sync: add 3 AtlasCloud models (2026-07-13)

README.md

@@ -457,6 +457,9 @@ This node pack focuses on **image / video / edit** — see the full **[node cata
 | AtlasCloud Grok Imagine Edit | xai/grok-imagine-image/edit |
 | AtlasCloud GPT Image-2 Edit | openai/gpt-image-2/edit |
 | AtlasCloud GPT Image-2 Developer Edit | openai/gpt-image-2-developer/edit |
+| AtlasCloud LTX 2.3 Quality Text-to-Video | ltx-2.3-quality/text-to-video |
+| AtlasCloud LTX 2.3 Quality Image-to-Video | ltx-2.3-quality/image-to-video |
+| AtlasCloud LTX 2.3 Quality Extend Video | ltx-2.3-quality/extend-video |
 
 > Nodes are continuously expanded as new models are added to AtlasCloud.
 

src/atlascloud_comfyui/nodes/video/ltx_2_3_quality_extend_video.py

@@ -0,0 +1,116 @@
+from __future__ import annotations
+
+from typing import Any, Dict, Tuple
+
+from ..auth.atlas_client_node import AtlasClientHandle
+
+_RESOLUTIONS = [
+    "square_hd",
+    "square",
+    "portrait_3_4",
+    "portrait_9_16",
+    "landscape_4_3",
+    "landscape_16_9",
+]
+
+_DEFAULT_PROMPT = "Continue the scene naturally, maintaining the same style, lighting, and motion dynamics"
+
+
+class AtlasLtx23QualityExtendVideo:
+    """
+    ComfyUI Node: AtlasCloud LTX 2.3 Quality Extend Video (remote API)
+    Outputs: video_url, prediction_id
+    """
+
+    CATEGORY = "AtlasCloud/Video"
+    FUNCTION = "run"
+    RETURN_TYPES = ("STRING", "STRING")
+    RETURN_NAMES = ("video_url", "prediction_id")
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "atlas_client": ("ATLAS_CLIENT",),
+                "video_url": ("STRING", {"default": "", "tooltip": "Input video URL or data:video/...;base64,..."}),
+                "prompt": ("STRING", {"multiline": True, "default": _DEFAULT_PROMPT, "tooltip": "Extended scene description (max 2000 chars)"}),
+            },
+            "optional": {
+                "extend_direction": (
+                    ["forward", "backward"],
+                    {"default": "forward", "tooltip": "Forward (append after) or Backward (prepend before)"},
+                ),
+                "num_frames": (
+                    "INT",
+                    {"default": 81, "min": 9, "max": 481, "tooltip": "Extension frames; multiple of 8 plus 1"},
+                ),
+                "resolution": (_RESOLUTIONS, {"default": "landscape_16_9", "tooltip": "Output resolution preset"}),
+                "match_input_fps": ("BOOLEAN", {"default": True, "tooltip": "Match the input video FPS"}),
+                "generate_audio": ("BOOLEAN", {"default": False, "tooltip": "Generate accompanying audio"}),
+                "randomize_seed": ("BOOLEAN", {"default": True, "tooltip": "开启后每次随机；关闭后使用下方固定 seed"}),
+                "seed": ("INT", {"default": 0, "min": 0, "max": 2**31 - 1, "tooltip": "固定 seed（仅在随机开关关闭时生效）"}),
+                "poll_interval_sec": ("FLOAT", {"default": 2.0, "min": 0.5, "max": 10.0, "tooltip": "Polling interval (seconds)"}),
+                "timeout_sec": ("INT", {"default": 900, "min": 30, "max": 7200, "tooltip": "Timeout (seconds)"}),
+            },
+        }
+
+    def run(
+        self,
+        atlas_client: AtlasClientHandle,
+        video_url: str,
+        prompt: str,
+        extend_direction: str = "forward",
+        num_frames: int = 81,
+        resolution: str = "landscape_16_9",
+        match_input_fps: bool = True,
+        generate_audio: bool = False,
+        randomize_seed: bool = True,
+        seed: int = 0,
+        poll_interval_sec: float = 2.0,
+        timeout_sec: int = 900,
+    ) -> Tuple[str, str]:
+        video_url = (video_url or "").strip()
+        if not video_url:
+            raise RuntimeError("video_url is required for LTX 2.3 Quality Extend Video")
+
+        p = (prompt or "").strip()
+        if not p:
+            raise RuntimeError("prompt is required for LTX 2.3 Quality Extend Video")
+
+        client = atlas_client.client
+
+        payload: Dict[str, Any] = {
+            "model": "ltx-2.3-quality/extend-video",
+            "prompt": p,
+            "video_url": video_url,
+            "extend_direction": extend_direction,
+            "num_frames": int(num_frames),
+            "resolution": resolution,
+            "match_input_fps": bool(match_input_fps),
+            "generate_audio": bool(generate_audio),
+        }
+        if not randomize_seed:
+            payload["seed"] = int(seed)
+
+        prediction_id = client.generate_video(payload)
+        result = client.poll_prediction(
+            prediction_id,
+            poll_interval_sec=float(poll_interval_sec),
+            timeout_sec=float(timeout_sec),
+        )
+
+        outputs = (result.get("data") or {}).get("outputs") or []
+        if not outputs:
+            raise RuntimeError(f"No outputs returned for prediction {prediction_id}: {result}")
+
+        first = outputs[0]
+        if isinstance(first, dict):
+            url = first.get("url") or first.get("video") or first.get("output")
+            if isinstance(url, str) and url.strip():
+                return (url, prediction_id)
+            raise RuntimeError(f"Unexpected output object for prediction {prediction_id}: {first}")
+
+        if not isinstance(first, str):
+            raise RuntimeError(f"Unexpected output type for prediction {prediction_id}: {type(first).__name__} {first!r}")
+
+        return (first, prediction_id)

