Skip to content

[qwenimage] add image_area kwarg to QwenImageEditPlusPipeline #13660

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
david-PHR wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[qwenimage] add image_area kwarg to QwenImageEditPlusPipeline #13660

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 9 additions & 3 deletions src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_plus.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -64,7 +64,6 @@
"""

CONDITION_IMAGE_SIZE = 384 * 384
VAE_IMAGE_SIZE = 1024 * 1024


# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift
Expand Down Expand Up @@ -534,6 +533,7 @@ def __call__(
true_cfg_scale: float = 4.0,
height: int | None = None,
width: int | None = None,
image_area: int = 1024 * 1024,
num_inference_steps: int = 50,
sigmas: list[float] | None = None,
guidance_scale: float | None = None,
Expand Down Expand Up @@ -579,6 +579,12 @@ def __call__(
The height in pixels of the generated image. This is set to 1024 by default for the best results.
width (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
The width in pixels of the generated image. This is set to 1024 by default for the best results.
image_area (`int`, *optional*, defaults to `1024 * 1024`):
Target pixel area used to derive (a) the default output `height`/`width` from the input image's
aspect ratio when those are not explicitly provided, and (b) the resolution at which the input
image(s) are encoded by the VAE. When `height` and `width` are both passed explicitly they
override the default-derivation, but `image_area` still controls the VAE-encoding size of the
input image(s).
num_inference_steps (`int`, *optional*, defaults to 50):
The number of denoising steps. More denoising steps usually lead to a higher quality image at the
expense of slower inference.
Expand Down Expand Up @@ -640,7 +646,7 @@ def __call__(
returning a tuple, the first element is a list with the generated images.
"""
image_size = image[-1].size if isinstance(image, list) else image.size
calculated_width, calculated_height = calculate_dimensions(1024 * 1024, image_size[0] / image_size[1])
calculated_width, calculated_height = calculate_dimensions(image_area, image_size[0] / image_size[1])
height = height or calculated_height
width = width or calculated_width

Expand Down Expand Up @@ -696,7 +702,7 @@ def __call__(
condition_width, condition_height = calculate_dimensions(
CONDITION_IMAGE_SIZE, image_width / image_height
)
vae_width, vae_height = calculate_dimensions(VAE_IMAGE_SIZE, image_width / image_height)
vae_width, vae_height = calculate_dimensions(image_area, image_width / image_height)
condition_image_sizes.append((condition_width, condition_height))
vae_image_sizes.append((vae_width, vae_height))
condition_images.append(self.image_processor.resize(img, condition_height, condition_width))
Expand Down
Loading