diff --git a/src/diffusers/pipelines/sana/pipeline_sana_sprint.py b/src/diffusers/pipelines/sana/pipeline_sana_sprint.py index 5a0bc81619f7..7c17c2708216 100644 --- a/src/diffusers/pipelines/sana/pipeline_sana_sprint.py +++ b/src/diffusers/pipelines/sana/pipeline_sana_sprint.py @@ -40,11 +40,7 @@ ) from ...utils.torch_utils import randn_tensor from ..pipeline_utils import DiffusionPipeline -from ..pixart_alpha.pipeline_pixart_alpha import ( - ASPECT_RATIO_512_BIN, - ASPECT_RATIO_1024_BIN, -) -from ..pixart_alpha.pipeline_pixart_sigma import ASPECT_RATIO_2048_BIN +from ..pixart_alpha.pipeline_pixart_alpha import ASPECT_RATIO_1024_BIN from .pipeline_output import SanaPipelineOutput @@ -64,49 +60,6 @@ import ftfy -ASPECT_RATIO_4096_BIN = { - "0.25": [2048.0, 8192.0], - "0.26": [2048.0, 7936.0], - "0.27": [2048.0, 7680.0], - "0.28": [2048.0, 7424.0], - "0.32": [2304.0, 7168.0], - "0.33": [2304.0, 6912.0], - "0.35": [2304.0, 6656.0], - "0.4": [2560.0, 6400.0], - "0.42": [2560.0, 6144.0], - "0.48": [2816.0, 5888.0], - "0.5": [2816.0, 5632.0], - "0.52": [2816.0, 5376.0], - "0.57": [3072.0, 5376.0], - "0.6": [3072.0, 5120.0], - "0.68": [3328.0, 4864.0], - "0.72": [3328.0, 4608.0], - "0.78": [3584.0, 4608.0], - "0.82": [3584.0, 4352.0], - "0.88": [3840.0, 4352.0], - "0.94": [3840.0, 4096.0], - "1.0": [4096.0, 4096.0], - "1.07": [4096.0, 3840.0], - "1.13": [4352.0, 3840.0], - "1.21": [4352.0, 3584.0], - "1.29": [4608.0, 3584.0], - "1.38": [4608.0, 3328.0], - "1.46": [4864.0, 3328.0], - "1.67": [5120.0, 3072.0], - "1.75": [5376.0, 3072.0], - "2.0": [5632.0, 2816.0], - "2.09": [5888.0, 2816.0], - "2.4": [6144.0, 2560.0], - "2.5": [6400.0, 2560.0], - "2.89": [6656.0, 2304.0], - "3.0": [6912.0, 2304.0], - "3.11": [7168.0, 2304.0], - "3.62": [7424.0, 2048.0], - "3.75": [7680.0, 2048.0], - "3.88": [7936.0, 2048.0], - "4.0": [8192.0, 2048.0], -} - EXAMPLE_DOC_STRING = """ Examples: ```py @@ -118,7 +71,7 @@ ... ) >>> pipe.to("cuda") - >>> image = pipe(prompt='a cyberpunk cat with a neon sign that says "Sana"')[0] + >>> image = pipe(prompt='a tiny astronaut hatching from an egg on the moon')[0] >>> image[0].save("output.png") ``` """ @@ -762,14 +715,8 @@ def __call__( # 1. Check inputs. Raise error if not correct if use_resolution_binning: - if self.transformer.config.sample_size == 128: - aspect_ratio_bin = ASPECT_RATIO_4096_BIN - elif self.transformer.config.sample_size == 64: - aspect_ratio_bin = ASPECT_RATIO_2048_BIN - elif self.transformer.config.sample_size == 32: + if self.transformer.config.sample_size == 32: aspect_ratio_bin = ASPECT_RATIO_1024_BIN - elif self.transformer.config.sample_size == 16: - aspect_ratio_bin = ASPECT_RATIO_512_BIN else: raise ValueError("Invalid sample size") orig_height, orig_width = height, width