{ "en": "LTX-2 field guide", "zh": "LTX-2 field guide", "ru": "LTX-2 field guide", "de": "LTX-2 field guide", "fr": "LTX-2 field guide", "ja": "LTX-2 field guide", "ko": "LTX-2 field guide", "pt": "LTX-2 field guide", "tr": "LTX-2 field guide", "es": "LTX-2 field guide" }
{ "en": "LTX-2 ComfyUI Guide: Nodes, Models and a Reliable Workflow", "zh": "LTX-2 ComfyUI 教程：节点、模型与可靠工作流", "ru": "LTX-2 ComfyUI Guide: Nodes, Models and a Reliable Workflow", "de": "LTX-2 ComfyUI Guide: Nodes, Models and a Reliable Workflow", "fr": "LTX-2 ComfyUI Guide: Nodes, Models and a Reliable Workflow", "ja": "LTX-2 ComfyUI ガイド：ノード、モデル、安定したワークフロー", "ko": "LTX-2 ComfyUI Guide: Nodes, Models and a Reliable Workflow", "pt": "LTX-2 ComfyUI Guide: Nodes, Models and a Reliable Workflow", "tr": "LTX-2 ComfyUI Guide: Nodes, Models and a Reliable Workflow", "es": "Guía de LTX-2 en ComfyUI: nodos, modelos y flujo fiable" }
{ "en": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "zh": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "ru": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "de": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "fr": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "ja": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "ko": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "pt": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "tr": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files.", "es": "Use ComfyUI when you want a visible, repeatable graph for LTX-2 text-to-video or image-to-video. The exact nodes and checkpoints depend on the provider package you install, so verify names against the package documentation before downloading files." }
1. Check the workflow before installing
Start from the workflow JSON supplied by the model or node package. Confirm the checkpoint filename, sampler, VAE and output node; do not assume a workflow for LTX-2.5 also works for LTX-2.3.
- Use a clean ComfyUI environment or a dedicated portable install.
- Keep model files in the package-defined folders.
- Record the package version and checkpoint name with every test run.
2. Use conservative first-pass settings
For a first render, use a short clip and 720p output. Move to 1080p only after the graph completes and the motion is stable. The hosted LTX-2.3 route supports 480p, 720p and 1080p for 5–20 second clips.
- Text-to-video: describe subject, action, camera and lighting.
- Image-to-video: keep the reference image dimensions consistent with the target aspect ratio.
- Lower resolution or frame count when a graph runs out of VRAM.
3. Diagnose failures by layer
A red node is usually a missing model file, an incompatible custom node, or insufficient memory. Read the first error in the console, then validate the checkpoint path before changing sampler settings.
Reproducible sample recipe
A locked-off medium shot of a red paper kite lifting from a beach, steady wind, warm late-afternoon light, natural camera motion.
LTX-2.3 · text-to-video · 720p · 10 seconds · 16:9 · seed recorded in workflow JSON
This is a recipe for a verifiable test. It is not a claim that every ComfyUI package exposes identical defaults.
온라인에서 워크플로 테스트
로컬 환경을 디버깅하기 전에 프롬프트를 확인하세요.