theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-18 22:08:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

add inference script for ltx-2 lora

Browse Source
This commit is contained in:
mi804
2026-02-10 14:32:30 +08:00
parent afd48cd706
commit b6e39c97af
18 changed files with 968 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
README.md
View File

@@ -32,6 +32,7 @@ We believe that a well-developed open-source code framework can lower the thresh
> DiffSynth-Studio has undergone major version updates, and some old features are no longer maintained. If you need to use old features, please switch to the [last historical version](https://github.com/modelscope/DiffSynth-Studio/tree/afd101f3452c9ecae0c87b79adfa2e22d65ffdc3) before the major version update.
> Currently, the development personnel of this project are limited, with most of the work handled by [Artiprocher](https://github.com/Artiprocher). Therefore, the progress of new feature development will be relatively slow, and the speed of responding to and resolving issues is limited. We apologize for this and ask developers to understand.
- **February 10, 2026** Added inference support for the LTX-2 audio-video generation model. See the documentation for details. Support for model training will be implemented in the future.
- **February 2, 2026** The first document of the Research Tutorial series is now available, guiding you through training a small 0.1B text-to-image model from scratch. For details, see the [documentation](/docs/en/Research_Tutorial/train_from_scratch.md) and [model](https://modelscope.cn/models/DiffSynth-Studio/AAAMyModel). We hope DiffSynth-Studio can evolve into a more powerful training framework for Diffusion models.
@@ -618,6 +619,13 @@ Example code for LTX-2 is available at: [/examples/ltx2/](/examples/ltx2/)
|[Lightricks/LTX-2: OneStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-OneStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-OneStage.py)|-|-|-|-|
|[Lightricks/LTX-2: TwoStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-TwoStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-TwoStage.py)|-|-|-|-|
|[Lightricks/LTX-2: DistilledPipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-DistilledPipeline.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-DistilledPipeline.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-In.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-In.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Up.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Up.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Down.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Down.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Static](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Static.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Static.py)|-|-|-|-|
</details>

8
README_zh.md
View File

@@ -32,6 +32,7 @@ DiffSynth 目前包括两个开源项目：
> DiffSynth-Studio 经历了大版本更新，部分旧功能已停止维护，如需使用旧版功能，请切换到大版本更新前的[最后一个历史版本](https://github.com/modelscope/DiffSynth-Studio/tree/afd101f3452c9ecae0c87b79adfa2e22d65ffdc3)。
> 目前本项目的开发人员有限，大部分工作由 [Artiprocher](https://github.com/Artiprocher) 负责，因此新功能的开发进展会比较缓慢，issue 的回复和解决速度有限，我们对此感到非常抱歉，请各位开发者理解。
- **2026年2月10日** 新增对[LTX-2](https://www.modelscope.cn/models/Lightricks/LTX-2)音视频生成模型的推理支持，详见[文档](docs/en/Model_Details/LTX-2.md)，后续将推进模型训练的支持。
- **2026年2月2日** Research Tutorial 的第一篇文档上线，带你从零开始训练一个 0.1B 的小型文生图模型，详见[文档](/docs/zh/Research_Tutorial/train_from_scratch.md)、[模型](https://modelscope.cn/models/DiffSynth-Studio/AAAMyModel)，我们希望 DiffSynth-Studio 能够成为一个更强大的 Diffusion 模型训练框架。
@@ -618,6 +619,13 @@ LTX-2 的示例代码位于：[/examples/ltx2/](/examples/ltx2/)
|[Lightricks/LTX-2: OneStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-OneStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-OneStage.py)|-|-|-|-|
|[Lightricks/LTX-2: TwoStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-TwoStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-TwoStage.py)|-|-|-|-|
|[Lightricks/LTX-2: DistilledPipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-DistilledPipeline.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-DistilledPipeline.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-In.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-In.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Up.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Up.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Down.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Down.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Static](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Static.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Static.py)|-|-|-|-|
</details>

7
docs/en/Model_Details/LTX-2.md
View File

@@ -73,6 +73,13 @@ write_video_audio_ltx2(
|[Lightricks/LTX-2: OneStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-OneStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-OneStage.py)|-|-|-|-|
|[Lightricks/LTX-2: TwoStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-TwoStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-TwoStage.py)|-|-|-|-|
|[Lightricks/LTX-2: DistilledPipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-DistilledPipeline.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-DistilledPipeline.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-In.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-In.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Up.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Up.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Down.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Down.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Static](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Static.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Static.py)|-|-|-|-|
## Model Inference

