Apply suggestion from @yiyixuxu

docs/source/en/api/pipelines/ltx2.md

@@ -106,6 +106,8 @@ video, audio = pipe(
     output_type="np",
     return_dict=False,
 )
+video = (video * 255).round().astype("uint8")
+video = torch.from_numpy(video)
 
 encode_video(
     video[0],
@@ -183,6 +185,8 @@ video, audio = pipe(
     output_type="np",
     return_dict=False,
 )
+video = (video * 255).round().astype("uint8")
+video = torch.from_numpy(video)
 
 encode_video(
     video[0],

docs/source/en/modular_diffusers/components_manager.md

@@ -12,179 +12,85 @@ specific language governing permissions and limitations under the License.
 
 # ComponentsManager
 
-The [`ComponentsManager`] is a model registry and management system for Modular Diffusers. It adds and tracks models, stores useful metadata (model size, device placement, adapters), prevents duplicate model instances, and supports offloading.
+The [`ComponentsManager`] is a model registry and management system for Modular Diffusers. It adds and tracks models, stores useful metadata (model size, device placement, adapters), and supports offloading.
 
 This guide will show you how to use [`ComponentsManager`] to manage components and device memory.
 
-## Add a component
+## Connect to a pipeline
 
-The [`ComponentsManager`] should be created alongside a [`ModularPipeline`] in either [`~ModularPipeline.from_pretrained`] or [`~ModularPipelineBlocks.init_pipeline`].
+Create a [`ComponentsManager`] and pass it to a [`ModularPipeline`] with either [`~ModularPipeline.from_pretrained`] or [`~ModularPipelineBlocks.init_pipeline`]. 
 
-> [!TIP]
-> The `collection` parameter is optional but makes it easier to organize and manage components.
 
 <hfoptions id="create">
 <hfoption id="from_pretrained">
 
 ```py
 from diffusers import ModularPipeline, ComponentsManager
+import torch
 
-comp = ComponentsManager()
-pipe = ModularPipeline.from_pretrained("YiYiXu/modular-demo-auto", components_manager=comp, collection="test1")
+manager = ComponentsManager()
+pipe = ModularPipeline.from_pretrained("Tongyi-MAI/Z-Image-Turbo", components_manager=manager)
+pipe.load_components(torch_dtype=torch.bfloat16)
 ```
 
 </hfoption>
 <hfoption id="init_pipeline">
 
 ```py
-from diffusers import ComponentsManager
-from diffusers.modular_pipelines import SequentialPipelineBlocks
-from diffusers.modular_pipelines.stable_diffusion_xl import TEXT2IMAGE_BLOCKS
-
-t2i_blocks = SequentialPipelineBlocks.from_blocks_dict(TEXT2IMAGE_BLOCKS)
-
-modular_repo_id = "YiYiXu/modular-loader-t2i-0704"
-components = ComponentsManager()
-t2i_pipeline = t2i_blocks.init_pipeline(modular_repo_id, components_manager=components)
+from diffusers import ModularPipelineBlocks, ComponentsManager
+import torch
+manager = ComponentsManager()
+blocks = ModularPipelineBlocks.from_pretrained("diffusers/Florence2-image-Annotator", trust_remote_code=True)
+pipe= blocks.init_pipeline(components_manager=manager)
+pipe.load_components(torch_dtype=torch.bfloat16)
 ```
 
