diff --git a/example_workflows/wanvideo_I2V_InfiniteTalk_example_02.json b/example_workflows/wanvideo_I2V_InfiniteTalk_example_03.json
similarity index 96%
rename from example_workflows/wanvideo_I2V_InfiniteTalk_example_02.json
rename to example_workflows/wanvideo_I2V_InfiniteTalk_example_03.json
index 2834730..e1c88ad 100644
--- a/example_workflows/wanvideo_I2V_InfiniteTalk_example_02.json
+++ b/example_workflows/wanvideo_I2V_InfiniteTalk_example_03.json
@@ -1,44 +1,9 @@
 {
   "id": "8b7a9a57-2303-4ef5-9fc2-bf41713bd1fc",
   "revision": 0,
-  "last_node_id": 304,
-  "last_link_id": 545,
+  "last_node_id": 310,
+  "last_link_id": 551,
   "nodes": [
-    {
-      "id": 120,
-      "type": "MultiTalkModelLoader",
-      "pos": [
-        592.73583984375,
-        -2520.13525390625
-      ],
-      "size": [
-        511.7737121582031,
-        66.51400756835938
-      ],
-      "flags": {},
-      "order": 0,
-      "mode": 0,
-      "inputs": [],
-      "outputs": [
-        {
-          "name": "model",
-          "type": "MULTITALKMODEL",
-          "links": [
-            355
-          ]
-        }
-      ],
-      "properties": {
-        "cnr_id": "ComfyUI-WanVideoWrapper",
-        "ver": "058286fc0f3b0651a2f6b68309df3f06e8332cc0",
-        "Node name for S&R": "MultiTalkModelLoader"
-      },
-      "widgets_values": [
-        "WanVideo\\InfiniteTalk\\InfiniteTalk\\Wan2_1-InfiniteTalk_Single_Q8.gguf"
-      ],
-      "color": "#223",
-      "bgcolor": "#335"
-    },
     {
       "id": 239,
       "type": "MarkdownNote",
@@ -51,7 +16,7 @@
         214.95587158203125
       ],
       "flags": {},
-      "order": 1,
+      "order": 0,
       "mode": 0,
       "inputs": [],
       "outputs": [],
@@ -75,7 +40,7 @@
         82
       ],
       "flags": {},
-      "order": 2,
+      "order": 1,
       "mode": 0,
       "inputs": [
         {
@@ -161,7 +126,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 36,
+      "order": 38,
       "mode": 0,
       "inputs": [
         {
@@ -199,7 +164,7 @@
         58
       ],
       "flags": {},
-      "order": 3,
+      "order": 2,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -276,7 +241,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 31,
+      "order": 30,
       "mode": 0,
       "inputs": [
         {
@@ -316,7 +281,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 32,
+      "order": 31,
       "mode": 0,
       "inputs": [
         {
@@ -356,7 +321,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 4,
+      "order": 3,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -390,7 +355,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 5,
+      "order": 4,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -422,7 +387,7 @@
         150
       ],
       "flags": {},
-      "order": 6,
+      "order": 5,
       "mode": 0,
       "inputs": [
         {
@@ -473,7 +438,7 @@
         202
       ],
       "flags": {},
-      "order": 7,
+      "order": 6,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -514,7 +479,7 @@
         58
       ],
       "flags": {},
-      "order": 8,
+      "order": 7,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -550,7 +515,7 @@
         58
       ],
       "flags": {},
-      "order": 9,
+      "order": 8,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -652,7 +617,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 10,
+      "order": 9,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -672,76 +637,6 @@
       "color": "#233",
       "bgcolor": "#355"
     },
-    {
-      "id": 242,
-      "type": "GetNode",
-      "pos": [
-        3670.50830078125,
-        -1982.0732421875
-      ],
-      "size": [
-        210,
-        60
-      ],
-      "flags": {
-        "collapsed": true
-      },
-      "order": 11,
-      "mode": 0,
-      "inputs": [],
-      "outputs": [
-        {
-          "name": "WANVAE",
-          "type": "WANVAE",
-          "links": [
-            438
-          ]
-        }
-      ],
-      "title": "Get_VAE",
-      "properties": {},
-      "widgets_values": [
-        "VAE"
-      ],