src/atlascloud_comfyui/nodes/video/ltx_2_3_quality_i2v.py

@@ -0,0 +1,105 @@
+from __future__ import annotations
+
+from typing import Any, Dict, Tuple
+
+from ..auth.atlas_client_node import AtlasClientHandle
+
+_RESOLUTIONS = [
+    "square_hd",
+    "square",
+    "portrait_3_4",
+    "portrait_9_16",
+    "landscape_4_3",
+    "landscape_16_9",
+]
+
+
+class AtlasLtx23QualityImageToVideo:
+    """
+    ComfyUI Node: AtlasCloud LTX 2.3 Quality Image-to-Video (remote API)
+    Outputs: video_url, prediction_id
+    """
+
+    CATEGORY = "AtlasCloud/Video"
+    FUNCTION = "run"
+    RETURN_TYPES = ("STRING", "STRING")
+    RETURN_NAMES = ("video_url", "prediction_id")
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "atlas_client": ("ATLAS_CLIENT",),
+                "prompt": ("STRING", {"multiline": True, "tooltip": "Motion/style prompt (max 2000 chars)"}),
+                "image_url": ("STRING", {"default": "", "tooltip": "Input image URL or data:image/...;base64,..."}),
+            },
+            "optional": {
+                "num_frames": (
+                    "INT",
+                    {"default": 121, "min": 25, "max": 257, "tooltip": "Frames; multiple of 8 plus 1"},
+                ),
+                "resolution": (_RESOLUTIONS, {"default": "landscape_16_9", "tooltip": "Output resolution preset"}),
+                "generate_audio": ("BOOLEAN", {"default": False, "tooltip": "Generate accompanying audio"}),
+                "randomize_seed": ("BOOLEAN", {"default": True, "tooltip": "开启后每次随机；关闭后使用下方固定 seed"}),
+                "seed": ("INT", {"default": 0, "min": 0, "max": 2**31 - 1, "tooltip": "固定 seed（仅在随机开关关闭时生效）"}),
+                "poll_interval_sec": ("FLOAT", {"default": 2.0, "min": 0.5, "max": 10.0, "tooltip": "Polling interval (seconds)"}),
+                "timeout_sec": ("INT", {"default": 900, "min": 30, "max": 7200, "tooltip": "Timeout (seconds)"}),
+            },
+        }
+
+    def run(
+        self,
+        atlas_client: AtlasClientHandle,
+        prompt: str,
+        image_url: str,
+        num_frames: int = 121,
+        resolution: str = "landscape_16_9",
+        generate_audio: bool = False,
+        randomize_seed: bool = True,
+        seed: int = 0,
+        poll_interval_sec: float = 2.0,
+        timeout_sec: int = 900,
+    ) -> Tuple[str, str]:
+        image_url = (image_url or "").strip()
+        if not image_url:
+            raise RuntimeError("image_url is required for LTX 2.3 Quality Image-to-Video")
+
+        p = (prompt or "").strip()
+        if not p:
+            raise RuntimeError("prompt is required for LTX 2.3 Quality Image-to-Video")
+
+        client = atlas_client.client
+
+        payload: Dict[str, Any] = {
+            "model": "ltx-2.3-quality/image-to-video",
+            "prompt": p,
+            "image_url": image_url,
+            "num_frames": int(num_frames),
+            "resolution": resolution,
+            "generate_audio": bool(generate_audio),
+        }
+        if not randomize_seed:
+            payload["seed"] = int(seed)
+
+        prediction_id = client.generate_video(payload)
+        result = client.poll_prediction(
+            prediction_id,
+            poll_interval_sec=float(poll_interval_sec),
+            timeout_sec=float(timeout_sec),
+        )
+
+        outputs = (result.get("data") or {}).get("outputs") or []
+        if not outputs:
+            raise RuntimeError(f"No outputs returned for prediction {prediction_id}: {result}")
+
+        first = outputs[0]
+        if isinstance(first, dict):
+            url = first.get("url") or first.get("video") or first.get("output")
+            if isinstance(url, str) and url.strip():
+                return (url, prediction_id)
+            raise RuntimeError(f"Unexpected output object for prediction {prediction_id}: {first}")
+
+        if not isinstance(first, str):
+            raise RuntimeError(f"Unexpected output type for prediction {prediction_id}: {type(first).__name__} {first!r}")
+
+        return (first, prediction_id)