 </hfoption>
 </hfoptions>
 
-Components are only loaded and registered when using [`~ModularPipeline.load_components`] or [`~ModularPipeline.load_components`]. The example below uses [`~ModularPipeline.load_components`] to create a second pipeline that reuses all the components from the first one, and assigns it to a different collection
+Components loaded by the pipeline are automatically registered in the manager. You can inspect them right away.
+
+## Inspect components
+
+Print the [`ComponentsManager`] to see all registered components, including their class, device placement, dtype, memory size, and load ID.
+
+The output below corresponds to the `from_pretrained` example above
 
 ```py
-pipe.load_components()
-pipe2 = ModularPipeline.from_pretrained("YiYiXu/modular-demo-auto", components_manager=comp, collection="test2")
+Components:
+=============================================================================================================================
+Models:
+-----------------------------------------------------------------------------------------------------------------------------
+Name_ID                      | Class                    | Device: act(exec) | Dtype          | Size (GB) | Load ID
+-----------------------------------------------------------------------------------------------------------------------------
+text_encoder_140458257514752 | Qwen3Model               | cpu               | torch.bfloat16 | 7.49      | Tongyi-MAI/Z-Image-Turbo|text_encoder|null|null
+vae_140458257515376          | AutoencoderKL            | cpu               | torch.bfloat16 | 0.16      | Tongyi-MAI/Z-Image-Turbo|vae|null|null
+transformer_140458257515616  | ZImageTransformer2DModel | cpu               | torch.bfloat16 | 11.46     | Tongyi-MAI/Z-Image-Turbo|transformer|null|null
+-----------------------------------------------------------------------------------------------------------------------------
+
+Other Components:
+-----------------------------------------------------------------------------------------------------------------------------
+ID                           | Class                           | Collection
+-----------------------------------------------------------------------------------------------------------------------------
+scheduler_140461023555264    | FlowMatchEulerDiscreteScheduler | N/A
+tokenizer_140458256346432    | Qwen2Tokenizer                  | N/A
+-----------------------------------------------------------------------------------------------------------------------------
 ```
 
-Use the [`~ModularPipeline.null_component_names`] property to identify any components that need to be loaded, retrieve them with [`~ComponentsManager.get_components_by_names`], and then call [`~ModularPipeline.update_components`] to add the missing components.
-
-```py
-pipe2.null_component_names 
-['text_encoder', 'text_encoder_2', 'tokenizer', 'tokenizer_2', 'image_encoder', 'unet', 'vae', 'scheduler', 'controlnet']
-
-comp_dict = comp.get_components_by_names(names=pipe2.null_component_names)
-pipe2.update_components(**comp_dict)
-```
-
-To add individual components, use the [`~ComponentsManager.add`] method. This registers a component with a unique id.
-
-```py
-from diffusers import AutoModel
-
-text_encoder = AutoModel.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", subfolder="text_encoder")
-component_id = comp.add("text_encoder", text_encoder)
-comp
-```
-
-Use [`~ComponentsManager.remove`] to remove a component using their id.
-
-```py
-comp.remove("text_encoder_139917733042864")
-```
-
-## Retrieve a component
-
-The [`ComponentsManager`] provides several methods to retrieve registered components.
-
-### get_one
-
-The [`~ComponentsManager.get_one`] method returns a single component and supports pattern matching for the `name` parameter. If multiple components match, [`~ComponentsManager.get_one`] returns an error.
-
-| Pattern     | Example                          | Description                               |
-|-------------|----------------------------------|-------------------------------------------|
-| exact       | `comp.get_one(name="unet")`      | exact name match                          |
-| wildcard    | `comp.get_one(name="unet*")`     | names starting with "unet"                |
-| exclusion   | `comp.get_one(name="!unet")`     | exclude components named "unet"           |
-| or          | `comp.get_one(name="unet&#124;vae")`  | name is "unet" or "vae"                   |
-
-[`~ComponentsManager.get_one`] also filters components by the `collection` argument or `load_id` argument.
-
-```py
-comp.get_one(name="unet", collection="sdxl")
-```
-
-### get_components_by_names
-
-The [`~ComponentsManager.get_components_by_names`] method accepts a list of names and returns a dictionary mapping names to components. This is especially useful with [`ModularPipeline`] since they provide lists of required component names and the returned dictionary can be passed directly to [`~ModularPipeline.update_components`].
-
-```py
-component_dict = comp.get_components_by_names(names=["text_encoder", "unet", "vae"])
-{"text_encoder": component1, "unet": component2, "vae": component3}
-```
-
-## Duplicate detection
-
-It is recommended to load model components with [`ComponentSpec`] to assign components with a unique id that encodes their loading parameters. This allows [`ComponentsManager`] to automatically detect and prevent duplicate model instances even when different objects represent the same underlying checkpoint.
-
-```py
-from diffusers import ComponentSpec, ComponentsManager
-from transformers import CLIPTextModel
-
-comp = ComponentsManager()
-
-# Create ComponentSpec for the first text encoder
-spec = ComponentSpec(name="text_encoder", repo="stabilityai/stable-diffusion-xl-base-1.0", subfolder="text_encoder", type_hint=AutoModel)
-# Create ComponentSpec for a duplicate text encoder (it is same checkpoint, from the same repo/subfolder)
-spec_duplicated = ComponentSpec(name="text_encoder_duplicated", repo="stabilityai/stable-diffusion-xl-base-1.0", subfolder="text_encoder", type_hint=CLIPTextModel)
-
-# Load and add both components - the manager will detect they're the same model
-comp.add("text_encoder", spec.load())
-comp.add("text_encoder_duplicated", spec_duplicated.load())
-```
-
-This returns a warning with instructions for removing the duplicate.
-
-```py
-ComponentsManager: adding component 'text_encoder_duplicated_139917580682672', but it has duplicate load_id 'stabilityai/stable-diffusion-xl-base-1.0|text_encoder|null|null' with existing components: text_encoder_139918506246832. To remove a duplicate, call `components_manager.remove('<component_id>')`.
-'text_encoder_duplicated_139917580682672'
-```
-
-You could also add a component without using [`ComponentSpec`] and duplicate detection still works in most cases even if you're adding the same component under a different name.
-
-However, [`ComponentManager`] can't detect duplicates when you load the same component into different objects. In this case, you should load a model with [`ComponentSpec`].
-
-```py
-text_encoder_2 = AutoModel.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", subfolder="text_encoder")
-comp.add("text_encoder", text_encoder_2)
-'text_encoder_139917732983664'
-```
-
-## Collections
-
-Collections are labels assigned to components for better organization and management. Add a component to a collection with the `collection` argument in [`~ComponentsManager.add`].
-
-Only one component per name is allowed in each collection. Adding a second component with the same name automatically removes the first component.
-
-```py
-from diffusers import ComponentSpec, ComponentsManager
-
-comp = ComponentsManager()
-# Create ComponentSpec for the first UNet
-spec = ComponentSpec(name="unet", repo="stabilityai/stable-diffusion-xl-base-1.0", subfolder="unet", type_hint=AutoModel)
-# Create ComponentSpec for a different UNet
-spec2 = ComponentSpec(name="unet", repo="RunDiffusion/Juggernaut-XL-v9", subfolder="unet", type_hint=AutoModel, variant="fp16")
-
-# Add both UNets to the same collection - the second one will replace the first
-comp.add("unet", spec.load(), collection="sdxl")
-comp.add("unet", spec2.load(), collection="sdxl")
-```
-
-This makes it convenient to work with node-based systems because you can:
-
-- Mark all models as loaded from one node with the `collection` label.
-- Automatically replace models when new checkpoints are loaded under the same name.
-- Batch delete all models in a collection when a node is removed.
+The table shows models (with device, dtype, and memory info) separately from other components like schedulers and tokenizers. If any models have LoRA adapters, IP-Adapters, or quantization applied, that information is displayed in an additional section at the bottom.
 
