatchEvent(new Event('change', { bubbles: true }));
+                }
+            }
+            
+            outputNode.setDirtyCanvas(true);
+        }
+    }
+    return false;
+}
+
 function notifyUser(message, type="info", title = "Runware", life = 4.5) {
     app.extensionManager.toast.add({
         severity: type, // 'info', 'success', 'warn', 'error' \\
@@ -1131,6 +1169,8 @@ function bytedanceProviderSettingsToggleHandler(bytedanceNode) {
     const fastModeWidget = bytedanceNode.widgets.find(w => w.name === "fastMode");
     const useAudioWidget = bytedanceNode.widgets.find(w => w.name === "useAudio");
     const audioWidget = bytedanceNode.widgets.find(w => w.name === "audio");
+    const useDraftWidget = bytedanceNode.widgets.find(w => w.name === "useDraft");
+    const draftWidget = bytedanceNode.widgets.find(w => w.name === "draft");
     
     // Helper function to toggle widget enabled state (exact same pattern)
     function toggleWidgetState(useWidget, paramWidget, paramName) {
@@ -1189,6 +1229,10 @@ function bytedanceProviderSettingsToggleHandler(bytedanceNode) {
     if (useAudioWidget && audioWidget) {
         toggleWidgetState(useAudioWidget, audioWidget, "audio");
     }
+
+    if (useDraftWidget && draftWidget) {
+        toggleWidgetState(useDraftWidget, draftWidget, "draft");
+    }
 }
 
 function openaiProviderSettingsToggleHandler(openaiNode) {
@@ -3149,6 +3193,7 @@ export {
     mediaUUIDHandler,
     captionNodeHandler,
     videoTranscriptionHandler,
+    videoOutputsHandler,
     handleCustomErrors,
     APIKeyHandler,
     videoInferenceDimensionsHandler,

modules/providerSettings/bytedanceImageProviderSettings.py

@@ -47,6 +47,14 @@ class RunwareBytedanceProviderSettings:
                     "tooltip": "Enable audio generation support when available.",
                     "default": False,
                 }),
+                "useDraft": ("BOOLEAN", {
+                    "tooltip": "Enable to include draft parameter in provider settings",
+                    "default": False,
+                }),
+                "draft": ("BOOLEAN", {
+                    "tooltip": "Enable draft mode for Bytedance provider",
+                    "default": False,
+                }),
             }
         }
     
@@ -63,12 +71,14 @@ class RunwareBytedanceProviderSettings:
         useMaxSequentialImages = kwargs.get("useMaxSequentialImages", False)
         useFastMode = kwargs.get("useFastMode", False)
         useAudio = kwargs.get("useAudio", False)
+        useDraft = kwargs.get("useDraft", False)
         
         # Get value parameters
         cameraFixed = kwargs.get("cameraFixed", False)
         maxSequentialImages = kwargs.get("maxSequentialImages", 1)
         fastMode = kwargs.get("fastMode", False)
         audio = kwargs.get("audio", False)
+        draft = kwargs.get("draft", False)
         
         # Build settings dictionary - only include what is enabled
         settings = {}
@@ -88,6 +98,10 @@ class RunwareBytedanceProviderSettings:
         # Add audio only if useAudio is enabled
         if useAudio:
             settings["audio"] = audio
+
+        # Add draft only if useDraft is enabled
+        if useDraft:
+            settings["draft"] = draft
         
         # Clean up None values
         settings = {k: v for k, v in settings.items() if v is not None}

modules/utils/runwareUtils.py

@@ -317,6 +317,19 @@ def sendVideoTranscription(transcriptionText, nodeID):
         },
     )
 
+
+def sendVideoOutputs(draftId, videoId, nodeID):
+    """Send draftId and videoId to frontend for display in Video Inference Outputs node textboxes."""
+    PromptServer.instance.send_sync(
+        "runwareVideoOutputs",
+        {
+            "success": True,
+            "draftId": draftId or "",
+            "videoId": videoId or "",
+            "nodeID": nodeID,
+        },
+    )
+
 def inferenecRequest(genConfig):
     global RUNWARE_API_KEY, RUNWARE_API_BASE_URL, SESSION_TIMEOUT
     RUNWARE_API_KEY = os.getenv("RUNWARE_API_KEY")

modules/videoBgRemoval.py

@@ -125,7 +125,8 @@ class videoBgRemoval:
                     if status == "success":
                         if "videoURL" in videoData or "videoBase64Data" in videoData:
                             videos = rwUtils.convertVideoB64List(pollResult, 1920, 1080)
-                            return videos
+                            # SaveVideo expects a single video object, not a tuple
+                            return (videos[0],) if len(videos) > 0 else (None,)
             
             comfy.model_management.throw_exception_if_processing_interrupted()
             

modules/videoInference.py

@@ -141,8 +141,8 @@ class txt2vid:
 
