LTX-2 Canny to Video
Generate high-quality videos from edge detection (Canny) guidance with synchronized audio-video generation using LTX-2. Features precise edge control, structural consistency, and efficient performance.
ワークフローをダウンロードCloudで試す API ドキュメント
概要
The 'LTX-2 Canny to Video' workflow is designed to transform edge-detected images into high-quality videos with synchronized audio. Utilizing the LTX-2 model, this workflow leverages the Canny edge detection technique to ensure precise edge control and structural consistency throughout the video. The process begins with the 'LoadVideo' node to import your source video, which is then decomposed into its components using the 'GetVideoComponents' node. The 'Canny' node plays a crucial role by detecting edges in the images, providing a guide for the video generation process.
This workflow is particularly useful for creators who need to maintain the structural integrity of the original footage while enhancing it with AI-generated elements. The LTX-2 model, developed by Lightricks, ensures that the generated video maintains high quality and coherence. The workflow also includes nodes like 'ImageScaleBy' and 'ResizeImageMaskNode' for adjusting image dimensions, and 'SaveVideo' to compile and save the final video. By following the recommended settings and parameters, users can achieve optimal results with minimal manual intervention.
入力するもの
- ワークフローを読み込み中
ComfyUI で実行できます。 このワークフローはこのページでは実行できません。すべてが読み込まれた状態でComfy Cloudで開くか、ご自身のComfyUI用にダウンロードしてください。
Cloudで試す例
ワークフローの内部
ノードグラフの読み取り専用プレビューです。ノードを編集するには、Comfy Cloudで開くかダウンロードしてください。
Loading the graph
Open full-size workflow preview- 作成者
- ComfyUI
カスタムノードが必要です
- ワークフローを読み込み中
コードからこのワークフローを使用する
可能な場合はワークフローグラフと入力から生成されます。それ以外の場合は保存された例が表示されます。料金が発生する場合があるため、コードを実行する前に認証情報とエンドポイントを確認してください。
1 API キーを取得 ↗
What you need
- Runs on
- Comfy Cloud
- キー
- COMFY_API_KEY
有料の Cloud プランとクレジットが必要です。
API JSONをダウンロードAPI ドキュメント ↗
# Python 3.10+: pip install comfy-sdk==0.4.0
import os
from comfy_sdk import Comfy
api_key = os.environ.get("COMFY_API_KEY", "").strip()
if not api_key: raise RuntimeError("Set COMFY_API_KEY")
os.environ.setdefault("COMFY_BASE_URL", "https://cloud.comfy.org")
client = Comfy(api_key=api_key)
workflow = client.workflows.from_json({
"127": {
"class_type": "LoadVideo",
"inputs": {
"file": "squirrel_in_flower_garden.mp4"
},
"_meta": {
"title": "Load Video"
}
},
"144": {
"class_type": "GetVideoComponents",
"inputs": {
"video": [
"127",
0
]
},
"_meta": {
"title": "Get Video Components"
}
},
"146": {
"class_type": "ImageScaleBy",
"inputs": {
"image": [
"164",
0
],
"scale_by": 0.5,
"upscale_method": "lanczos"
},
"_meta": {
"title": "Upscale Image By"
}
},
"152": {
"class_type": "ImageFromBatch",
"inputs": {
"batch_index": 0,
"image": [
"144",
0
],
"length": [
"168",
0
]
},
"_meta": {
"title": "Get Image from Batch"
}
},
"158": {
"class_type": "Canny",
"inputs": {
"high_threshold": 0.8,
"image": [
"146",
0
],
"low_threshold": 0.4
},
"_meta": {
"title": "Detect Edges (Canny)"
}
},
"160": {
"class_type": "LoadImage",
"inputs": {
"image": "ltx_2_canny_1st_frame.png"
},
"_meta": {
"title": "Load Image"
}
},
"164": {
"class_type": "ResizeImageMaskNode",
"inputs": {
"input": [
"152",
0
],
"resize_type": "scale dimensions",
"resize_type.crop": "center",
"resize_type.height": [
"167",
0
],
"resize_type.width": [
"166",
0
],
"scale_method": "lanczos"
},
"_meta": {
"title": "Resize Image/Mask"
}
},
"165": {
"class_type": "ResizeImageMaskNode",
"inputs": {
"input": [
"160",
0
],
"resize_type": "scale dimensions",
"resize_type.crop": "center",
"resize_type.height": [
"167",
0
],
"resize_type.width": [
"166",
0
],
"scale_method": "lanczos"
},
"_meta": {
"title": "Resize Image/Mask"
}
},
"166": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 1280
},
"_meta": {
"title": "Width"
}
},
"167": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 720
},
"_meta": {
"title": "Height"
}
},
"168": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 121
},
"_meta": {
"title": "Frame Count"
}
},
"169": {
"class_type": "PrimitiveBoolean",
"inputs": {
"value": False
},
"_meta": {
"title": "Ignore First Frame？"
}
},
"175": {
"class_type": "SaveVideo",
"inputs": {
"filename_prefix": "video/LTX_2.0",
"format": "auto",
"format.codec": "auto",
