Open Model vs Photoroom#
9 examples. Quantitative metrics (MGE, MAE, Connectivity) require ground-truth alpha mattes. Use the view controls on each row to switch between transparency, chroma key, and alpha matte on all outputs. Metric scores are shown as ? until ground truth is added.
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Alexander Dummer on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Andreas Rasmussen on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Camila Quintero Franco on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Conor Samuel on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Ilja Tulit on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Ivan Dostal on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Jonny Clow on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Niko Tsviliov on Unsplash
Not Available


- MGE
- ?
- MAE
- ?
- Conn.
- ?
- MGE
- ?
- MAE
- ?
- Conn.
- ?
Photo by Vinicius Wiesehofer on Unsplash
Methodology
- Alpha mattes are grayscale masks encoding per-pixel foreground probability. They capture fine edge detail (hair, fur, semi-transparent regions) that binary masks discard.
- Metrics are computed by comparing predicted mattes against a ground-truth reference: MGE measures edge sharpness, MAE measures pixel-level accuracy, and Connectivity penalises fragmented foreground regions. All three are lower-is-better.
- Scores are shown as ? when no ground-truth matte is available for that image. See the alpha matting evaluation benchmark for details on the evaluation methodology.
Photo Credits
Source images are courtesy of the photographers below via Unsplash.