Spaces:
Running
on
Zero
Running
on
Zero
Add paper abstract
Browse files
app.py
CHANGED
@@ -1678,6 +1678,15 @@ if __name__ == "__main__":
|
|
1678 |
with gr.Blocks(css=css) as demo:
|
1679 |
with gr.Column(elem_id="col-container"):
|
1680 |
gr.Markdown("# AccDiffusion: An Accurate Method for Higher-Resolution Image Generation")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1681 |
gr.HTML("""
|
1682 |
<div style="display:flex;column-gap:4px;">
|
1683 |
<a href='https://lzhxmu.github.io/accdiffusion/accdiffusion.html'>
|
|
|
1678 |
with gr.Blocks(css=css) as demo:
|
1679 |
with gr.Column(elem_id="col-container"):
|
1680 |
gr.Markdown("# AccDiffusion: An Accurate Method for Higher-Resolution Image Generation")
|
1681 |
+
gr.Markdown("### Paper Abstract")
|
1682 |
+
gr.Markdown("""
|
1683 |
+
This paper attempts to address the object repetition issue in patch-wise higher-resolution image generation.
|
1684 |
+
We propose AccDiffusion, an accurate method for patch-wise higher-resolution image generation without training.
|
1685 |
+
An in-depth analysis in this paper reveals an identical text prompt for different patches causes repeated object generation, while no prompt compromises the image details.
|
1686 |
+
Therefore, our AccDiffusion, for the first time, proposes to decouple the vanilla image-content-aware prompt into a set of patch-content-aware prompts, each of which serves as a more precise description of an image patch.
|
1687 |
+
Besides, AccDiffusion also introduces dilated sampling with window interaction for better global consistency in higher-resolution image generation.
|
1688 |
+
Experimental comparison with existing methods demonstrates that our AccDiffusion effectively addresses the issue of repeated object generation and leads to better performance in higher-resolution image generation.
|
1689 |
+
""")
|
1690 |
gr.HTML("""
|
1691 |
<div style="display:flex;column-gap:4px;">
|
1692 |
<a href='https://lzhxmu.github.io/accdiffusion/accdiffusion.html'>
|