src/atlascloud_comfyui/nodes/video/ltx_2_3_quality_t2v.py

@@ -0,0 +1,98 @@
+from __future__ import annotations
+
+from typing import Any, Dict, Tuple
+
+from ..auth.atlas_client_node import AtlasClientHandle
+
+_RESOLUTIONS = [
+    "square_hd",
+    "square",
+    "portrait_3_4",
+    "portrait_9_16",
+    "landscape_4_3",
+    "landscape_16_9",
+]
+
+
+class AtlasLtx23QualityTextToVideo:
+    """
+    ComfyUI Node: AtlasCloud LTX 2.3 Quality Text-to-Video (remote API)
+    Outputs: video_url, prediction_id
+    """
+
+    CATEGORY = "AtlasCloud/Video"
+    FUNCTION = "run"
+    RETURN_TYPES = ("STRING", "STRING")
+    RETURN_NAMES = ("video_url", "prediction_id")
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "atlas_client": ("ATLAS_CLIENT",),
+                "prompt": ("STRING", {"multiline": True, "tooltip": "Text prompt (max 2000 chars)"}),
+            },
+            "optional": {
+                "num_frames": (
+                    "INT",
+                    {"default": 121, "min": 25, "max": 257, "tooltip": "Frames; multiple of 8 plus 1"},
+                ),
+                "resolution": (_RESOLUTIONS, {"default": "landscape_16_9", "tooltip": "Output resolution preset"}),
+                "generate_audio": ("BOOLEAN", {"default": False, "tooltip": "Generate accompanying audio"}),
+                "randomize_seed": ("BOOLEAN", {"default": True, "tooltip": "开启后每次随机；关闭后使用下方固定 seed"}),
+                "seed": ("INT", {"default": 0, "min": 0, "max": 2**31 - 1, "tooltip": "固定 seed（仅在随机开关关闭时生效）"}),
+                "poll_interval_sec": ("FLOAT", {"default": 2.0, "min": 0.5, "max": 10.0, "tooltip": "Polling interval (seconds)"}),
+                "timeout_sec": ("INT", {"default": 900, "min": 30, "max": 7200, "tooltip": "Timeout (seconds)"}),
+            },
+        }
+
+    def run(
+        self,
+        atlas_client: AtlasClientHandle,
+        prompt: str,
+        num_frames: int = 121,
+        resolution: str = "landscape_16_9",
+        generate_audio: bool = False,
+        randomize_seed: bool = True,
+        seed: int = 0,
+        poll_interval_sec: float = 2.0,
+        timeout_sec: int = 900,
+    ) -> Tuple[str, str]:
+        p = (prompt or "").strip()
+        if not p:
+            raise RuntimeError("prompt is required for LTX 2.3 Quality Text-to-Video")
+
+        client = atlas_client.client
+
+        payload: Dict[str, Any] = {
+            "model": "ltx-2.3-quality/text-to-video",
+            "prompt": p,
+            "num_frames": int(num_frames),
+            "resolution": resolution,
+            "generate_audio": bool(generate_audio),
+        }
+        if not randomize_seed:
+            payload["seed"] = int(seed)
+
+        prediction_id = client.generate_video(payload)
+        result = client.poll_prediction(
+            prediction_id,
+            poll_interval_sec=float(poll_interval_sec),
+            timeout_sec=float(timeout_sec),
+        )
+
+        outputs = (result.get("data") or {}).get("outputs") or []
+        if not outputs:
+            raise RuntimeError(f"No outputs returned for prediction {prediction_id}: {result}")
+
+        first = outputs[0]
+        if isinstance(first, dict):
+            url = first.get("url") or first.get("video") or first.get("output")
+            if isinstance(url, str) and url.strip():
+                return (url, prediction_id)
+            raise RuntimeError(f"Unexpected output object for prediction {prediction_id}: {first}")
+
+        if not isinstance(first, str):
+            raise RuntimeError(f"Unexpected output type for prediction {prediction_id}: {type(first).__name__} {first!r}")
+
+        return (first, prediction_id)

