LTX-2.5: FLF2V
Generate a video from first and last frame reference images using LTX-2.5, blending both frames into a single animated sequence. Ideal for creating transition shots, morphing between two subjects or scenes, and building multi-shot narratives from still references.
查看工作流内部
源模板的只读预览。在 Cloud 中打开，即可查看节点并进行修改。
Loading the graph
Open full-size workflow preview- 输出
- 视频
- 作者
- ComfyUI
从代码中调用该模型
以下请求使用与体验区相同的预设图和设置。每次提交都可能开始一次付费运行；如果提交结果未知，请勿自动重试。
What you need
- Runs on
- Comfy Cloud
- 密钥
- COMFY_API_KEY
- 你的文件
- 由代码根据链接上传
需要付费的 Cloud 方案和额度。
# 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({
"31": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_first_frame_FILENAME"
},
"_meta": {
"title": "Load First Frame"
}
},
"39": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_last_frame_FILENAME"
},
"_meta": {
"title": "Load Last Frame"
}
},
"68": {
"class_type": "SaveVideo",
"inputs": {
"filename_prefix": "video/ltx2.5_flf2v",
"format": "auto",
"format.codec": "auto",
"video": [
"251:218",
0
]
},
"_meta": {
"title": "Save Video"
}
},
"251:195": {
"class_type": "LTXVPreprocess",
"inputs": {
"image": [
"251:214",
0
],
"img_compression": 18
},
"_meta": {
"title": "LTXV Preprocess"
}
},
"251:196": {
"class_type": "RandomNoise",
"inputs": {
"noise_seed": 873293130933086
},
"_meta": {
"title": "RandomNoise"
}
},
"251:197": {
"class_type": "LTXVEmptyLatentAudio",
"inputs": {
"audio_vae": [
"251:227",
0
],
"batch_size": 1,
"frame_rate": [
"251:205",
0
],
"frames_number": [
"251:226",
1
]
},
"_meta": {
"title": "Empty LTXV Latent Audio"
}
},
"251:198": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 5
},
"_meta": {
"title": "Duration"
}
},
"251:199": {
"class_type": "LTXVPreprocess",
"inputs": {
"image": [
"251:213",
0
],
"img_compression": 18
},
"_meta": {
"title": "LTXV Preprocess"
}
},
"251:200": {
"class_type": "LTXVCropGuides",
"inputs": {
"latent": [
"251:221",
0
],
"negative": [
"251:204",
1
],
"positive": [
"251:204",
0
]
},
"_meta": {
"title": "LTXVCropGuides"
}
},
"251:201": {
"class_type": "EmptyLTXVLatentVideo",
"inputs": {
"batch_size": 1,
"height": [
"251:203",
1
],
"length": [
"251:226",
1
],
"width": [
"251:203",
0
]
},
"_meta": {
"title": "Empty LTXV Latent Video"
}
},
"251:202": {
"class_type": "LTXVConditioning",
"inputs": {
"frame_rate": [
"251:212",
0
],
"negative": [
"251:217",
0
],
"positive": [
"251:222",
0
]
},
"_meta": {
"title": "LTXVConditioning"
}
},
"251:203": {
"class_type": "GetImageSize",
"inputs": {
"image": [
"251:213",
0
]
},
"_meta": {
"title": "Get Image Size"
}
},
"251:204": {
"class_type": "LTXVAddGuide",
"inputs": {
"frame_idx": -1,
"image": [
"251:195",
0
],
"latent": [
"251:206",
2
],
"negative": [
"251:206",
1
],
"positive": [
"251:206",
0
],
"strength": 0.7,
"vae": [
"251:229",
0
]
},
"_meta": {
"title": "LTXVAddGuide"
}
},
"251:205": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 24
},
"_meta": {
"title": "Frame Rate(int)"
}
},
"251:206": {
"class_type": "LTXVAddGuide",
"inputs": {
"frame_idx": 0,
"image": [
"251:199",
0
],
"latent": [
"251:201",
0
],
"negative": [
"251:202",
1
],
"positive": [
"251:202",
0
],
"strength": 0.7,
"vae": [
"251:229",
0
]
},
"_meta": {
"title": "LTXVAddGuide"
}
},
"251:208": {
"class_type": "SamplerEulerAncestral",
"inputs": {
"eta": 0,
"s_noise": 1
},
"_meta": {
"title": "SamplerEulerAncestral"
}
},
"251:210": {
"class_type": "LTXVConcatAVLatent",
"inputs": {
"audio_latent": [
"251:197",
0
],
"video_latent": [
"251:204",
2
]
},
"_meta": {
"title": "Concat AV Latent"
}
},
"251:211": {
"class_type": "SamplerCustomAdvanced",
"inputs": {
"guider": [
"251:235",
0
],
"latent_image": [
"251:210",
0
],
"noise": [
"251:196",
0
],
"sampler": [
"251:208",
0
],
"sigmas": [
"251:239",
0
]
},
"_meta": {
"title": "SamplerCustomAdvanced"
}
},
"251:212": {
"class_type": "ComfyMathExpression",
"inputs": {
"expression": "a",
"values.a": [
"251:205",
0
]
},
"_meta": {
"title": "Math Expression"
}
},
"251:213": {
"class_type": "ResizeImageMaskNode",
"inputs": {
"input": [
"31",
0
],
"resize_type": "scale dimensions",
"resize_type.crop": "center",
"resize_type.height": [
"251:216",
0
],
"resize_type.width": [
"251:215",
0
],
"scale_method": "nearest-exact"
},
"_meta": {
"title": "Resize Image/Mask"
}
},
"251:214": {
"class_type": "ResizeImageMaskNode",
"inputs": {
"input": [
"39",
0
],
"resize_type": "scale dimensions",
"resize_type.crop": "center",