7
docs/zh/Model_Details/LTX-2.md
View File

@@ -73,6 +73,13 @@ write_video_audio_ltx2(
|[Lightricks/LTX-2: OneStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-OneStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-OneStage.py)|-|-|-|-|
|[Lightricks/LTX-2: TwoStagePipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-TwoStage.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-TwoStage.py)|-|-|-|-|
|[Lightricks/LTX-2: DistilledPipeline-I2AV](https://www.modelscope.cn/models/Lightricks/LTX-2)|`input_images`|[code](/examples/ltx2/model_inference/LTX-2-I2AV-DistilledPipeline.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-I2AV-DistilledPipeline.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-In.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-In.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Out.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Left.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Right.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Up.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Up.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Down.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Down.py)|-|-|-|-|
|[Lightricks/LTX-2-19b-LoRA-Camera-Control-Static](https://www.modelscope.cn/models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static)||[code](/examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Static.py)|[code](/examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Static.py)|-|-|-|-|
## 模型推理

65
examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-In.py Normal file
View File

@@ -0,0 +1,65 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.bfloat16,
"offload_device": "cpu",
"onload_dtype": torch.bfloat16,
"onload_device": "cuda",
"preparing_dtype": torch.bfloat16,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-in.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In/ltx-2-19b-lora-camera-control-dolly-in.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-in shot: A cheerful girl smiles brightly and says, 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' The camera smoothly moves closer to her face, highlighting her enthusiasm and sincerity."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_dolly_in_lora.mp4',
fps=24,
audio_sample_rate=24000,
)

65
examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Left.py Normal file
View File

@@ -0,0 +1,65 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.bfloat16,
"offload_device": "cpu",
"onload_dtype": torch.bfloat16,
"onload_device": "cuda",
"preparing_dtype": torch.bfloat16,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-left.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left/ltx-2-19b-lora-camera-control-dolly-left.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-left shot: A joyful young woman sits at a minimalist desk with a laptop running Diffsynth-Studio, code and generative visuals glowing on screen. She turns slightly toward the camera and says with a smile, 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' As she speaks, the camera smoothly dollies left, revealing a wall of framed open-source project posters, a whiteboard covered in neural network sketches, and a shelf stacked with AI/graphics books beside her."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_dolly_left_lora.mp4',
fps=24,
audio_sample_rate=24000,
)

66
examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Out.py Normal file
View File

@@ -0,0 +1,66 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.bfloat16,
"offload_device": "cpu",
"onload_dtype": torch.bfloat16,
"onload_device": "cuda",
"preparing_dtype": torch.bfloat16,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-out.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out/ltx-2-19b-lora-camera-control-dolly-out.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-out shot: A joyful young woman smiles warmly and says: 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' As she speaks, the camera slowly dollies out, revealing a bright, modern creative studio filled with plants, whiteboards full of diagrams, and soft natural light from large windows."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path=f'ltx2_dolly_out.mp4',
fps=24,
audio_sample_rate=24000,
)

65
examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Dolly-Right.py Normal file
View File

@@ -0,0 +1,65 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.bfloat16,
"offload_device": "cpu",
"onload_dtype": torch.bfloat16,
"onload_device": "cuda",
"preparing_dtype": torch.bfloat16,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-right.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right/ltx-2-19b-lora-camera-control-dolly-right.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-right shot: A happy girl looks up and says happily, 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' She sits before a sunlit café table, her open laptop displaying the Github interface. The camera glides right to show a barista crafting coffee in the background, shelves of artisan beans, and a chalkboard menu softly blurred in the bokeh."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_dolly_right_lora.mp4',
fps=24,
audio_sample_rate=24000,
)

75
examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Down.py Normal file
View File

@@ -0,0 +1,75 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.bfloat16,
"offload_device": "cpu",
"onload_dtype": torch.bfloat16,
"onload_device": "cuda",
"preparing_dtype": torch.bfloat16,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down",
allow_file_pattern="ltx-2-19b-lora-camera-control-jib-down.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down/ltx-2-19b-lora-camera-control-jib-down.safetensors", alpha=1.0, hotload=True)
prompt = (
"A girl is very happy, standing on a clean studio floor with soft ambient lighting. "
"She is speaking directly to the camera: “I enjoy working with Diffsynth-Studio, it's a perfect framework.” "
"The shot begins with a medium close-up framing her from the waist up. As the camera performs a smooth jib-down movement—"
"descending vertically downward—it gradually reveals more of the lower portion of the scene. "
"During the descent, the following elements become visible near the bottom of the frame: "
"- The polished concrete floor with subtle reflections of the girl’s shoes, "
"- A small branded mat labeled “Diffsynth-Studio” placed just beneath her feet, "
"- The lower part of a sleek workstation desk with a glowing logo on its front panel, partially hidden at the start but fully revealed as the camera lowers. "
"This downward motion provides a dynamic reveal of contextual details that reinforce the professional and creative environment, "
"while maintaining focus on the girl’s enthusiastic expression throughout."
)
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_camera_jib_down.mp4',
fps=24,
audio_sample_rate=24000,
)