src/atlascloud_comfyui/registry.py

@@ -265,6 +265,10 @@
 from atlascloud_comfyui.nodes.image.ideogram_v4_turbo_t2i import AtlasIdeogramV4TurboTextToImage
 from atlascloud_comfyui.nodes.image.ideogram_v4_quality_t2i import AtlasIdeogramV4QualityTextToImage
 
+from atlascloud_comfyui.nodes.video.ltx_2_3_quality_t2v import AtlasLtx23QualityTextToVideo
+from atlascloud_comfyui.nodes.video.ltx_2_3_quality_i2v import AtlasLtx23QualityImageToVideo
+from atlascloud_comfyui.nodes.video.ltx_2_3_quality_extend_video import AtlasLtx23QualityExtendVideo
+
 from atlascloud_comfyui.nodes.video.kling_v16_multi_i2v_pro import AtlasKlingV16MultiI2VPro
 from atlascloud_comfyui.nodes.video.kling_v16_multi_i2v_standard import AtlasKlingV16MultiI2VStandard
 from atlascloud_comfyui.nodes.video.kling_v16_i2v_standard import AtlasKlingV16I2VStandard
@@ -723,6 +727,9 @@
     "AtlasCloud Cosmos 3 Super Image-to-Video": AtlasCosmos3SuperImageToVideo,
     "AtlasCloud Ideogram V4 Turbo Text-to-Image": AtlasIdeogramV4TurboTextToImage,
     "AtlasCloud Ideogram V4 Quality Text-to-Image": AtlasIdeogramV4QualityTextToImage,
+    "AtlasCloud LTX 2.3 Quality Text-to-Video": AtlasLtx23QualityTextToVideo,
+    "AtlasCloud LTX 2.3 Quality Image-to-Video": AtlasLtx23QualityImageToVideo,
+    "AtlasCloud LTX 2.3 Quality Extend Video": AtlasLtx23QualityExtendVideo,
 }
 
 
@@ -1063,6 +1070,9 @@
     "AtlasCloud Cosmos 3 Super Image-to-Video": "AtlasCloud Cosmos 3 Super Image-to-Video",
     "AtlasCloud Ideogram V4 Turbo Text-to-Image": "AtlasCloud Ideogram V4 Turbo Text-to-Image",
     "AtlasCloud Ideogram V4 Quality Text-to-Image": "AtlasCloud Ideogram V4 Quality Text-to-Image",
+    "AtlasCloud LTX 2.3 Quality Text-to-Video": "AtlasCloud LTX 2.3 Quality Text-to-Video",
+    "AtlasCloud LTX 2.3 Quality Image-to-Video": "AtlasCloud LTX 2.3 Quality Image-to-Video",
+    "AtlasCloud LTX 2.3 Quality Extend Video": "AtlasCloud LTX 2.3 Quality Extend Video",
 }