見出し画像

WSL2でVistaDreamを試してみる

noguchi-shoji

「単一ビュー画像から3Dシーンを再構築する新しいフレームワーク。最近の拡散モデルにより、単一ビューの入力画像から高品質の新しいビュー画像を生成できる」らしいVistaDreamを試してみます。

使用するPCはドスパラさんの「GALLERIA UL9C-R49」。スペックは
・CPU: Intel® Core™ i9-13900HX Processor
・Mem: 64 GB
・GPU: NVIDIA® GeForce RTX™ 4090 Laptop GPU(16GB)
・GPU: NVIDIA® GeForce RTX™ 4090 (24GB)
・OS: Ubuntu22.04 on WSL2（Windows 11）
です。


1. 準備

環境構築

python3 -m venv vistadream
cd $_
source bin/activate

リポジトリをクローンして、インストールします。

git clone https://github.com/WHU-USI3DV/VistaDream.git
cd VistaDream

requirements.txtがないので、environment.yamlから変換してます。

envfile=./environment.yaml 
grep -n "pip:" ${envfile} \
 	| cut -d":" -f1 \
 	| xargs -I{} expr {} + 1 \
 	| xargs -I{} tail -n +{} ${envfile} \
 	| grep -e '^[[:space:]]*- ' \
 	| sed 's/^[[:space:]]*- //' 	> requirements.txt

requirements.txtプラスαのパッケージのインストールです。

pip install wheel
pip install torchmetrics==1.3.2
# detectron2
pip install 'git+https://github.com/facebookresearch/detectron2.git'
#
pip install -r requirements.txt

続いて、tools配下のパッケージをインストールしていきます。

cd tools/DepthPro
pip install -e .
cd ../..

MultiScaleDeformableAttention

pushd tools/OneFormer/oneformer/modeling/pixel_decoder/ops/
sh make.sh
popd

モデルのダウンロード

tools下の所定のディレクトリにモデルをダウンロードします。

bash download_weights.sh

2. ファイルの修正

※※※　こちらの修正は、最新コードで対処済みとなっています　※※※

pipe/cfgs/basic.yaml内のckptパラメータ値内のディレクトリ名に誤り、シングルクォーテーションが足りない…とかがあるため、修正します。

diff --git a/pipe/cfgs/basic.yaml b/pipe/cfgs/basic.yaml
index 44bb4eb..cf86633 100644
--- a/pipe/cfgs/basic.yaml
+++ b/pipe/cfgs/basic.yaml
@@ -4,14 +4,14 @@ model:
   sky:
     value: 1e5 # to be update
     oneformer:
-      ckpt: 'tools/OneFormer/ckpts/coco_pretrain_1280x1280_150_16_dinat_l_oneformer_ade20k_160k.pth'
+      ckpt: 'tools/OneFormer/checkpoints/coco_pretrain_1280x1280_150_16_dinat_l_oneformer_ade20k_160k.pth'
       yaml: 'tools/OneFormer/configs/ade20k/dinat/coco_pretrain_oneformer_dinat_large_bs16_160k_1280x1280.yaml'
   vlm:
     llava:
       ckpt: 'llava-hf/bakLlava-v1-hf' # downloaded from hugging face
   mde:
     dpt_pro:
-      ckpt: tools/DepthPro/checkpoints/depth_pro.pt'
+      ckpt: 'tools/DepthPro/checkpoints/depth_pro.pt'
   paint:
     fooocus:
       pass # it will load required checkpoints automaticly