 ## Offloading
 
 The [`~ComponentsManager.enable_auto_cpu_offload`] method is a global offloading strategy that works across all models regardless of which pipeline is using them. Once enabled, you don't need to worry about device placement if you add or remove components.
 
 ```py
-comp.enable_auto_cpu_offload(device="cuda")
+manager.enable_auto_cpu_offload(device="cuda")
 ```
 
 All models begin on the CPU and [`ComponentsManager`] moves them to the appropriate device right before they're needed, and moves other models back to the CPU when GPU memory is low.
 
-You can set your own rules for which models to offload first.
+To disable offloading, call [~ComponentsManager.disable_auto_cpu_offload].
+
+```py
+manager.disable_auto_cpu_offload()
+```

src/diffusers/pipelines/ltx2/export_utils.py

@@ -13,15 +13,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from collections.abc import Iterator
 from fractions import Fraction
-from itertools import chain
-from typing import List, Optional, Union
+from typing import Optional
 
-import numpy as np
-import PIL.Image
 import torch
-from tqdm import tqdm
 
 from ...utils import is_av_available
 
@@ -106,54 +101,11 @@ def _write_audio(
 
 
 def encode_video(
-    video: Union[List[PIL.Image.Image], np.ndarray, torch.Tensor, Iterator[torch.Tensor]],
-    fps: int,
-    audio: Optional[torch.Tensor],
-    audio_sample_rate: Optional[int],
-    output_path: str,
-    video_chunks_number: int = 1,
+    video: torch.Tensor, fps: int, audio: Optional[torch.Tensor], audio_sample_rate: Optional[int], output_path: str
 ) -> None:
-    """
-    Encodes a video with audio using the PyAV library. Based on code from the original LTX-2 repo:
-    https://github.com/Lightricks/LTX-2/blob/4f410820b198e05074a1e92de793e3b59e9ab5a0/packages/ltx-pipelines/src/ltx_pipelines/utils/media_io.py#L182
+    video_np = video.cpu().numpy()
 