     DESCRIPTION = "Generates Videos Lightning Fast With Runware Video Inference Engine."
     FUNCTION = "generateVideo"
-    RETURN_TYPES = ("VIDEO", "RUNWARETASK")
-    RETURN_NAMES = ("VIDEO", "RW-Task")
+    RETURN_TYPES = ("VIDEO", "RUNWARETASK", "RUNWARETASK")
+    RETURN_NAMES = ("VIDEO", "RW-Task", "OUTPUT")
     CATEGORY = "Runware"
 
     def generateVideo(self, **kwargs):
@@ -325,7 +325,7 @@ class txt2vid:
             genConfig[0]["acceleration"] = acceleration
 
         if (multiInferenceMode):
-            return (None, genConfig)
+            return (None, genConfig, None)
         else:
             try:
                 # Debug: Print the request being sent
@@ -388,7 +388,9 @@ class txt2vid:
                         if status == "success":
                             if "videoURL" in video_data or "videoBase64Data" in video_data:
                                 videos = rwUtils.convertVideoB64List(pollResult, width, height)
-                                return videos
+                                # SaveVideo expects a single video object, not a tuple
+                                video_output = videos[0] if len(videos) > 0 else None
+                                return (video_output, genConfig, pollResult)
                         
                         # If status is "processing", continue polling
                 

modules/videoInferenceInputs.py

@@ -43,6 +43,16 @@ class videoInferenceInputs:
                 "Reference Voices": ("RUNWAREREFERENCEVOICES", {
                     "tooltip": "Connect the Reference Voices node to provide reference voice URLs or mediaUUIDs."
                 }),
+                "draftId": ("STRING", {
+                    "default": "",
+                    "multiline": False,
+                    "tooltip": "Draft ID from a previous Bytedance video generation. Used when continuing or extending a draft output",
+                }),
+                "videoId": ("STRING", {
+                    "default": "",
+                    "multiline": False,
+                    "tooltip": "Video ID from a previous generation. Used when extending or editing an existing OpenAI Sora video",
+                }),
             }
         }
 
@@ -65,6 +75,8 @@ class videoInferenceInputs:
         references = kwargs.get("References", None)
         referenceVideos = kwargs.get("Reference Videos", None)
         referenceVoices = kwargs.get("Reference Voices", None)
+        draftId = kwargs.get("draftId", "")
+        videoId = kwargs.get("videoId", "")
 
         inputs = {}
 
@@ -112,4 +124,10 @@ class videoInferenceInputs:
         if referenceVoices is not None and len(referenceVoices) > 0:
             inputs["referenceVoices"] = referenceVoices
 
+        if draftId is not None and draftId.strip() != "":
+            inputs["draftId"] = draftId.strip()
+
+        if videoId is not None and videoId.strip() != "":
+            inputs["videoId"] = videoId.strip()
+
         return (inputs,)

modules/videoOutputs.py

@@ -0,0 +1,75 @@
+from .utils import runwareUtils as rwUtils
+
+
+class RunwareVideoInferenceOutputs:
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "Output": (
+                    "RUNWARETASK",
+                    {
+                        "tooltip": "Connect the OUTPUT from Runware Video Inference.",
+                    },
+                ),
+            },
+            "optional": {
+                "draftId": ("STRING", {
+                    "placeholder": "Draft ID will appear here after generation.",
+                    "tooltip": "This field will be automatically populated with the draft ID from the video inference output.",
+                }),
+                "videoId": ("STRING", {
+                    "placeholder": "Video ID will appear here after generation.",
+                    "tooltip": "This field will be automatically populated with the video ID from the video inference output.",
+                }),
+            },
+            "hidden": {"node_id": "UNIQUE_ID"},
+        }
+
+    DESCRIPTION = "Extracts Runware video inference outputs such as draftId and videoId."
+    FUNCTION = "get_outputs"
+    RETURN_TYPES = ("STRING", "STRING")
+    RETURN_NAMES = ("Draft ID", "Video ID")
+    CATEGORY = "Runware"
+    OUTPUT_NODE = True  # Ensure node executes even if outputs aren't used
+
+    def get_outputs(self, Output, draftId=None, videoId=None, **kwargs):
+        draft_id = ""
+        video_id = ""
+
+        try:
+            data_list = None
+
+            # Handle different possible response shapes defensively
+            if isinstance(Output, dict):
+                if "data" in Output:
+                    data_list = Output.get("data", [])
+                elif "response" in Output and isinstance(Output["response"], dict):
+                    data_list = Output["response"].get("data", [])
+
+            if data_list and len(data_list) > 0:
+                first = data_list[0]
+                outputs = first.get("outputs") or {}
+                if isinstance(outputs, dict):
+                    # draftId is used by some providers
+                    draft_id = outputs.get("draftId", "") or outputs.get("draftID", "")
+                    # videoId / videoID used by other providers
+                    video_id = outputs.get("videoId", "") or outputs.get("videoID", "")
+
+            # Send to frontend so values appear in textboxes (like Media Upload)
+            rwUtils.sendVideoOutputs(draft_id, video_id, kwargs.get("node_id"))
+
+        except Exception as e:
+            print(f"[Runware Video Outputs] Failed to parse outputs: {e}")
+
+        return (draft_id, video_id)
+
+
+NODE_CLASS_MAPPINGS = {
+    "RunwareVideoInferenceOutputs": RunwareVideoInferenceOutputs,
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "RunwareVideoInferenceOutputs": "Runware Video Inference Outputs",
+}
+

modules/videoUpscaler.py

@@ -121,7 +121,8 @@ class videoUpscaler:
                         if status == "success":
                             if "mediaURL" in video_data or "videoURL" in video_data or "videoBase64Data" in video_data:
                                 videos = rwUtils.convertVideoB64List(pollResult, 1920, 1080)
-                                return videos
+                                # SaveVideo expects a single video object, not a tuple
+                                return (videos[0],) if len(videos) > 0 else (None,)
                         
                         # If status is "processing", continue polling