3. 試してみる

(1) プロセスkillされる…

試してみます。

CUDA_VISIBLE_DEVICES=0 python demo.py

いろいろたくさん表示され、、、

/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:48: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  @custom_fwd(cast_inputs=torch.float16)
/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:61: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  def backward(ctx, grad_out):
/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:82: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  @custom_fwd(cast_inputs=torch.float16)
/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:94: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  def backward(ctx, grad_out):
/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:115: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  @custom_fwd(cast_inputs=torch.float16)
/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:128: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  def backward(ctx, grad_out):
/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:149: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
  @custom_fwd(cast_inputs=torch.float16)
/path/to/venv/vistadreamlib/python3.10/site-packages/natten/functional.py:161: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
  def backward(ctx, grad_out):
/path/to/venv/vistadreamVistaDream/tools/OneFormer/oneformer/modeling/pixel_decoder/msdeformattn.py:318: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
  @autocast(enabled=False)
Load default preset failed.
[Errno 2] No such file or directory: '/path/to/venv/vistadreamVistaDream/presets/default.json'
No presets found.
No presets found.
Total VRAM 24564 MB, total RAM 31948 MB
Set vram state to: NORMAL_VRAM
Always offload VRAM
Device: cuda:0 NVIDIA GeForce RTX 4090 : native
VAE dtype: torch.bfloat16
Using pytorch cross attention
Refiner unloaded.
model_type EPS
UNet ADM Dimension 2816
Using pytorch attention in VAE
Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
Using pytorch attention in VAE
extra {'cond_stage_model.clip_l.logit_scale', 'cond_stage_model.clip_l.text_projection'}
left over keys: dict_keys(['cond_stage_model.clip_l.transformer.text_model.embeddings.position_ids'])
Base model loaded: /path/to/venv/vistadreamVistaDream/tools/Fooocus/models/checkpoints/juggernautXL_v8Rundiffusion.safetensors
VAE loaded: None
Request to load LoRAs [] for model [/path/to/venv/vistadreamVistaDream/tools/Fooocus/models/checkpoints/juggernautXL_v8Rundiffusion.safetensors].
Fooocus V2 Expansion: Vocab with 642 words.
Fooocus Expansion engine loaded for cuda:0, use_fp16 = True.
Requested to load SDXLClipModel
Requested to load GPT2LMHeadModel
Loading 2 new models
[Fooocus Model Management] Moving model(s) has taken 1.01 seconds
Started worker with PID 11607
/path/to/venv/vistadreamlib/python3.10/site-packages/diffusers/utils/outputs.py:63: FutureWarning: `torch.utils._pytree._register_pytree_node` is deprecated. Please use `torch.utils._pytree.register_pytree_node` instead.
  torch.utils._pytree._register_pytree_node(
Loading config tools/OneFormer/configs/ade20k/dinat/coco_pretrain_oneformer_dinat_large_bs16_160k_1280x1280.yaml with yaml.unsafe_load. Your machine may be at risk if the file contains malicious content.
Loading config tools/OneFormer/configs/ade20k/dinat/../Base-ADE20K-UnifiedSegmentation.yaml with yaml.unsafe_load. Your machine may be at risk if the file contains malicious content.
/path/to/venv/vistadreamVistaDream/tools/Fooocus/modules/patch.py:465: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  result = original_loader(*args, **kwargs)
The checkpoint state_dict contains keys that are not used by the model:
  text_encoder.positional_embedding
  text_encoder.transformer.resblocks.0.attn.{in_proj_bias, in_proj_weight}
  text_encoder.transformer.resblocks.0.attn.out_proj.{bias, weight}
  text_encoder.transformer.resblocks.0.ln_1.{bias, weight}
  text_encoder.transformer.resblocks.0.mlp.c_fc.{bias, weight}
  text_encoder.transformer.resblocks.0.mlp.c_proj.{bias, weight}
  text_encoder.transformer.resblocks.0.ln_2.{bias, weight}