69
examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Jib-Up.py Normal file
View File

@@ -0,0 +1,69 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.bfloat16,
"offload_device": "cpu",
"onload_dtype": torch.bfloat16,
"onload_device": "cuda",
"preparing_dtype": torch.bfloat16,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up",
allow_file_pattern="ltx-2-19b-lora-camera-control-jib-up.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up/ltx-2-19b-lora-camera-control-jib-up.safetensors", alpha=1.0, hotload=True)
prompt = (
"A girl stands happily at a sleek desk with a glowing 'Diffsynth-Studio' logo, saying: “I enjoy working with Diffsynth-Studio, it's a perfect framework.” "
"The shot starts low—framing her waist, shoes, and a branded floor mat—and smoothly jib-ups upward. "
"As the camera rises, it reveals her smiling face, upper body, and behind her: a bright creative studio with wall art and a large window showing daylight sky. "
"The final frame fully shows the inspiring workspace above the initial view, ensuring spatial continuity."
)
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_camera_jib_up.mp4',
fps=24,
audio_sample_rate=24000,
)

64
examples/ltx2/model_inference/LTX-2-T2AV-Camera-Control-Static.py Normal file
View File

@@ -0,0 +1,64 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.bfloat16,
"offload_device": "cpu",
"onload_dtype": torch.bfloat16,
"onload_device": "cuda",
"preparing_dtype": torch.bfloat16,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Static",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static",
allow_file_pattern="ltx-2-19b-lora-camera-control-static.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static/ltx-2-19b-lora-camera-control-static.safetensors", alpha=1.0, hotload=True)
prompt = "A beautiful sunset over the ocean."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_camera_static.mp4',
fps=24,
audio_sample_rate=24000,
)

65
examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-In.py Normal file
View File

@@ -0,0 +1,65 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.float8_e5m2,
"offload_device": "cpu",
"onload_dtype": torch.float8_e5m2,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e5m2,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-in.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-In/ltx-2-19b-lora-camera-control-dolly-in.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-in shot: A cheerful girl smiles brightly and says, 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' The camera smoothly moves closer to her face, highlighting her enthusiasm and sincerity."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_dolly_in_lora.mp4',
fps=24,
audio_sample_rate=24000,
)

65
examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Left.py Normal file
View File

@@ -0,0 +1,65 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.float8_e5m2,
"offload_device": "cpu",
"onload_dtype": torch.float8_e5m2,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e5m2,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-left.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Left/ltx-2-19b-lora-camera-control-dolly-left.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-left shot: A joyful young woman sits at a minimalist desk with a laptop running Diffsynth-Studio, code and generative visuals glowing on screen. She turns slightly toward the camera and says with a smile, 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' As she speaks, the camera smoothly dollies left, revealing a wall of framed open-source project posters, a whiteboard covered in neural network sketches, and a shelf stacked with AI/graphics books beside her."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_dolly_left_lora.mp4',
fps=24,
audio_sample_rate=24000,
)

66
examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Out.py Normal file
View File

@@ -0,0 +1,66 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.float8_e5m2,
"offload_device": "cpu",
"onload_dtype": torch.float8_e5m2,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e5m2,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-out.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Out/ltx-2-19b-lora-camera-control-dolly-out.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-out shot: A joyful young woman smiles warmly and says: 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' As she speaks, the camera slowly dollies out, revealing a bright, modern creative studio filled with plants, whiteboards full of diagrams, and soft natural light from large windows."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path=f'ltx2_dolly_out.mp4',
fps=24,
audio_sample_rate=24000,
)

65
examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Dolly-Right.py Normal file
View File

@@ -0,0 +1,65 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.float8_e5m2,
"offload_device": "cpu",
"onload_dtype": torch.float8_e5m2,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e5m2,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right",
allow_file_pattern="ltx-2-19b-lora-camera-control-dolly-right.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Dolly-Right/ltx-2-19b-lora-camera-control-dolly-right.safetensors", alpha=1.0, hotload=True)
prompt = "Dolly-right shot: A happy girl looks up and says happily, 'I enjoy working with Diffsynth-Studio, it's a perfect framework.' She sits before a sunlit café table, her open laptop displaying the Github interface. The camera glides right to show a barista crafting coffee in the background, shelves of artisan beans, and a chalkboard menu softly blurred in the bokeh."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_dolly_right_lora.mp4',
fps=24,
audio_sample_rate=24000,
)

75
examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Down.py Normal file
View File