-      "color": "#322",
-      "bgcolor": "#533"
-    },
-    {
-      "id": 270,
-      "type": "INTConstant",
-      "pos": [
-        2655.538330078125,
-        -2261.7578125
-      ],
-      "size": [
-        210,
-        58
-      ],
-      "flags": {},
-      "order": 12,
-      "mode": 0,
-      "inputs": [],
-      "outputs": [
-        {
-          "name": "value",
-          "type": "INT",
-          "links": [
-            472
-          ]
-        }
-      ],
-      "title": "Max frames",
-      "properties": {
-        "cnr_id": "comfyui-kjnodes",
-        "ver": "e435e999e4b1a828a6b5f6d8f037e66f4a798324",
-        "Node name for S&R": "INTConstant"
-      },
-      "widgets_values": [
-        1000
-      ],
-      "color": "#1b4669",
-      "bgcolor": "#29699c"
-    },
     {
       "id": 294,
       "type": "SetNode",
@@ -827,7 +722,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 33,
+      "order": 35,
       "mode": 0,
       "inputs": [
         {
@@ -960,7 +855,7 @@
         143.7332305908203
       ],
       "flags": {},
-      "order": 13,
+      "order": 10,
       "mode": 0,
       "inputs": [],
       "outputs": [],
@@ -983,7 +878,7 @@
         314
       ],
       "flags": {},
-      "order": 30,
+      "order": 34,
       "mode": 0,
       "inputs": [
         {
@@ -1011,7 +906,7 @@
           "link": null
         },
         {
-          "name": "vace_model",
+          "name": "extra_model",
           "shape": 7,
           "type": "VACEPATH",
           "link": null
@@ -1035,7 +930,7 @@
           "link": null
         },
         {
-          "name": "extra_model",
+          "name": "vace_model",
           "shape": 7,
           "type": "VACEPATH",
           "link": null
@@ -1077,7 +972,7 @@
         202
       ],
       "flags": {},
-      "order": 14,
+      "order": 11,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1118,7 +1013,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 15,
+      "order": 12,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1138,141 +1033,12 @@
       "color": "#1b4669",
       "bgcolor": "#29699c"
     },
-    {
-      "id": 130,
-      "type": "WanVideoDecode",
-      "pos": [
-        3675.3330078125,
-        -1893.888671875
-      ],
-      "size": [
-        315,
-        198
-      ],
-      "flags": {},
-      "order": 45,
-      "mode": 0,
-      "inputs": [
-        {
-          "name": "vae",
-          "type": "WANVAE",
-          "link": 438
-        },
-        {
-          "name": "samples",
-          "type": "LATENT",
-          "link": 530
-        }
-      ],
-      "outputs": [
-        {
-          "name": "images",
-          "type": "IMAGE",
-          "slot_index": 0,
-          "links": [
-            493
-          ]
-        }
-      ],
-      "properties": {
-        "cnr_id": "ComfyUI-WanVideoWrapper",
-        "ver": "c3ee35f3ece76e38099dc516182d69b406e16772",
-        "Node name for S&R": "WanVideoDecode"
-      },
-      "widgets_values": [
-        false,
-        272,
-        272,
-        144,
-        128,
-        "default"
-      ],
-      "color": "#322",
-      "bgcolor": "#533"
-    },
-    {
-      "id": 131,
-      "type": "VHS_VideoCombine",
-      "pos": [
-        3689.34814453125,
-        -1589.9613037109375
-      ],
-      "size": [
-        991.5499877929688,
-        1319.550048828125
-      ],
-      "flags": {},
-      "order": 46,
-      "mode": 0,
-      "inputs": [
-        {
-          "name": "images",
-          "type": "IMAGE",
-          "link": 493
-        },
-        {
-          "name": "audio",
-          "shape": 7,
-          "type": "AUDIO",
-          "link": 449
-        },
-        {
-          "name": "meta_batch",
-          "shape": 7,
-          "type": "VHS_BatchManager",
-          "link": null
-        },
-        {
-          "name": "vae",
-          "shape": 7,
-          "type": "VAE",
-          "link": null
-        }
-      ],
-      "outputs": [
-        {
-          "name": "Filenames",
-          "type": "VHS_FILENAMES",
-          "links": null
-        }
-      ],