  text_encoder.transformer.resblocks.1.attn.{in_proj_bias, in_proj_weight}
  text_encoder.transformer.resblocks.1.attn.out_proj.{bias, weight}
  text_encoder.transformer.resblocks.1.ln_1.{bias, weight}
  text_encoder.transformer.resblocks.1.mlp.c_fc.{bias, weight}
  text_encoder.transformer.resblocks.1.mlp.c_proj.{bias, weight}
  text_encoder.transformer.resblocks.1.ln_2.{bias, weight}
  text_encoder.transformer.resblocks.2.attn.{in_proj_bias, in_proj_weight}
  text_encoder.transformer.resblocks.2.attn.out_proj.{bias, weight}
  text_encoder.transformer.resblocks.2.ln_1.{bias, weight}
  text_encoder.transformer.resblocks.2.mlp.c_fc.{bias, weight}
  text_encoder.transformer.resblocks.2.mlp.c_proj.{bias, weight}
  text_encoder.transformer.resblocks.2.ln_2.{bias, weight}
  text_encoder.transformer.resblocks.3.attn.{in_proj_bias, in_proj_weight}
  text_encoder.transformer.resblocks.3.attn.out_proj.{bias, weight}
  text_encoder.transformer.resblocks.3.ln_1.{bias, weight}
  text_encoder.transformer.resblocks.3.mlp.c_fc.{bias, weight}
  text_encoder.transformer.resblocks.3.mlp.c_proj.{bias, weight}
  text_encoder.transformer.resblocks.3.ln_2.{bias, weight}
  text_encoder.transformer.resblocks.4.attn.{in_proj_bias, in_proj_weight}
  text_encoder.transformer.resblocks.4.attn.out_proj.{bias, weight}
  text_encoder.transformer.resblocks.4.ln_1.{bias, weight}
  text_encoder.transformer.resblocks.4.mlp.c_fc.{bias, weight}
  text_encoder.transformer.resblocks.4.mlp.c_proj.{bias, weight}
  text_encoder.transformer.resblocks.4.ln_2.{bias, weight}
  text_encoder.transformer.resblocks.5.attn.{in_proj_bias, in_proj_weight}
  text_encoder.transformer.resblocks.5.attn.out_proj.{bias, weight}
  text_encoder.transformer.resblocks.5.ln_1.{bias, weight}
  text_encoder.transformer.resblocks.5.mlp.c_fc.{bias, weight}
  text_encoder.transformer.resblocks.5.mlp.c_proj.{bias, weight}
  text_encoder.transformer.resblocks.5.ln_2.{bias, weight}
  text_encoder.ln_final.{bias, weight}
  text_encoder.token_embedding.weight
  text_projector.layers.0.{bias, weight}
  text_projector.layers.1.{bias, weight}
  prompt_ctx.weight
/path/to/venv/vistadreamlib/python3.10/site-packages/transformers/models/llava/configuration_llava.py:100: FutureWarning: The `vocab_size` argument is deprecated and will be removed in v4.42, since it can be inferred from the `text_config`. Passing this argument has no effect
  warnings.warn(
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████| 4/4 [00:01<00:00,  3.77it/s]
Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
Inpaint-Caption[1/3] Move llava.model to GPU...
Inpaint-Caption[2/3] Llava inpainting instruction:
We detected that you are passing `past_key_values` as a tuple and this is deprecated and will be removed in v4.43. Please use an appropriate `Cache` class (https://huggingface.co/docs/transformers/v4.41.3/en/internal/generation_utils#transformers.Cache)
 a cozy library room in a house. The room features a large bookshelf filled with numerous books, a chair, a couch, and a coffee table. The books on the shelves come in various sizes and colors, creating an inviting atmosphere for reading and relaxation. The chair and couch provide comfortable seating options for people to enjoy the space and indulge in their favorite books. The coffee table, placed between the chair and couch, adds functionality to the room, allowing for the placement of drinks, snacks, or other items. Overall, the room exudes a warm and welcoming ambiance, perfect for spending quality time alone or with others who share a love for books and reading.
Inpaint-Caption[3/3] Move llava.model to CPU...
Inpaint-Fooocus[1/2] Fooocus inpainting...
[Parameters] Adaptive CFG = 7.0
[Parameters] CLIP Skip = 2
[Parameters] Sharpness = 2.0
[Parameters] ControlNet Softness = 0.25
[Parameters] ADM Scale = 1.5 : 0.8 : 0.3
[Parameters] CFG = 7.0
[Parameters] Seed = 7106521602475165645
[Fooocus] Downloading upscale models ...
[Fooocus] Downloading inpainter ...
[Inpaint] Current inpaint model is /path/to/venv/vistadreamVistaDream/tools/Fooocus/models/inpaint/inpaint_v26.fooocus.patch
[Parameters] Sampler = dpmpp_2m_sde_gpu - karras
