Spaces:
Running
on
Zero
Running
on
Zero
File size: 5,317 Bytes
22a452a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 |
# Copyright 2025 The HuggingFace Team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import unittest
import numpy as np
import torch
from diffusers import AutoencoderKLLTXVideo, LTXLatentUpsamplePipeline
from diffusers.pipelines.ltx.modeling_latent_upsampler import LTXLatentUpsamplerModel
from diffusers.utils.testing_utils import enable_full_determinism
from ..test_pipelines_common import PipelineTesterMixin, to_np
enable_full_determinism()
class LTXLatentUpsamplePipelineFastTests(PipelineTesterMixin, unittest.TestCase):
pipeline_class = LTXLatentUpsamplePipeline
params = {"video", "generator"}
batch_params = {"video", "generator"}
required_optional_params = frozenset(["generator", "latents", "return_dict"])
test_xformers_attention = False
supports_dduf = False
def get_dummy_components(self):
torch.manual_seed(0)
vae = AutoencoderKLLTXVideo(
in_channels=3,
out_channels=3,
latent_channels=8,
block_out_channels=(8, 8, 8, 8),
decoder_block_out_channels=(8, 8, 8, 8),
layers_per_block=(1, 1, 1, 1, 1),
decoder_layers_per_block=(1, 1, 1, 1, 1),
spatio_temporal_scaling=(True, True, False, False),
decoder_spatio_temporal_scaling=(True, True, False, False),
decoder_inject_noise=(False, False, False, False, False),
upsample_residual=(False, False, False, False),
upsample_factor=(1, 1, 1, 1),
timestep_conditioning=False,
patch_size=1,
patch_size_t=1,
encoder_causal=True,
decoder_causal=False,
)
vae.use_framewise_encoding = False
vae.use_framewise_decoding = False
torch.manual_seed(0)
latent_upsampler = LTXLatentUpsamplerModel(
in_channels=8,
mid_channels=32,
num_blocks_per_stage=1,
dims=3,
spatial_upsample=True,
temporal_upsample=False,
)
components = {
"vae": vae,
"latent_upsampler": latent_upsampler,
}
return components
def get_dummy_inputs(self, device, seed=0):
if str(device).startswith("mps"):
generator = torch.manual_seed(seed)
else:
generator = torch.Generator(device=device).manual_seed(seed)
video = torch.randn((5, 3, 32, 32), generator=generator, device=device)
inputs = {
"video": video,
"generator": generator,
"height": 16,
"width": 16,
"output_type": "pt",
}
return inputs
def test_inference(self):
device = "cpu"
components = self.get_dummy_components()
pipe = self.pipeline_class(**components)
pipe.to(device)
pipe.set_progress_bar_config(disable=None)
inputs = self.get_dummy_inputs(device)
video = pipe(**inputs).frames
generated_video = video[0]
self.assertEqual(generated_video.shape, (5, 3, 32, 32))
expected_video = torch.randn(5, 3, 32, 32)
max_diff = np.abs(generated_video - expected_video).max()
self.assertLessEqual(max_diff, 1e10)
def test_vae_tiling(self, expected_diff_max: float = 0.25):
generator_device = "cpu"
components = self.get_dummy_components()
pipe = self.pipeline_class(**components)
pipe.to("cpu")
pipe.set_progress_bar_config(disable=None)
# Without tiling
inputs = self.get_dummy_inputs(generator_device)
inputs["height"] = inputs["width"] = 128
output_without_tiling = pipe(**inputs)[0]
# With tiling
pipe.vae.enable_tiling(
tile_sample_min_height=96,
tile_sample_min_width=96,
tile_sample_stride_height=64,
tile_sample_stride_width=64,
)
inputs = self.get_dummy_inputs(generator_device)
inputs["height"] = inputs["width"] = 128
output_with_tiling = pipe(**inputs)[0]
self.assertLess(
(to_np(output_without_tiling) - to_np(output_with_tiling)).max(),
expected_diff_max,
"VAE tiling should not affect the inference results",
)
@unittest.skip("Test is not applicable.")
def test_callback_inputs(self):
pass
@unittest.skip("Test is not applicable.")
def test_attention_slicing_forward_pass(
self, test_max_difference=True, test_mean_pixel_difference=True, expected_max_diff=1e-3
):
pass
@unittest.skip("Test is not applicable.")
def test_inference_batch_consistent(self):
pass
@unittest.skip("Test is not applicable.")
def test_inference_batch_single_identical(self):
pass
|