-      "properties": {
-        "cnr_id": "comfyui-videohelpersuite",
-        "ver": "0a75c7958fe320efcb052f1d9f8451fd20c730a8",
-        "Node name for S&R": "VHS_VideoCombine"
-      },
-      "widgets_values": {
-        "frame_rate": 25,
-        "loop_count": 0,
-        "filename_prefix": "WanVideo2_1_InfiniteTalk",
-        "format": "video/h264-mp4",
-        "pix_fmt": "yuv420p",
-        "crf": 19,
-        "save_metadata": true,
-        "trim_to_audio": false,
-        "pingpong": false,
-        "save_output": false,
-        "videopreview": {
-          "hidden": false,
-          "paused": false,
-          "params": {
-            "filename": "WanVideo2_1_InfiniteTalk_00004-audio.mp4",
-            "subfolder": "",
-            "type": "temp",
-            "format": "video/h264-mp4",
-            "frame_rate": 25,
-            "workflow": "WanVideo2_1_InfiniteTalk_00004.png",
-            "fullpath": "N:\\AI\\ComfyUI\\temp\\WanVideo2_1_InfiniteTalk_00004-audio.mp4"
-          }
-        }
-      }
-    },
     {
       "id": 254,
       "type": "GetNode",
       "pos": [
-        3683.283935546875,
-        -1643.584228515625
+        3663.662841796875,
+        -1884.486328125
       ],
       "size": [
         210,
@@ -1281,7 +1047,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 16,
+      "order": 13,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1315,7 +1081,7 @@
       "flags": {
         "collapsed": false
       },
-      "order": 37,
+      "order": 36,
       "mode": 0,
       "inputs": [
         {
@@ -1363,174 +1129,6 @@
       },
       "widgets_values": []
     },
-    {
-      "id": 241,
-      "type": "WanVideoTextEncodeCached",
-      "pos": [
-        2281.288330078125,
-        -1946.174072265625
-      ],
-      "size": [
-        465.5179138183594,
-        386.354248046875
-      ],
-      "flags": {},
-      "order": 17,
-      "mode": 0,
-      "inputs": [
-        {
-          "name": "extender_args",
-          "shape": 7,
-          "type": "WANVIDEOPROMPTEXTENDER_ARGS",
-          "link": null
-        }
-      ],
-      "outputs": [
-        {
-          "name": "text_embeds",
-          "type": "WANVIDEOTEXTEMBEDS",
-          "links": [
-            437
-          ]
-        },
-        {
-          "name": "negative_text_embeds",
-          "type": "WANVIDEOTEXTEMBEDS",
-          "links": null
-        },
-        {
-          "name": "positive_prompt",
-          "type": "STRING",
-          "links": null
-        }
-      ],
-      "properties": {
-        "cnr_id": "ComfyUI-WanVideoWrapper",
-        "ver": "ff779c91714d8ee3484cd4119b082c72a1734b72",
-        "Node name for S&R": "WanVideoTextEncodeCached"
-      },
-      "widgets_values": [
-        "umt5-xxl-enc-bf16.safetensors",
-        "bf16",
-        "a man is talking",
-        "bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards",
-        "disabled",
-        false,
-        "gpu"
-      ],
-      "color": "#432",
-      "bgcolor": "#653"
-    },
-    {
-      "id": 192,
-      "type": "WanVideoImageToVideoMultiTalk",
-      "pos": [
-        2836.906982421875,
-        -1288.3162841796875
-      ],
-      "size": [
-        349.6015625,
-        266
-      ],
-      "flags": {},
-      "order": 41,
-      "mode": 0,
-      "inputs": [
-        {
-          "name": "vae",
-          "type": "WANVAE",
-          "link": 440
-        },
-        {
-          "name": "start_image",
-          "shape": 7,
-          "type": "IMAGE",
-          "link": 540
-        },
-        {
-          "name": "clip_embeds",
-          "shape": 7,
-          "type": "WANVIDIMAGE_CLIPEMBEDS",
-          "link": 433
-        },
-        {
-          "name": "width",
-          "type": "INT",
-          "widget": {
-            "name": "width"
-          },
-          "link": 516
-        },
-        {
-          "name": "height",
-          "type": "INT",
-          "widget": {
-            "name": "height"
-          },
-          "link": 517
-        }
-      ],
-      "outputs": [
-        {
-          "name": "image_embeds",