[Parameters] Steps = 30 - 24
[Fooocus] Initializing ...
[Fooocus] Loading models ...
Synthetic Refiner Activated
Synthetic Refiner Activated
Request to load LoRAs [('/path/to/venv/vistadreamVistaDream/tools/Fooocus/models/inpaint/inpaint_v26.fooocus.patch', 1.0)] for model [/path/to/venv/vistadreamVistaDream/tools/Fooocus/models/checkpoints/juggernautXL_v8Rundiffusion.safetensors].
Loaded LoRA [/path/to/venv/vistadreamVistaDream/tools/Fooocus/models/inpaint/inpaint_v26.fooocus.patch] for UNet [/path/to/venv/vistadreamVistaDream/tools/Fooocus/models/checkpoints/juggernautXL_v8Rundiffusion.safetensors] with 960 keys at weight 1.0.
Request to load LoRAs [] for model [/path/to/venv/vistadreamVistaDream/tools/Fooocus/models/checkpoints/juggernautXL_v8Rundiffusion.safetensors].
Requested to load SDXLClipModel
Loading 1 new model
unload clone 1
[Fooocus Model Management] Moving model(s) has taken 1.48 seconds
[Fooocus] Processing prompts ...
[Fooocus] Preparing Fooocus text #1 ...
[Prompt Expansion] a cozy library room in a house. The room features a large bookshelf filled with numerous books, a chair, a couch, and a coffee table. The books on the shelves come in various sizes and colors, creating an inviting atmosphere for reading and relaxation. The chair and couch provide comfortable seating options for people to enjoy the space and indulge in their favorite books. The coffee table, placed between the chair and couch, adds functionality to the room, allowing for the placement of drinks, snacks, or other items. Overall, the room exudes a warm and welcoming ambiance, perfect for spending quality time alone or with others who share a love for books and reading. 8K, no large circles, no cameras, no fisheye, beautiful, glowing, crisp, focus, very inspirational, bright colorful, cinematic, dynamic light, epic composition, elegant, rich deep color, stunning intricate detailed artistic surreal background, creative, positive, unique, healing, pure, rational, loving, vibrant, calm, relaxed, emotional, whole, joyful, complex, highly enigmatic, amazing, lucid, thought, magic
[Fooocus] Encoding positive #1 ...
[Fooocus] Encoding negative #1 ...
[Fooocus] Image processing ...
Upscaling image with shape (920, 920, 3) ...
[Fooocus] VAE Inpaint encoding ...
Requested to load AutoencoderKL
Loading 1 new model
[Fooocus Model Management] Moving model(s) has taken 0.81 seconds
[Fooocus] VAE encoding ...
Final resolution is (920, 920), latent is (1024, 1024).
[Parameters] Denoising Strength = 1.0
[Parameters] Initial Latent shape: torch.Size([1, 4, 128, 128])
Preparation time: 8.49 seconds
[Fooocus] Preparing task 1/1 ...
[Sampler] refiner_swap_method = joint
[Sampler] sigma_min = 0.0291671771556139, sigma_max = 14.614643096923828
Requested to load SDXL
Loading 1 new model
[Fooocus Model Management] Moving model(s) has taken 3.11 seconds
 80%|████████████████████████████████████████████████████████████████████▊                 | 24/30 [00:04<00:01,  5.25it/s]Requested to load SDXL
Loading 1 new model
unload clone 0
[Fooocus Model Management] Moving model(s) has taken 3.01 seconds
Refiner Swapped
100%|██████████████████████████████████████████████████████████████████████████████████████| 30/30 [00:08<00:00,  3.40it/s]
Requested to load AutoencoderKL
Loading 1 new model
[Fooocus Model Management] Moving model(s) has taken 1.86 seconds
Generating and saving time: 14.07 seconds
Inpaint-Fooocus[2/2] Assign Frame...
Pro_dpt[1/3] Move Pro_dpt.model to GPU...
Pro_dpt[2/3] Pro_dpt Estimation...
Pro_dpt[3/3] Move Pro_dpt.model to GPU...
Pro_dpt[1/3] Move Pro_dpt.model to GPU...
Pro_dpt[2/3] Pro_dpt Estimation...
Pro_dpt[3/3] Move Pro_dpt.model to GPU...
/path/to/venv/vistadreamlib/python3.10/site-packages/torch/functional.py:534: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at ../aten/src/ATen/native/TensorShape.cpp:3595.)
  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
  0%|                                                                                              | 0/100 [00:00<?, ?it/s]( ●    ) gsplat: Setting up CUDA (This may take a few minutes the first time)Killed