"video": [
"143:106",
0
]
},
"_meta": {
"title": "Save Video"
}
},
"143:100": {
"class_type": "LatentUpscaleModelLoader",
"inputs": {
"model_name": "ltx-2-spatial-upscaler-x2-1.0.safetensors"
},
"_meta": {
"title": "Load Latent Upscale Model"
}
},
"143:101": {
"class_type": "LTXVConcatAVLatent",
"inputs": {
"audio_latent": [
"143:122",
1
],
"video_latent": [
"143:151",
0
]
},
"_meta": {
"title": "Concat AV Latent"
}
},
"143:103": {
"class_type": "CheckpointLoaderSimple",
"inputs": {
"ckpt_name": "ltx-2-19b-dev-fp8.safetensors"
},
"_meta": {
"title": "Load Checkpoint"
}
},
"143:104": {
"class_type": "LTXVCropGuides",
"inputs": {
"latent": [
"143:122",
0
],
"negative": [
"143:132",
1
],
"positive": [
"143:132",
0
]
},
"_meta": {
"title": "LTXVCropGuides"
}
},
"143:105": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "ltx-2-19b-distilled-lora-384.safetensors",
"model": [
"143:134",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"143:106": {
"class_type": "CreateVideo",
"inputs": {
"audio": [
"143:139",
0
],
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": [
"143:148",
0
],
"images": [
"143:144",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"143:107": {
"class_type": "SamplerCustomAdvanced",
"inputs": {
"guider": [
"143:108",
0
],
"latent_image": [
"143:101",
0
],
"noise": [
"143:143",
0
],
"sampler": [
"143:98",
0
],
"sigmas": [
"143:99",
0
]
},
"_meta": {
"title": "SamplerCustomAdvanced"
}
},
"143:108": {
"class_type": "CFGGuider",
"inputs": {
"cfg": 1,
"model": [
"143:105",
0
],
"negative": [
"143:104",
1
],
"positive": [
"143:104",
0
]
},
"_meta": {
"title": "CFG Guider"
}
},
"143:110": {
"class_type": "GetImageSize",
"inputs": {
"image": [
"158",
0
]
},
"_meta": {
"title": "Get Image Size"
}
},
"143:111": {
"class_type": "LTXVEmptyLatentAudio",
"inputs": {
"audio_vae": [
"143:96",
0
],
"batch_size": 1,
"frame_rate": [
"143:145",
0
],
"frames_number": [
"143:110",
2
]
},
"_meta": {
"title": "Empty LTXV Latent Audio"
}
},
"143:112": {
"class_type": "LTXVLatentUpsampler",
"inputs": {
"samples": [
"143:104",
2
],
"upscale_model": [
"143:100",
0
],
"vae": [
"143:103",
2
]
},
"_meta": {
"title": "spatial"
}
},
"143:113": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"143:138",
0
],
"vae": [
"143:103",
2
]
},
"_meta": {
"title": "VAE Decode"
}
},
"143:114": {
"class_type": "LTXVConditioning",
"inputs": {
"frame_rate": [
"143:148",
0
],
"negative": [
"143:119",
0
],
"positive": [
"143:124",
0
]
},
"_meta": {
"title": "LTXVConditioning"
}
},
"143:115": {
"class_type": "EmptyLTXVLatentVideo",
"inputs": {
"batch_size": 1,
"height": [
"143:110",
1
],
"length": [
"143:110",
2
],
"width": [
"143:110",
0
]
},
"_meta": {
"title": "Empty LTXV Latent Video"
}
},
"143:116": {
"class_type": "LTXVConcatAVLatent",
"inputs": {
"audio_latent": [
"143:111",
0
],
"video_latent": [
"143:132",
2
]
},
"_meta": {
"title": "Concat AV Latent"
}
},
"143:119": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"143:97",
0
],
"text": "blurry, low quality, still frame, frames, watermark, overlay, titles, has blurbox, has subtitles"
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"143:122": {
"class_type": "LTXVSeparateAVLatent",
"inputs": {
"av_latent": [
"143:123",
0
]
},
"_meta": {
"title": "Separate AV Latent"
}
},
"143:123": {
"class_type": "SamplerCustomAdvanced",
"inputs": {
"guider": [
"143:93",
0
],
"latent_image": [
"143:116",
0
],
"noise": [
"143:126",
0
],
"sampler": [
"143:94",
0
],
"sigmas": [
"143:95",
0
]
},
"_meta": {
"title": "SamplerCustomAdvanced"
}
},
"143:124": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"143:97",
0
],
"text": "A wide shot reveals a whimsical outdoor scene with a decorative tree-like structure crafted from brown branches. Golden spherical ornaments and delicate white artificial flowers adorn the branches, catching bright daylight. A small grey and white squirrel stands on a round, light-colored wooden platform at the base, its bushy tail slightly raised. The squirrel dips its head down, nibbling at something on the platform. The camera slowly pushes in, focusing on the squirrel's gentle movements. Soft forest sounds fill the air—rustling leaves, distant birdsong, and a gentle breeze. The golden ornaments shimmer in the sunlight as the squirrel continues to eat."