@@ -0,0 +1,75 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.float8_e5m2,
"offload_device": "cpu",
"onload_dtype": torch.float8_e5m2,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e5m2,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down",
allow_file_pattern="ltx-2-19b-lora-camera-control-jib-down.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Down/ltx-2-19b-lora-camera-control-jib-down.safetensors", alpha=1.0, hotload=True)
prompt = (
"A girl is very happy, standing on a clean studio floor with soft ambient lighting. "
"She is speaking directly to the camera: “I enjoy working with Diffsynth-Studio, it's a perfect framework.” "
"The shot begins with a medium close-up framing her from the waist up. As the camera performs a smooth jib-down movement—"
"descending vertically downward—it gradually reveals more of the lower portion of the scene. "
"During the descent, the following elements become visible near the bottom of the frame: "
"- The polished concrete floor with subtle reflections of the girl’s shoes, "
"- A small branded mat labeled “Diffsynth-Studio” placed just beneath her feet, "
"- The lower part of a sleek workstation desk with a glowing logo on its front panel, partially hidden at the start but fully revealed as the camera lowers. "
"This downward motion provides a dynamic reveal of contextual details that reinforce the professional and creative environment, "
"while maintaining focus on the girl’s enthusiastic expression throughout."
)
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_camera_jib_down.mp4',
fps=24,
audio_sample_rate=24000,
)

69
examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Jib-Up.py Normal file
View File

@@ -0,0 +1,69 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.float8_e5m2,
"offload_device": "cpu",
"onload_dtype": torch.float8_e5m2,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e5m2,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up",
allow_file_pattern="ltx-2-19b-lora-camera-control-jib-up.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Jib-Up/ltx-2-19b-lora-camera-control-jib-up.safetensors", alpha=1.0, hotload=True)
prompt = (
"A girl stands happily at a sleek desk with a glowing 'Diffsynth-Studio' logo, saying: “I enjoy working with Diffsynth-Studio, it's a perfect framework.” "
"The shot starts low—framing her waist, shoes, and a branded floor mat—and smoothly jib-ups upward. "
"As the camera rises, it reveals her smiling face, upper body, and behind her: a bright creative studio with wall art and a large window showing daylight sky. "
"The final frame fully shows the inspiring workspace above the initial view, ensuring spatial continuity."
)
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_camera_jib_up.mp4',
fps=24,
audio_sample_rate=24000,
)

64
examples/ltx2/model_inference_low_vram/LTX-2-T2AV-Camera-Control-Static.py Normal file
View File

@@ -0,0 +1,64 @@
import torch
from diffsynth.pipelines.ltx2_audio_video import LTX2AudioVideoPipeline, ModelConfig
from diffsynth.utils.data.media_io_ltx2 import write_video_audio_ltx2
from modelscope import snapshot_download
vram_config = {
"offload_dtype": torch.float8_e5m2,
"offload_device": "cpu",
"onload_dtype": torch.float8_e5m2,
"onload_device": "cpu",
"preparing_dtype": torch.float8_e5m2,
"preparing_device": "cuda",
"computation_dtype": torch.bfloat16,
"computation_device": "cuda",
}
pipe = LTX2AudioVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized", origin_file_pattern="model-*.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-dev.safetensors", **vram_config),
ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-spatial-upscaler-x2-1.0.safetensors", **vram_config),
],
tokenizer_config=ModelConfig(model_id="google/gemma-3-12b-it-qat-q4_0-unquantized"),
stage2_lora_config=ModelConfig(model_id="Lightricks/LTX-2", origin_file_pattern="ltx-2-19b-distilled-lora-384.safetensors"),
)
snapshot_download(
"Lightricks/LTX-2-19b-LoRA-Camera-Control-Static",
local_dir="models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static",
allow_file_pattern="ltx-2-19b-lora-camera-control-static.safetensors",
)
pipe.load_lora(pipe.dit, "models/Lightricks/LTX-2-19b-LoRA-Camera-Control-Static/ltx-2-19b-lora-camera-control-static.safetensors", alpha=1.0, hotload=True)
prompt = "A beautiful sunset over the ocean."
negative_prompt = (
"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, 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, wrong gaze direction, "
"mismatched lip sync, silent or muted audio, distorted voice, robotic voice, echo, background noise, "
"off-sync audio, incorrect dialogue, added dialogue, repetitive speech, jittery movement, awkward "
"pauses, incorrect timing, unnatural transitions, inconsistent framing, tilted camera, flat lighting, "
"inconsistent tone, cinematic oversaturation, stylized filters, or AI artifacts."
)
height, width, num_frames = 512 * 2, 768 * 2, 121
video, audio = pipe(
prompt=prompt,
negative_prompt=negative_prompt,
seed=43,
height=height,
width=width,
num_frames=num_frames,
tiled=True,
use_two_stage_pipeline=True,
)
write_video_audio_ltx2(
video=video,
audio=audio,
output_path='ltx2_camera_static.mp4',
fps=24,
audio_sample_rate=24000,
)