Killed

プロセス、Killされた…。

こんなことで負けていられません（何に。とりあえず、WSL2に割り当てているメモリを増量します。

PS C:> type .\.wslconfig
# Settings apply across all Linux distros running on WSL 2
[wsl2]

# Limits VM memory to use no more than 4 GB, this can be set as whole numbers using GB or MB
memory=49GB

# Sets amount of swap storage space to 8GB, default is 25% of available RAM
#swap=8GB

#networkingMode=mirrored
#dnsTunneling=true
#firewall=false

32GBから49GBに増量して再度試したところ、プロセスkillされずに先に進みました。CPUもGPUもメモリが大事ですね。

(2) JSONファイルが見つからない

※※※　こちらのエラーは、最新コードで対処済みとなっています　※※※

しばし待つと、またもやエラー。今度はJSONファイルが見つからない、らしい。

Traceback (most recent call last):
  File "/path/to/venv/vistadream/VistaDream/demo.py", line 7, in <module>
    vistadream()
  File "/path/to/venv/vistadream/VistaDream/pipe/c2f_recons.py", line 182, in __call__
    self._MCS_Refinement()
  File "/path/to/venv/vistadream/VistaDream/pipe/c2f_recons.py", line 149, in _MCS_Refinement
    refiner = HackSD_MCS(device='cuda',use_lcm=True,denoise_steps=self.mcs_iterations)
  File "/path/to/venv/vistadream/VistaDream/ops/mcs.py", line 33, in __init__
    self._load_model()
  File "/path/to/venv/vistadream/VistaDream/ops/mcs.py", line 39, in _load_model
    self.model = Hack_SDPipe_Stepwise.from_pretrained(self.sd_ckpt)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
    return fn(*args, **kwargs)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/diffusers/pipelines/pipeline_utils.py", line 1117, in from_pretrained
    config_dict = cls.load_config(cached_folder)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
    return fn(*args, **kwargs)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/diffusers/configuration_utils.py", line 370, in load_config
    raise EnvironmentError(
OSError: Error no file named model_index.json found in directory tools/StableDiffusion/ckpt.
Traceback (most recent call last):
  File "/path/to/venv/vistadream/VistaDream/demo.py", line 7, in <module>
    vistadream()
  File "/path/to/venv/vistadream/VistaDream/pipe/c2f_recons.py", line 182, in __call__
    self._MCS_Refinement()
  File "/path/to/venv/vistadream/VistaDream/pipe/c2f_recons.py", line 149, in _MCS_Refinement
    refiner = HackSD_MCS(device='cuda',use_lcm=True,denoise_steps=self.mcs_iterations)
  File "/path/to/venv/vistadream/VistaDream/ops/mcs.py", line 33, in __init__
    self._load_model()
  File "/path/to/venv/vistadream/VistaDream/ops/mcs.py", line 39, in _load_model
    self.model = Hack_SDPipe_Stepwise.from_pretrained(self.sd_ckpt)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
    return fn(*args, **kwargs)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/diffusers/pipelines/pipeline_utils.py", line 1271, in from_pretrained
    loaded_sub_model = load_sub_model(
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/diffusers/pipelines/pipeline_utils.py", line 528, in load_sub_model
    loaded_sub_model = load_method(cached_folder, **loading_kwargs)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/transformers/image_processing_base.py", line 206, in from_pretrained
    image_processor_dict, kwargs = cls.get_image_processor_dict(pretrained_model_name_or_path, **kwargs)
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/transformers/image_processing_base.py", line 335, in get_image_processor_dict
    resolved_image_processor_file = cached_file(
  File "/path/to/venv/vistadream/lib/python3.10/site-packages/transformers/utils/hub.py", line 373, in cached_file
    raise EnvironmentError(
OSError: tools/StableDiffusion/ckpt does not appear to have a file named preprocessor_config.json. Checkout 'https://huggingface.co/tools/StableDiffusion/ckpt/tree/main' for available files.

さすがに存在しないファイルはどうしようもないので、issueを登録する。
すると15分後にcontributorから回答あり。「commitしたから、pullしてみてくれ」ということで、git pull して試します。

(3) 再度試します

CUDA_VISIBLE_DEVICES=0 python demo.py

しばし待つと、完了しそうなメッセージが！

100%|██████████████████████████████████████████████████████████████████████████████████████| 10/10 [04:28<00:00, 26.84s/it]
[INFO] rendering final video with 250 frames...

正常終了。そして、ファイルを探すと、

$ find . -type f -ls | grep 'Oct 25'
 30160965     412 -rw-r--r--   1 users users     419478 Oct 25 00:13 ./data/sd_readingroom/video_dpt.mp4
 30160924  215928 -rw-r--r--   1 users users  221107092 Oct 25 00:12 ./data/sd_readingroom/scene.pth
 30160964     944 -rw-r--r--   1 users users     963279 Oct 25 00:13 ./data/sd_readingroom/video_rgb.mp4
$

ありました！

これら.mp4がこちら。

ちゃんと動きました。

4. まとめ

リソース使用状況を中心にまとめておきます。

  • MEMは32GBでは足りず、プロセスkillされる。（弊環境では49GBで正常終了）

  • GPU VRAM使用量は23.8GBほど。24GBないと厳しい。

VRAM使用量

この記事が気に入ったらサポートをしてみませんか？