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"143:126": {
"class_type": "RandomNoise",
"inputs": {
"noise_seed": 0
},
"_meta": {
"title": "RandomNoise"
}
},
"143:132": {
"class_type": "LTXVAddGuide",
"inputs": {
"frame_idx": 0,
"image": [
"158",
0
],
"latent": [
"143:149",
0
],
"negative": [
"143:114",
1
],
"positive": [
"143:114",
0
],
"strength": 1,
"vae": [
"143:103",
2
]
},
"_meta": {
"title": "LTXVAddGuide"
}
},
"143:134": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "ltx-2-19b-ic-lora-canny-control.safetensors",
"model": [
"143:103",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"143:138": {
"class_type": "LTXVSeparateAVLatent",
"inputs": {
"av_latent": [
"143:107",
1
]
},
"_meta": {
"title": "Separate AV Latent"
}
},
"143:139": {
"class_type": "LTXVAudioVAEDecode",
"inputs": {
"audio_vae": [
"143:96",
0
],
"samples": [
"143:138",
1
]
},
"_meta": {
"title": "LTXV Audio VAE Decode"
}
},
"143:143": {
"class_type": "RandomNoise",
"inputs": {
"noise_seed": 0
},
"_meta": {
"title": "RandomNoise"
}
},
"143:144": {
"class_type": "VAEDecodeTiled",
"inputs": {
"overlap": 64,
"samples": [
"143:138",
0
],
"temporal_overlap": 8,
"temporal_size": 4096,
"tile_size": 512,
"vae": [
"143:103",
2
]
},
"_meta": {
"title": "VAE Decode (Tiled)"
}
},
"143:145": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 24
},
"_meta": {
"title": "Int"
}
},
"143:148": {
"class_type": "PrimitiveFloat",
"inputs": {
"value": 24
},
"_meta": {
"title": "Float"
}
},
"143:149": {
"class_type": "LTXVImgToVideoInplace",
"inputs": {
"bypass": [
"169",
0
],
"image": [
"165",
0
],
"latent": [
"143:115",
0
],
"strength": 1,
"vae": [
"143:103",
2
]
},
"_meta": {
"title": "LTXVImgToVideoInplace"
}
},
"143:151": {
"class_type": "LTXVImgToVideoInplace",
"inputs": {
"bypass": [
"169",
0
],
"image": [
"165",
0
],
"latent": [
"143:112",
0
],
"strength": 1,
"vae": [
"143:103",
2
]
},
"_meta": {
"title": "LTXVImgToVideoInplace"
}
},
"143:93": {
"class_type": "CFGGuider",
"inputs": {
"cfg": 3,
"model": [
"143:134",
0
],
"negative": [
"143:132",
1
],
"positive": [
"143:132",
0
]
},
"_meta": {
"title": "CFG Guider"
}
},
"143:94": {
"class_type": "KSamplerSelect",
"inputs": {
"sampler_name": "euler"
},
"_meta": {
"title": "KSamplerSelect"
}
},
"143:95": {
"class_type": "LTXVScheduler",
"inputs": {
"base_shift": 0.95,
"latent": [
"143:116",
0
],
"max_shift": 2.05,
"steps": 20,
"stretch": True,
"terminal": 0.1
},
"_meta": {
"title": "LTXVScheduler"
}
},
"143:96": {
"class_type": "LTXVAudioVAELoader",
"inputs": {
"ckpt_name": "ltx-2-19b-dev-fp8.safetensors"
},
"_meta": {
"title": "Load LTXV Audio VAE"
}
},
"143:97": {
"class_type": "LTXAVTextEncoderLoader",
"inputs": {
"ckpt_name": "ltx-2-19b-dev-fp8.safetensors",
"device": "default",
"text_encoder": "gemma_3_12B_it_fp4_mixed.safetensors"
},
"_meta": {
"title": "Load LTXV Audio Text Encoder"
}
},
"143:98": {
"class_type": "KSamplerSelect",
"inputs": {
"sampler_name": "gradient_estimation"
},
"_meta": {
"title": "KSamplerSelect"
}
},
"143:99": {
"class_type": "ManualSigmas",
"inputs": {
"sigmas": "0.909375, 0.725, 0.421875, 0.0"
},
"_meta": {
"title": "ManualSigmas"
}
}
})
# run() also passes api_key to the partner nodes in this workflow.
job = client.run(workflow, api_key=api_key)よくある質問
The Canny node detects edges in video frames, providing a guide for the LTX-2 model to generate videos with precise edge control.
Yes, you can use high-resolution settings like 1920x1088, but ensure that width and height are divisible by 64, and frame count by 8+1.
The workflow will not produce errors but will adjust to the closest valid parameters to ensure smooth operation.
The LTX-2 model maintains structural consistency and enhances visual details, ensuring high-quality video output.