zylim0702/bokeh_prediction 🔢🖼️ → 🖼️

▶️ 622 runs 📅 Jul 2023 ⚙️ Cog 0.9.4 🔗 GitHub ⚖️ License

bokeh depth-of-field-simulation image-editing

About

Bokeh Prediction, a hybrid bokeh rendering framework that combines a neural renderer with a classical approach. It generates high-resolution, adjustable bokeh effects from a single image and potentially imperfect disparity maps.

Example Output

Output

Performance Metrics

24.28s Prediction Time

24.83s Total Time

All Input Parameters

{
  "K": 20,
  "focus": 0.87,
  "image": "https://replicate.delivery/pbxt/KIwMwjiIssQuFO7dAFO9X8hZOZSNmbpUlrGh0WRFH4oTuvhK/If-Elon-Musk-Had-Been-a-Happy-Child%2C-Would-He-Still-Be-Launching-Rockets--Business-Redux-h_16082330.webp"
}

Input Parameters

K Type: numberDefault: 10: Blur strength
focus Type: numberDefault: 0.8: Focus distance
image (required) Type: string: Image to enlarge
x_coord_focus Type: integerDefault: 0: x-coordinate of the image/coord to focus ( 0: use focus distance)
y_coord_focus Type: integerDefault: 0: y-coordinate of the image/coord to focus ( 0: use focus distance)

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

tmp/bokeh_neural.png

Version Details

Version ID: a03867e7e035d1ab286fd8fe7c7f8461b0908b59f6daace4e97d8f68c3a1d05a
Version Created: February 4, 2024

Run on Replicate →