-          "type": "WANVIDIMAGE_EMBEDS",
-          "links": [
-            413
-          ]
-        }
-      ],
-      "properties": {
-        "cnr_id": "ComfyUI-WanVideoWrapper",
-        "ver": "49430f900bf812a1e281560bafafd04c43dfdf13",
-        "Node name for S&R": "WanVideoImageToVideoMultiTalk"
-      },
-      "widgets_values": [
-        832,
-        480,
-        81,
-        9,
-        false,
-        "disabled",
-        false,
-        "infinitetalk"
-      ],
-      "color": "#322",
-      "bgcolor": "#533"
-    },
-    {
-      "id": 244,
-      "type": "GetNode",
-      "pos": [
-        2825.2236328125,
-        -1325.1824951171875
-      ],
-      "size": [
-        210,
-        34
-      ],
-      "flags": {
-        "collapsed": true
-      },
-      "order": 18,
-      "mode": 0,
-      "inputs": [],
-      "outputs": [
-        {
-          "name": "WANVAE",
-          "type": "WANVAE",
-          "links": [
-            440
-          ]
-        }
-      ],
-      "title": "Get_VAE",
-      "properties": {},
-      "widgets_values": [
-        "VAE"
-      ],
-      "color": "#322",
-      "bgcolor": "#533"
-    },
     {
       "id": 283,
       "type": "GetNode",
@@ -1545,7 +1143,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 19,
+      "order": 14,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1577,7 +1175,7 @@
         314
       ],
       "flags": {},
-      "order": 20,
+      "order": 15,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1616,7 +1214,7 @@
         336
       ],
       "flags": {},
-      "order": 34,
+      "order": 32,
       "mode": 0,
       "inputs": [
         {
@@ -1699,7 +1297,7 @@
         88
       ],
       "flags": {},
-      "order": 21,
+      "order": 16,
       "mode": 0,
       "inputs": [],
       "outputs": [],
@@ -1724,7 +1322,7 @@
       "flags": {
         "collapsed": true
       },
-      "order": 35,
+      "order": 33,
       "mode": 0,
       "inputs": [
         {
@@ -1764,7 +1362,7 @@
         136
       ],
       "flags": {},
-      "order": 22,
+      "order": 17,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1799,7 +1397,7 @@
         106
       ],
       "flags": {},
-      "order": 23,
+      "order": 18,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1834,7 +1432,7 @@
         106
       ],
       "flags": {},
-      "order": 24,
+      "order": 19,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1867,7 +1465,7 @@
         46
       ],
       "flags": {},
-      "order": 38,
+      "order": 37,
       "mode": 0,
       "inputs": [
         {
@@ -1899,7 +1497,8 @@
         "cnr_id": "ComfyUI-MelBandRoFormer",
         "ver": "b68d9077815387b64d596f8c39607052b95b6eba",
         "Node name for S&R": "MelBandRoFormerSampler"
-      }
+      },
+      "widgets_values": []
     },
     {
       "id": 303,
@@ -1913,7 +1512,7 @@
         125.51774597167969
       ],
       "flags": {},
-      "order": 25,
+      "order": 20,
       "mode": 0,
       "inputs": [],
       "outputs": [],
@@ -1937,7 +1536,7 @@
         91.6759033203125
       ],
       "flags": {},
-      "order": 26,
+      "order": 21,
       "mode": 0,
       "inputs": [],
       "outputs": [],
@@ -1961,7 +1560,7 @@
         58
       ],
       "flags": {},
-      "order": 27,
+      "order": 22,
       "mode": 0,
       "inputs": [],
       "outputs": [
@@ -1982,6 +1581,364 @@
         "MelBandRoFormer\\MelBandRoformer_fp16.safetensors"
       ]
     },
+    {
+      "id": 120,
+      "type": "MultiTalkModelLoader",
+      "pos": [
+        592.73583984375,
+        -2520.13525390625
+      ],
+      "size": [
+        511.7737121582031,
+        66.51400756835938
+      ],
+      "flags": {},
+      "order": 23,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "model",
+          "type": "MULTITALKMODEL",
+          "links": [
+            355
+          ]
+        }
+      ],
+      "properties": {
+        "cnr_id": "ComfyUI-WanVideoWrapper",
+        "ver": "058286fc0f3b0651a2f6b68309df3f06e8332cc0",
+        "Node name for S&R": "MultiTalkModelLoader"
+      },