"resize_type.height": [
"251:216",
0
],
"resize_type.width": [
"251:215",
0
],
"scale_method": "nearest-exact"
},
"_meta": {
"title": "Resize Image/Mask"
}
},
"251:215": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 1280
},
"_meta": {
"title": "Width"
}
},
"251:216": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 720
},
"_meta": {
"title": "height"
}
},
"251:217": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"251:228",
0
],
"text": "blurry, out of focus, overexposed, underexposed, low contrast, washed out colors, excessive noise, grainy texture, poor lighting, flickering, motion blur, distorted proportions, unnatural skin tones, deformed facial features, asymmetrical face, missing facial features, extra limbs, disfigured hands, wrong hand count, artifacts around text, unreadable text on shirt or hat, incorrect lettering on cap (“PNTR”), incorrect t-shirt slogan (“JUST DO IT”), missing microphone, misplaced microphone, inconsistent perspective, camera shake, incorrect depth of field, background too sharp, background clutter, distracting reflections, harsh shadows, inconsistent lighting direction, color banding, cartoonish rendering, 3D CGI look, unrealistic materials, uncanny valley effect, incorrect ethnicity, wrong gender, exaggerated expressions, smiling, laughing, exaggerated sadness, wrong gaze direction, eyes looking at camera, mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, off-sync audio, missing sniff sounds, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, missing door or shelves, missing shallow depth of field, flat lighting, inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"251:218": {
"class_type": "CreateVideo",
"inputs": {
"audio": [
"251:220",
0
],
"bit_depth": 8,
"codec": "none",
"color_space": "sRGB",
"fps": [
"251:212",
0
],
"images": [
"251:219",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"251:219": {
"class_type": "VAEDecodeTiled",
"inputs": {
"overlap": 64,
"samples": [
"251:200",
2
],
"temporal_overlap": 64,
"temporal_size": 4096,
"tile_size": 768,
"vae": [
"251:229",
0
]
},
"_meta": {
"title": "VAE Decode (Tiled)"
}
},
"251:220": {
"class_type": "LTXVAudioVAEDecode",
"inputs": {
"audio_vae": [
"251:227",
0
],
"samples": [
"251:221",
1
]
},
"_meta": {
"title": "LTXV Audio VAE Decode"
}
},
"251:221": {
"class_type": "LTXVSeparateAVLatent",
"inputs": {
"av_latent": [
"251:211",
1
]
},
"_meta": {
"title": "Separate AV Latent"
}
},
"251:222": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"251:228",
0
],
"text": [
"251:248",
0
]
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"251:226": {
"class_type": "ComfyMathExpression",
"inputs": {
"expression": "a * b + 1",
"values.a": [
"251:198",
0
],
"values.b": [
"251:205",
0
]
},
"_meta": {
"title": "Math Expression"
}
},
"251:227": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "ltx-2.5-audio-vae-bf16.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"251:228": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot.safetensors",
"device": "default",
"type": "ltxv"
},
"_meta": {
"title": "Load CLIP"
}
},
"251:229": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "ltx-2.5-video-vae-bf16.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"251:230": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "ltx-2.5-22b-distilled-transformer-comfy-int8-convrot.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"251:235": {
"class_type": "LTXVDualCFGGuider",
"inputs": {
"audio_cfg": 1,
"model": [
"251:230",
0
],
"negative": [
"251:204",
1
],
"positive": [
"251:204",
0
],
"video_cfg": 1
},
"_meta": {
"title": "LTXV Dual CFG Guider"
}
},
"251:239": {
"class_type": "ManualSigmas",
"inputs": {
"sigmas": "1.0, 0.99375, 0.9875, 0.98125, 0.975, 0.909375, 0.725, 0.421875, 0.0"
},
"_meta": {
"title": "ManualSigmas"
}
},
"251:246": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "gemma4_e2b_it_bf16.safetensors",
"device": "default",
"type": "ltxv"
},
"_meta": {
"title": "Load CLIP"
}
},
"251:247": {
"class_type": "TextGenerateLTX2Prompt",
"inputs": {
"clip": [
"251:246",
0
],
"image": [
"251:199",
0
],
"max_length": 600,
"mtp": "auto",
"prompt": [
"251:252",
0
],
"sampling_mode": "on",
"sampling_mode.min_p": 0.05,
"sampling_mode.presence_penalty": 0,
"sampling_mode.repetition_penalty": 1.15,
"sampling_mode.seed": 0,
"sampling_mode.temperature": 0.7,
"sampling_mode.top_k": 64,
"sampling_mode.top_p": 0.95,
