Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix dimension errors when inpainting model is used with hires-fix (#2440)

Browse Source
This commit is contained in:
Kevin Turner 2023-01-31 11:04:02 -08:00 committed by GitHub
commit bde5874707
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 9 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
ldm/invoke/generator/txt2img2img.py
View File

@ -3,10 +3,10 @@ ldm.invoke.generator.txt2img inherits from ldm.invoke.generator
''' '''
import math import math
from diffusers.utils.logging import get_verbosity, set_verbosity, set_verbosity_error
from typing import Callable, Optional from typing import Callable, Optional
import torch import torch
from diffusers.utils.logging import get_verbosity, set_verbosity, set_verbosity_error
from ldm.invoke.generator.base import Generator from ldm.invoke.generator.base import Generator
from ldm.invoke.generator.diffusers_pipeline import trim_to_multiple_of, StableDiffusionGeneratorPipeline, \ from ldm.invoke.generator.diffusers_pipeline import trim_to_multiple_of, StableDiffusionGeneratorPipeline, \
@ -129,17 +129,12 @@ class Txt2Img2Img(Generator):
scaled_height = height scaled_height = height
device = self.model.device device = self.model.device
channels = self.latent_channels
if channels == 9:
channels = 4 # we don't really want noise for all the mask channels
shape = (1, channels,
scaled_height // self.downsampling_factor, scaled_width // self.downsampling_factor)
if self.use_mps_noise or device.type == 'mps': if self.use_mps_noise or device.type == 'mps':
return torch.randn([1, return torch.randn(shape, dtype=self.torch_dtype(), device='cpu').to(device)
self.latent_channels,
scaled_height // self.downsampling_factor,
scaled_width // self.downsampling_factor],
dtype=self.torch_dtype(),
device='cpu').to(device)
else: else:
return torch.randn([1, return torch.randn(shape, dtype=self.torch_dtype(), device=device)
self.latent_channels,
scaled_height // self.downsampling_factor,
scaled_width // self.downsampling_factor],
dtype=self.torch_dtype(),
device=device)