+      "widgets_values": [
+        "WanVideo\\InfiniteTalk\\Wan2_1-InfiniteTalk_Single_Q8.gguf"
+      ],
+      "color": "#223",
+      "bgcolor": "#335"
+    },
+    {
+      "id": 244,
+      "type": "GetNode",
+      "pos": [
+        2835.034423828125,
+        -1351.3436279296875
+      ],
+      "size": [
+        210,
+        34
+      ],
+      "flags": {
+        "collapsed": true
+      },
+      "order": 24,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "WANVAE",
+          "type": "WANVAE",
+          "links": [
+            440
+          ]
+        }
+      ],
+      "title": "Get_VAE",
+      "properties": {},
+      "widgets_values": [
+        "VAE"
+      ],
+      "color": "#322",
+      "bgcolor": "#533"
+    },
+    {
+      "id": 131,
+      "type": "VHS_VideoCombine",
+      "pos": [
+        3669.72705078125,
+        -1830.8634033203125
+      ],
+      "size": [
+        991.5499877929688,
+        1319.550048828125
+      ],
+      "flags": {},
+      "order": 46,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "link": 551
+        },
+        {
+          "name": "audio",
+          "shape": 7,
+          "type": "AUDIO",
+          "link": 449
+        },
+        {
+          "name": "meta_batch",
+          "shape": 7,
+          "type": "VHS_BatchManager",
+          "link": null
+        },
+        {
+          "name": "vae",
+          "shape": 7,
+          "type": "VAE",
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "Filenames",
+          "type": "VHS_FILENAMES",
+          "links": null
+        }
+      ],
+      "properties": {
+        "cnr_id": "comfyui-videohelpersuite",
+        "ver": "0a75c7958fe320efcb052f1d9f8451fd20c730a8",
+        "Node name for S&R": "VHS_VideoCombine"
+      },
+      "widgets_values": {
+        "frame_rate": 25,
+        "loop_count": 0,
+        "filename_prefix": "WanVideo2_1_InfiniteTalk",
+        "format": "video/h264-mp4",
+        "pix_fmt": "yuv420p",
+        "crf": 19,
+        "save_metadata": true,
+        "trim_to_audio": false,
+        "pingpong": false,
+        "save_output": false,
+        "videopreview": {
+          "hidden": false,
+          "paused": false,
+          "params": {
+            "filename": "WanVideo2_1_InfiniteTalk_00003-audio.mp4",
+            "subfolder": "",
+            "type": "temp",
+            "format": "video/h264-mp4",
+            "frame_rate": 25,
+            "workflow": "WanVideo2_1_InfiniteTalk_00003.png",
+            "fullpath": "N:\\AI\\ComfyUI\\temp\\WanVideo2_1_InfiniteTalk_00003-audio.mp4"
+          }
+        }
+      }
+    },
+    {
+      "id": 309,
+      "type": "WanVideoPassImagesFromSamples",
+      "pos": [
+        3642.054443359375,
+        -1970.1678466796875
+      ],
+      "size": [
+        321.5958557128906,
+        26
+      ],
+      "flags": {},
+      "order": 45,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "samples",
+          "type": "LATENT",
+          "link": 550
+        }
+      ],
+      "outputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "links": [
+            551
+          ]
+        }
+      ],
+      "properties": {
+        "cnr_id": "ComfyUI-WanVideoWrapper",
+        "ver": "e5955d83958f1c808c86e5be3804ca42b39ab3fa",
+        "Node name for S&R": "WanVideoPassImagesFromSamples"
+      }
+    },
+    {
+      "id": 241,
+      "type": "WanVideoTextEncodeCached",
+      "pos": [
+        2281.288330078125,
+        -1946.174072265625
+      ],
+      "size": [
+        465.5179138183594,
+        386.354248046875
+      ],
+      "flags": {},
+      "order": 25,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "extender_args",
+          "shape": 7,
+          "type": "WANVIDEOPROMPTEXTENDER_ARGS",
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "text_embeds",
+          "type": "WANVIDEOTEXTEMBEDS",
+          "links": [
+            437
+          ]
+        },
+        {
+          "name": "negative_text_embeds",
+          "type": "WANVIDEOTEXTEMBEDS",
+          "links": null
+        },
+        {
