valknar/runpod
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: completely rebuild CogVideoX I2V workflow with correct configurations

Browse Source 
Major fixes:
- Replace DualCLIPLoader with CLIPLoader using t5xxl_fp16.safetensors
- Fix CogVideoSampler parameter order: [num_frames, steps, cfg, seed, control, scheduler, denoise]
- Fix CogVideoImageEncode input: 'image' -> 'start_image'
- Remove CogVideoXVAELoader, use VAE directly from DownloadAndLoadCogVideoModel
- Add CogVideoTextEncode strength and force_offload parameters
- Simplify to 8 nodes (removed node 10)
- All nodes properly connected with correct link IDs

Version: 1.2.0
Tested against: ComfyUI-CogVideoXWrapper example workflows
This commit is contained in:
Sebastian Krüger
2025-11-23 09:41:01 +01:00
parent 80a81aa12f
commit 47824ab987
1 changed files with 30 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

81
comfyui/workflows/image-to-video/cogvideox-i2v-production-v1.json
View File

@@ -1,6 +1,6 @@
{
"last_node_id": 10,
"last_link_id": 14,
"last_node_id": 8,
"last_link_id": 10,
"nodes": [
{
"id": 1,
@@ -52,23 +52,23 @@
{
"name": "vae",
"type": "VAE",
"links": [3],
"links": [3, 10],
"slot_index": 1
}
]
},
{
"id": 7,
"type": "DualCLIPLoader",
"type": "CLIPLoader",
"pos": [50, 650],
"size": [350, 100],
"widgets_values": ["t5xxl_fp16.safetensors", "clip_l.safetensors", "flux"],
"title": "CLIP Loader",
"widgets_values": ["t5xxl_fp16.safetensors", "sd3"],
"title": "T5 CLIP Loader",
"flags": {},
"order": 2,
"mode": 0,
"properties": {
"Node name for S&R": "DualCLIPLoader"
"Node name for S&R": "CLIPLoader"
},
"outputs": [
{
@@ -84,7 +84,7 @@
"type": "CogVideoTextEncode",
"pos": [450, 100],
"size": [400, 200],
"widgets_values": ["Camera movement description, action, scene details"],
"widgets_values": ["Camera movement description, action, scene details", 1, false],
"title": "API Video Prompt (Positive)",
"flags": {},
"order": 3,
@@ -118,7 +118,7 @@
"type": "CogVideoTextEncode",
"pos": [450, 350],
"size": [400, 200],
"widgets_values": ["low quality, blurry, distorted"],
"widgets_values": ["low quality, blurry, distorted, watermark", 1, true],
"title": "API Video Prompt (Negative)",
"flags": {},
"order": 4,
@@ -167,7 +167,7 @@
"link": 3
},
{
"name": "image",
"name": "start_image",
"type": "IMAGE",
"link": 1
}
@@ -186,7 +186,7 @@
"type": "CogVideoSampler",
"pos": [900, 100],
"size": [315, 474],
"widgets_values": [42, "CogVideoXDDIM", 49, 50, 6.0],
"widgets_values": [49, 50, 6.0, 42, "fixed", "CogVideoXDDIM", 1.0],
"title": "CogVideoX Sampler (6s @ 8fps)",
"flags": {},
"order": 6,
@@ -259,28 +259,6 @@
}
]
},
{
"id": 10,
"type": "CogVideoXVAELoader",
"pos": [900, 600],
"size": [315, 100],
"widgets_values": ["THUDM/CogVideoX-5b-I2V"],
"title": "VAE Loader",
"flags": {},
"order": 8,
"mode": 0,
"properties": {
"Node name for S&R": "CogVideoXVAELoader"
},
"outputs": [
{
"name": "vae",
"type": "VAE",
"links": [10],
"slot_index": 0
}
]
},
{
"id": 6,
"type": "VHS_VideoCombine",
@@ -289,7 +267,7 @@
"widgets_values": [8, 0, "cogvideox_output", "video/h264-mp4", "yuv420p", 19, true, false],
"title": "Combine Video Frames",
"flags": {},
"order": 9,
"order": 8,
"mode": 0,
"properties": {
"Node name for S&R": "VHS_VideoCombine"
@@ -320,7 +298,7 @@
[7, 8, 0, 4, 2, "CONDITIONING"],
[8, 9, 0, 4, 3, "LATENT"],
[9, 4, 0, 5, 1, "LATENT"],
[10, 10, 0, 5, 0, "VAE"],
[10, 2, 1, 5, 0, "VAE"],
[11, 5, 0, 6, 0, "IMAGE"]
],
"groups": [],
@@ -328,12 +306,12 @@
"extra": {
"workflow_info": {
"name": "CogVideoX Image-to-Video Production",
"version": "1.1.0",
"version": "1.2.0",
"description": "AI-driven image-to-video using CogVideoX-5b-I2V. Generate 6-second videos (49 frames @ 8fps) from input images with camera movement and action.",
"category": "image-to-video",
"tags": ["cogvideox", "i2v", "video-generation", "production"],
"requirements": {
"models": ["CogVideoX-5b-I2V", "t5xxl", "clip_l"],
"models": ["CogVideoX-5b-I2V", "T5-XXL FP16"],
"custom_nodes": ["ComfyUI-VideoHelperSuite", "ComfyUI-CogVideoXWrapper"],
"vram_min": "20GB",
"vram_recommended": "24GB"
@@ -359,20 +337,12 @@
"widget_index": 0,
"type": "string",
"required": false,
"default": "low quality, blurry, distorted",
"default": "low quality, blurry, distorted, watermark",
"description": "Undesired elements to avoid"
},
"seed": {
"node_id": 4,
"widget_index": 0,
"type": "integer",
"required": false,
"default": 42,
"description": "Random seed for reproducibility"
},
"num_frames": {
"node_id": 4,
"widget_index": 2,
"widget_index": 0,
"type": "integer",
"required": false,
"default": 49,
@@ -382,7 +352,7 @@
},
"steps": {
"node_id": 4,
"widget_index": 3,
"widget_index": 1,
"type": "integer",
"required": false,
"default": 50,
@@ -392,14 +362,22 @@
},
"cfg": {
"node_id": 4,
"widget_index": 4,
"widget_index": 2,
"type": "float",
"required": false,
"default": 6.0,
"min": 1.0,
"max": 15.0,
"max": 30.0,
"description": "Classifier-free guidance scale"
},
"seed": {
"node_id": 4,
"widget_index": 3,
"type": "integer",
"required": false,
"default": 42,
"description": "Random seed for reproducibility"
},
"fps": {
"node_id": 6,
"widget_index": 0,
@@ -433,7 +411,8 @@
"CogVideoX-5b-I2V is specifically trained for image-to-video generation",
"Model will download automatically on first use (~10GB)",
"Enable VAE tiling to reduce VRAM usage",
"Higher steps (50-100) improve quality but increase generation time"
"Higher steps (50-100) improve quality but increase generation time",
"T5-XXL text encoder required - automatically linked from SD3.5"
]
}
},