Video · text-to-video · image-to-video

VideoCrafter

The VideoCrafter2 HF card exposes Apache-2.0; older Text2Video/Image2Video cards need license review.

Integrated统一环境源码已固定videocrafter

构建运行命令

选择仓库中已记录的 variant，然后复制对应的准备、检查或运行命令。

worldfoundry.pipeline

Variant

任务text-to-video

环境worldfoundry-unified-cu128

设备CUDA 12.8

worldfoundry-eval evaluate \
  --mode model \
  --model-id videocrafter \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/videocrafter \
  --metric artifact_count \
  --json

兼容性与版本

由 Manifest 生成

集成状态: integrated
Runner 证据: pending
环境: worldfoundry-unified-cu128
Python: 3.11
CUDA: CUDA 12.8
PyTorch: torch
源码 revision: f6c7f7f28f71d3069a9ee0e002d2f4b019eb78f6
Checkpoint revision: 7b4079838622fcba3fc2c9575210df3ed1cb8d69

Runtime profile: videocrafter2-t2v
Pipeline binding: videocrafter
Runner: worldfoundry.pipeline
Pipeline target: worldfoundry.pipelines.videocrafter.pipeline_videocrafter2_t2v:VideoCrafter2T2VPipeline
Backend stage: official_in_tree_runtime
Runtime 状态: official_in_tree_runtime_ready
Driver 状态: compatible_unified_default
环境类型: 统一环境

安装环境

环境解析器会读取已记录 profile，并选择此处显示的统一或独立环境。

bash scripts/setup/model_env_install.sh --model videocrafter

依赖版本约束 4

torch
torchvision
torchaudio
xfuser

Conda 依赖 5

python=3.11
pip
ffmpeg
setuptools
wheel

Checkpoint 与资产

分配算力前先做本地检查；gated、private 与 license 字段直接来自 checkpoint manifest。

worldfoundry-eval zoo model-download --model-id videocrafter --check-local --json

VideoCrafter/VideoCrafter2

Revision: 7b4079838622fcba3fc2c9575210df3ed1cb8d69
License: apache-2.0
Gated: false
Private: false

VideoCrafter/Text2Video-1024

Revision: 3919b21f62d5c14498e4d25c831c6df031da18a6
License: —
Gated: false
Private: false

VideoCrafter/Image2Video-512

Revision: 36d5fca839f041121c576b1cd76c58d4eb238408
License: —
Gated: false
Private: false

VideoCrafter/VideoCrafter2

Revision: —
License: —
Gated: —
Private: —

运行与输出

生成的命令通过共享 evaluation 边界运行，并持久保存结果 manifest 与 artifact。

worldfoundry-eval evaluate \
  --mode model \
  --model-id videocrafter \
  --model-runner worldfoundry:pipeline \
  --model-manifest-dir worldfoundry/data/models/catalog \
  --requests-path tmp/requests.jsonl \
  --output-dir tmp/model_eval/videocrafter \
  --metric artifact_count \
  --json

输入契约

字段	已记录契约
`prompt`	Required
`image`	Optional
`video`	Optional

Artifact 契约

Artifact 类型	文件名 / 路径
`generated_video`	`videocrafter2-t2v.mp4`

证据与来源

Catalog 集成、原生 demo parity 与 runner parity 是三条独立记录。

集成状态: integrated
Runner 证据: pending
原生 Demo 证据: pending
验证 Imports: torch, torchvision, numpy, transformers, diffusers, xfuser, worldfoundry

配方溯源

worldfoundry/data/models/catalog/video/videocrafter.yamlGitHubf6c7f7f28f71d3069a9ee0e002d2f4b019eb78f6

构建运行命令

输入契约

Artifact 契约

相关配方