+          "name": "positive_prompt",
+          "type": "STRING",
+          "links": null
+        }
+      ],
+      "properties": {
+        "cnr_id": "ComfyUI-WanVideoWrapper",
+        "ver": "ff779c91714d8ee3484cd4119b082c72a1734b72",
+        "Node name for S&R": "WanVideoTextEncodeCached"
+      },
+      "widgets_values": [
+        "umt5-xxl-enc-bf16.safetensors",
+        "bf16",
+        "a woman is singing",
+        "bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards",
+        "disabled",
+        false,
+        "gpu"
+      ],
+      "color": "#432",
+      "bgcolor": "#653"
+    },
+    {
+      "id": 270,
+      "type": "INTConstant",
+      "pos": [
+        2655.538330078125,
+        -2261.7578125
+      ],
+      "size": [
+        210,
+        58
+      ],
+      "flags": {},
+      "order": 26,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "value",
+          "type": "INT",
+          "links": [
+            472
+          ]
+        }
+      ],
+      "title": "Max frames",
+      "properties": {
+        "cnr_id": "comfyui-kjnodes",
+        "ver": "e435e999e4b1a828a6b5f6d8f037e66f4a798324",
+        "Node name for S&R": "INTConstant"
+      },
+      "widgets_values": [
+        500
+      ],
+      "color": "#1b4669",
+      "bgcolor": "#29699c"
+    },
+    {
+      "id": 192,
+      "type": "WanVideoImageToVideoMultiTalk",
+      "pos": [
+        2836.906982421875,
+        -1288.3162841796875
+      ],
+      "size": [
+        349.6015625,
+        290
+      ],
+      "flags": {},
+      "order": 41,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "vae",
+          "type": "WANVAE",
+          "link": 440
+        },
+        {
+          "name": "start_image",
+          "shape": 7,
+          "type": "IMAGE",
+          "link": 540
+        },
+        {
+          "name": "clip_embeds",
+          "shape": 7,
+          "type": "WANVIDIMAGE_CLIPEMBEDS",
+          "link": 433
+        },
+        {
+          "name": "width",
+          "type": "INT",
+          "widget": {
+            "name": "width"
+          },
+          "link": 516
+        },
+        {
+          "name": "height",
+          "type": "INT",
+          "widget": {
+            "name": "height"
+          },
+          "link": 517
+        }
+      ],
+      "outputs": [
+        {
+          "name": "image_embeds",
+          "type": "WANVIDIMAGE_EMBEDS",
+          "links": [
+            413
+          ]
+        },
+        {
+          "name": "output_path",
+          "type": "STRING",
+          "links": []
+        }
+      ],
+      "properties": {
+        "cnr_id": "ComfyUI-WanVideoWrapper",
+        "ver": "49430f900bf812a1e281560bafafd04c43dfdf13",
+        "Node name for S&R": "WanVideoImageToVideoMultiTalk"
+      },
+      "widgets_values": [
+        832,
+        480,
+        81,
+        9,
+        false,
+        "disabled",
+        false,
+        "infinitetalk",
+        ""
+      ],
+      "color": "#322",
+      "bgcolor": "#533"
+    },
     {
       "id": 128,
       "type": "WanVideoSampler",
@@ -2104,7 +2061,7 @@
           "type": "LATENT",
           "slot_index": 0,
           "links": [
-            530
+            550
           ]
         },
         {
@@ -2134,6 +2091,29 @@
         -1,
         true
       ]
+    },
+    {
+      "id": 310,
+      "type": "Note",
+      "pos": [
+        2849.673095703125,
+        -937.4252319335938
+      ],
+      "size": [
+        340.64837646484375,
+        183.92144775390625
+      ],
+      "flags": {},
+      "order": 27,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "properties": {},
+      "widgets_values": [
+        "This node makes the sampling into a loop to process all the needed frames. Each window is the same size, if the frame count doesn't divide evenly the audio will be padded with silence and extra frames generated.\n\nBetween each window the frames have to be re-encoded, thus no latents are saved and end result is already decoded.\n\nYou can also set an output_path to save each window's result as images during the process."