"thinking": False,
"use_default_template": True
},
"_meta": {
"title": "Generate LTX2 Prompt"
}
},
"251:248": {
"class_type": "ComfySwitchNode",
"inputs": {
"on_false": [
"251:252",
0
],
"on_true": [
"251:247",
0
],
"switch": [
"251:250",
0
]
},
"_meta": {
"title": "If/Else Switch"
}
},
"251:249": {
"class_type": "PreviewAny",
"inputs": {
"source": [
"251:248",
0
]
},
"_meta": {
"title": "Preview as Text"
}
},
"251:250": {
"class_type": "PrimitiveBoolean",
"inputs": {
"value": True
},
"_meta": {
"title": "Boolean (Enable Prompt Enhance)"
}
},
"251:252": {
"class_type": "PrimitiveStringMultiline",
"inputs": {
"value": "The video opens low on the black water, the colossal closed bud standing motionless under the stormy midnight sky while thin fog crawls across the surface and a cold wind hisses. A single beam of moonlight shifts across the frosted outer petals. With a deep creaking sound, the bud begins to loosen: the outer petals peel back one by one, slowly and heavily, shedding flakes of frost into the water with soft splashes. Ripples spread outward in rings, catching faint cyan glints from below. As the inner petals unfurl, a warm amber glow swells from the heart of the flower, rising into the mist and lighting the undersides of the petals and the nearby basalt rocks. The camera holds steady at water level, then drifts slightly closer as the flower opens fully and its glow hums softly."
},
"_meta": {
"title": "Prompt"
}
}
})
workflow.set_input("31", "image", client.assets.from_url("https://comfy.org/catalog-media/4f6040be-625a-4b93-a379-fe59a630c054/a4ded0e5-806b-42b3-ae29-8ebd7edc22f1"))
workflow.set_input("39", "image", client.assets.from_url("https://comfy.org/catalog-media/d12742d8-ff7e-4fd7-a6b6-dd80ef65305c/696609a3-c17b-4b2e-af69-29bb504a5295"))
# run() also passes api_key to the partner nodes in this workflow.
job = client.run(workflow, api_key=api_key)
for output in job.get_outputs("68"):
output.to_file(output.name)试试一个示例
关于此工作流
LTX-2.5: FLF2V is a focused ComfyUI workflow for turning two still images—your intended first and last frames—into a single, coherent video. It’s ideal for creating smooth transitions, morphs between subjects, or bridging shots in a sequence when you only have still references. Two LoadImage nodes supply the start and end visuals, which feed into a single LTX-2.5 FLF2V generator node (cf70afc4-5a03-47ce-8210-734b1de6c6bc). The resulting frames are handed to SaveVideo for encoding to a playable clip. A MarkdownNote is included with inline tips.
Under the hood, the LTX-2.5 node leverages Lightricks’ open video generation model and its keyframes-first “Pixel Diffusion” approach to synthesize in-between frames from your bookend images. Typical controls you’ll use are target resolution, number of frames (or duration), frames-per-second, and a random seed for reproducibility. Many builds also expose blending/easing options to bias the motion toward the first or last frame. SaveVideo then encodes the output at your chosen fps (MP4/WebM, depending on your ffmpeg setup).
For best results, match the aspect ratio and general lighting/composition of your two reference images. Larger frame counts and consistent content between the first and last frames usually produce smoother motion. If you see stretching or cropping, pre-crop your inputs to the same size before loading them into the workflow.
常见问答
Matching aspect ratios is strongly recommended. The generator will resize inputs to its working resolution, but mismatched aspects can introduce stretching or unwanted cropping. Pre-crop or pad both images to the same dimensions before loading for the cleanest results.
The perceived duration is frames ÷ fps. To slow the transition and make it smoother, increase the number of frames while keeping fps the same. Some builds also include a blend/easing control—use it to ease-in or ease-out of the start/end frame for a more natural feel.
Yes. Set a fixed seed in the LTX-2.5 FLF2V node (if exposed). Using the same inputs and parameters with the same seed will reproduce the same video frames, which is useful for iterative tweaks.
Reduce the target resolution and/or the number of frames, and ensure both images are pre-sized to the model’s working aspect to avoid extra scaling. A modern GPU is recommended for smooth local generation; if you still see out-of-memory errors, lower the resolution further and iterate.