-    Args:
-        video (`List[PIL.Image.Image]` or `np.ndarray` or `torch.Tensor`):
-            A video tensor of shape [frames, height, width, channels] with integer pixel values in [0, 255]. If the
-            input is a `np.ndarray`, it is expected to be a float array with values in [0, 1] (which is what pipelines
-            usually return with `output_type="np"`).
-        fps (`int`)
-            The frames per second (FPS) of the encoded video.
-        audio (`torch.Tensor`, *optional*):
-            An audio waveform of shape [audio_channels, samples].
-        audio_sample_rate: (`int`, *optional*):
-            The sampling rate of the audio waveform. For LTX 2, this is typically 24000 (24 kHz).
-        output_path (`str`):
-            The path to save the encoded video to.
-        video_chunks_number (`int`, *optional*, defaults to `1`):
-            The number of chunks to split the video into for encoding. Each chunk will be encoded separately. The
-            number of chunks to use often depends on the tiling config for the video VAE.
-    """
-    if isinstance(video, list) and isinstance(video[0], PIL.Image.Image):
-        # Pipeline output_type="pil"; assumes each image is in "RGB" mode
-        video_frames = [np.array(frame) for frame in video]
-        video = np.stack(video_frames, axis=0)
-        video = torch.from_numpy(video)
-    elif isinstance(video, np.ndarray):
-        # Pipeline output_type="np"
-        is_denormalized = np.logical_and(np.zeros_like(video) <= video, video <= np.ones_like(video))
-        if np.all(is_denormalized):
-            video = (video * 255).round().astype("uint8")
-        video = torch.from_numpy(video)
-
-    if isinstance(video, torch.Tensor):
-        # Split into video_chunks_number along the frame dimension
-        video = torch.tensor_split(video, video_chunks_number, dim=0)
-        video = iter(video)
-
-    first_chunk = next(video)
-
-    _, height, width, _ = first_chunk.shape
+    _, height, width, _ = video_np.shape
 
     container = av.open(output_path, mode="w")
     stream = container.add_stream("libx264", rate=int(fps))
@@ -167,12 +119,10 @@ def encode_video(
 
         audio_stream = _prepare_audio_stream(container, audio_sample_rate)
 
-    for video_chunk in tqdm(chain([first_chunk], video), total=video_chunks_number, desc="Encoding video chunks"):
-        video_chunk_cpu = video_chunk.to("cpu").numpy()
-        for frame_array in video_chunk_cpu:
-            frame = av.VideoFrame.from_ndarray(frame_array, format="rgb24")
-            for packet in stream.encode(frame):
-                container.mux(packet)
+    for frame_array in video_np:
+        frame = av.VideoFrame.from_ndarray(frame_array, format="rgb24")
+        for packet in stream.encode(frame):
+            container.mux(packet)
 
     # Flush encoder
     for packet in stream.encode():

src/diffusers/pipelines/ltx2/pipeline_ltx2.py

@@ -69,6 +69,8 @@ EXAMPLE_DOC_STRING = """
         ...     output_type="np",
         ...     return_dict=False,
         ... )
+        >>> video = (video * 255).round().astype("uint8")
+        >>> video = torch.from_numpy(video)
 
         >>> encode_video(
         ...     video[0],

src/diffusers/pipelines/ltx2/pipeline_ltx2_image2video.py

@@ -75,6 +75,8 @@ EXAMPLE_DOC_STRING = """
         ...     output_type="np",
         ...     return_dict=False,
         ... )
+        >>> video = (video * 255).round().astype("uint8")
+        >>> video = torch.from_numpy(video)
 
         >>> encode_video(
         ...     video[0],

src/diffusers/pipelines/ltx2/pipeline_ltx2_latent_upsample.py

@@ -76,6 +76,8 @@ EXAMPLE_DOC_STRING = """
         ...     output_type="np",
         ...     return_dict=False,
         ... )[0]
+        >>> video = (video * 255).round().astype("uint8")
+        >>> video = torch.from_numpy(video)
 
         >>> encode_video(
         ...     video[0],