inswapper-512-live
This project contains information about the inswapper-512-live model and the application releases based on it.
Update Logs
2025-11-18 Our Real-time Live Face Swap macOS App released based on inswapper-512-live models.
2025-08-18 Our On-device Live Face Swap iOS App released based on inswapper-512-live models.
Applications
1. Picsi.Ai Live Face Swap macOS App
The new Picsi.Ai macOS app brings our most advanced real-time technologies straight to your desk—opening the door to instant, on-device face swapping through your webcam, paired with the complete Pro Studio built natively for macOS.
To download the App, visit https://www.picsi.ai/app
To check the user guide, visit https://www.picsi.ai/macos-docs
System requirements: Apple Silicon Macs(M series) and macOS >= 15.0.
2. Picsi.Ai Live Face Swap iOS App
Based on the inswapper_512_live model, this iOS app provides on-device, real-time face swapping functionalities, including live camera face swapping and video face swapping with a real-time preview.
To download the App, visit https://www.picsi.ai/app
To check the user guide, visit https://www.picsi.ai/ios-docs
Model Introduction
inswapper-512-live is an ultra-lightweight face swapping model that provides native 512x512 resolution. Compared to inswapper_128, it requires less than 1/10 of the computational resources yet produces outputs with 16 times the pixel count(512x512 vs 128x128).
Evaluation
The numbers in the table may fluctuate as we continue to gather more diverse test data.
| Model | Output | Params | GFLOPs | Similarity | Realism | Attributes |
|---|
| INSwapper_128 | RGB 128x128 | 138.3M | 174.7G | 86.9 | 63.3 | 78.8 |
| INSwapper_Cyn | - | - | - | 92.8 | 75.9 | 80.6 |
| INSwapper_Dax | - | - | - | 92.5 | 86.0 | 81.5 |
| INSwapper_Dax(w Optimizer) | - | - | - | 93.4 | 90.2 | 87.7 |
| INSwapper_512_live_base | RGB 512x512 | 9.5M | 12.1G | 87.0 | 73.7 | 80.1 |
| INSwapper_512_live_mini | RGB 512x512 | 4.7M | 6.3G | | | |
Params: Number of parameters in the network, with lower values indicating smaller model file size.
GFLOPs: Computational complexity measured in FP32(half if using FP16), with lower values indicating faster inference.
Evaluation Metric Explanation:
-
Similarity: The similarity between the generated face and the source face is evaluated using a third-party, high-precision face recognition model. The similarity score is calculated using the cosine similarity between the feature embeddings of the source and output images, scaled to a range of 0-100, where higher scores indicate greater similarity;
-
Realism: This metric evaluates the realism and clarity of the generated face. We employ a comprehensive evaluation framework that combines the FID score, custom-designed metrics for facial clarity, and human perceptual evaluation. The score ranges from 0 to 100, with higher scores indicating a more realistic and clear result. We used a test set consisting of hundreds of high-resolution facial images, and randomly resized between 160x160 and 640x640 pixels in testing time;
-
Attributes: This metric assesses the correlation between the attributes of the generated face and the target face, including factors such as pose, gaze direction, expression, etc. We employ third-party facial attribute models to extract feature representations of both the generated and target images. The similarities between these feature vectors are then computed to obtain the final metric score, which ranges from 0 to 100, with higher scores indicating better attribute alignment.
License
All resources in this project, including models, libraries and applications, are strictly for academic use or personal testing and cannot be used for any commercial purposes.
Contact
contact#picsi.ai or contact#insightface.ai
/inswapper-512-live