+      ],
+      "color": "#432",
+      "bgcolor": "#653"
     }
   ],
   "links": [
@@ -2209,14 +2189,6 @@
       2,
       "WANVIDEOTEXTEMBEDS"
     ],
-    [
-      438,
-      242,
-      0,
-      130,
-      0,
-      "WANVAE"
-    ],
     [
       440,
       244,
@@ -2289,14 +2261,6 @@
       0,
       "*"
     ],
-    [
-      493,
-      130,
-      0,
-      131,
-      0,
-      "IMAGE"
-    ],
     [
       494,
       283,
@@ -2361,14 +2325,6 @@
       6,
       "INT"
     ],
-    [
-      530,
-      128,
-      0,
-      130,
-      1,
-      "LATENT"
-    ],
     [
       538,
       281,
@@ -2424,6 +2380,22 @@
       302,
       1,
       "AUDIO"
+    ],
+    [
+      550,
+      128,
+      0,
+      309,
+      0,
+      "LATENT"
+    ],
+    [
+      551,
+      309,
+      0,
+      131,
+      0,
+      "IMAGE"
     ]
   ],
   "groups": [
@@ -2470,13 +2442,13 @@
   "config": {},
   "extra": {
     "ds": {
-      "scale": 0.6115909044842021,
+      "scale": 0.6727499949326234,
       "offset": [
-        -305.72724541367705,
-        2418.5059970103503
+        -916.8927060576216,
+        2084.4270316397833
       ]
     },
-    "frontendVersion": "1.26.6",
+    "frontendVersion": "1.27.0",
     "node_versions": {
       "ComfyUI-WanVideoWrapper": "0a11c67a0c0062b534178920a0d6dcaa75e7b5fe",
       "comfy-core": "0.3.43",
diff --git a/multitalk/nodes.py b/multitalk/nodes.py
index dd2da93..b305237 100644
--- a/multitalk/nodes.py
+++ b/multitalk/nodes.py
@@ -6,6 +6,7 @@ import torch
 from ..utils import log, set_module_tensor_to_device
 import os
 import json
+import datetime
 
 script_directory = os.path.dirname(os.path.abspath(__file__))
 folder_paths.add_model_folder_path("wav2vec2", os.path.join(folder_paths.models_dir, "wav2vec2"))
@@ -389,17 +390,19 @@ class WanVideoImageToVideoMultiTalk:
                     "auto",
                     "multitalk",
                     "infinitetalk"
-                ], {"default": "auto", "tooltip": "The sampling strategy to use in the long video generation loop, should match the model used"})
+                ], {"default": "auto", "tooltip": "The sampling strategy to use in the long video generation loop, should match the model used"}),
+                "output_path": ("STRING", {"default": "", "tooltip": "If set, will save each window's resulting frames to this folder"}),
+
             }
         }
 
-    RETURN_TYPES = ("WANVIDIMAGE_EMBEDS",)
-    RETURN_NAMES = ("image_embeds",)
+    RETURN_TYPES = ("WANVIDIMAGE_EMBEDS", "STRING",)
+    RETURN_NAMES = ("image_embeds", "output_path")
     FUNCTION = "process"
     CATEGORY = "WanVideoWrapper"
     DESCRIPTION = "Enables Multi/InfiniteTalk long video generation sampling method, the video is created in windows with overlapping frames. Not compatible or necessary to be used with context windows and many other features besides Multi/InfiniteTalk."
 
-    def process(self, vae, width, height, frame_window_size, motion_frame, force_offload, colormatch, start_image=None, tiled_vae=False, clip_embeds=None, mode="multitalk"):
+    def process(self, vae, width, height, frame_window_size, motion_frame, force_offload, colormatch, start_image=None, tiled_vae=False, clip_embeds=None, mode="multitalk", output_path=""):
 
         H = height
         W = width
@@ -416,6 +419,11 @@ class WanVideoImageToVideoMultiTalk:
         target_shape = (16, (num_frames - 1) // VAE_STRIDE[0] + 1,
                         height // VAE_STRIDE[1],
                         width // VAE_STRIDE[2])
+        
+        if output_path:
+            timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
+            output_path = os.path.join(output_path, f"{timestamp}_{mode}_output")
+            os.makedirs(output_path, exist_ok=True)
 
         image_embeds = {
             "multitalk_sampling": True,
@@ -430,10 +438,11 @@ class WanVideoImageToVideoMultiTalk:
             "target_shape": target_shape,
             "clip_context": clip_embeds.get("clip_embeds", None) if clip_embeds is not None else None,
             "colormatch": colormatch,
-            "multitalk_mode": mode
+            "multitalk_mode": mode,
+            "output_path": output_path
         }
 
-        return (image_embeds,)
+        return (image_embeds, output_path)
     
 NODE_CLASS_MAPPINGS = {
     "MultiTalkModelLoader": MultiTalkModelLoader,
diff --git a/nodes.py b/nodes.py
index a44d710..7fc6d20 100644
--- a/nodes.py
+++ b/nodes.py
@@ -5,6 +5,7 @@ import numpy as np
 from tqdm import tqdm
 import inspect
 import copy
+from PIL import Image
 import hashlib
 from diffusers.schedulers import FlowMatchEulerDiscreteScheduler
 
@@ -3391,6 +3392,9 @@ class WanVideoSampler:
                         if original_images is None:
                             original_images = torch.zeros([noise.shape[0], 1, target_h, target_w], device=device)
 
+                        output_path = image_embeds.get("output_path", "")
+                        img_counter = 0
+
                         if len(multitalk_embeds['audio_features'])==2 and (multitalk_embeds['ref_target_masks'] is None):
                             face_scale = 0.1
                             x_min, x_max = int(target_h * face_scale), int(target_h * (1 - face_scale))
@@ -3746,7 +3750,17 @@ class WanVideoSampler:
                         
                                 videos = torch.stack(cm_result_list, dim=0).permute(3, 0, 1, 2)
 
-                            # cache generated samples
+                            # optionally save generated samples to disk
+                            if output_path:
+                                video_np = videos.clamp(-1.0, 1.0).add(1.0).div(2.0).mul(255).cpu().float().numpy().transpose(1, 2, 3, 0).astype('uint8')
+                                num_frames_to_save = video_np.shape[0] if is_first_clip else video_np.shape[0] - cur_motion_frames_num
+                                log.info(f"Saving {num_frames_to_save} generated frames to {output_path}")
+                                start_idx = 0 if is_first_clip else cur_motion_frames_num
+                                for i in range(start_idx, video_np.shape[0]):
+                                    im = Image.fromarray(video_np[i])
+                                    im.save(os.path.join(output_path, f"frame_{img_counter:05d}.png"))
+                                    img_counter += 1
+                                
                             gen_video_list.append(videos if is_first_clip else videos[:, cur_motion_frames_num:])
 
                             current_condframe_index += 1
diff --git a/nodes_utility.py b/nodes_utility.py
index bb4e408..bbb195d 100644
--- a/nodes_utility.py
+++ b/nodes_utility.py
@@ -448,6 +448,26 @@ class NormalizeAudioLoudness:
         normalized_audio = pyloudnorm.normalize.loudness(audio_array, loudness, lufs)
         return normalized_audio
     
+class WanVideoPassImagesFromSamples:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required": {
+                    "samples": ("LATENT",),
+                }
+                }
+
+    RETURN_TYPES = ("IMAGE",)
+    RETURN_NAMES = ("images",)
+    FUNCTION = "decode"
+    CATEGORY = "WanVideoWrapper"
+    DESCRIPTION = "Gets possible already decoded images from the samples dictionary, used with Multi/InfiniteTalk sampling"
+
+    def decode(self, samples):
+        video = samples.get("video", None)
+        video.clamp_(-1.0, 1.0)
+        video.add_(1.0).div_(2.0)
+        return video.cpu().float(),
+    
 NODE_CLASS_MAPPINGS = {
     "WanVideoImageResizeToClosest": WanVideoImageResizeToClosest,
     "WanVideoVACEStartToEndFrame": WanVideoVACEStartToEndFrame,
@@ -457,7 +477,8 @@ NODE_CLASS_MAPPINGS = {
     "WanVideoLatentReScale": WanVideoLatentReScale,
     "CreateScheduleFloatList": CreateScheduleFloatList,
     "WanVideoSigmaToStep": WanVideoSigmaToStep,
-    "NormalizeAudioLoudness": NormalizeAudioLoudness
+    "NormalizeAudioLoudness": NormalizeAudioLoudness,
+    "WanVideoPassImagesFromSamples": WanVideoPassImagesFromSamples,
 }
 NODE_DISPLAY_NAME_MAPPINGS = {
     "WanVideoImageResizeToClosest": "WanVideo Image Resize To Closest",
@@ -468,5 +489,6 @@ NODE_DISPLAY_NAME_MAPPINGS = {
     "WanVideoLatentReScale": "WanVideo Latent ReScale",
     "CreateScheduleFloatList": "Create Schedule Float List",
     "WanVideoSigmaToStep": "WanVideo Sigma To Step",
-    "NormalizeAudioLoudness": "Normalize Audio Loudness"
+    "NormalizeAudioLoudness": "Normalize Audio Loudness",
+    "WanVideoPassImagesFromSamples": "WanVideo Pass Images From Samples",
 }
\ No newline at end of file