Create video face swaps for social clips, creative edits, and project previews. Choose cloud processing or keep the work on your PC with Offline FaceSwap Master.
To replace a face in a video, pair the clip with a clear portrait of the replacement face. GiliSoft Face Swap provides a Windows workflow for choosing the target person and processing the replacement. Start with a short clip, check the moving result, then use the same approach for the complete video.
Add the replacement portrait in SOURCE and the video to change in TARGET. Review PREVIEW and the face-selection settings, choose an OUTPUT PATH, and process a short sample. Check the saved video for stable facial detail and motion before running the complete clip. Choose cloud processing or the local workflow in Offline FaceSwap Master according to your project.
The two inputs have different jobs. The source portrait (SOURCE) supplies the replacement identity; the target video (TARGET) contains the person, setting, expressions, and motion you want to work with.
| Input | What to choose | What to avoid for the first test |
|---|---|---|
| Replacement face portrait (SOURCE) | A sharp JPEG, PNG, or BMP photo with visible facial features and a reasonably similar angle | Blur, heavy filters, strong shadows, or hair covering the eyes and mouth |
| Target video (TARGET) | An MP4 or AVI clip where the intended person's face is large enough to inspect | Tiny distant faces, constant obstruction, or a clip made up only of fast movement |
| Test segment | A short section containing a clear view plus representative speech or a head turn | Judging the entire project from one easy, motionless frame |
| Output copy | A separate saved result you can compare with the original | Overwriting the only copy of the source footage |
Use SOURCE for the replacement portrait and TARGET for the video you want to change. The screenshot shows the workspace before media is loaded, with preview and face-selection controls on the right.
In SOURCE, use Click to Upload or drag a portrait into Drop File Here. This is the face you want to insert into the video. Choose a clear JPEG, PNG, or BMP photo with visible eyes, nose, and mouth, and an angle reasonably close to the person in the target clip.
In TARGET, upload or drop the video containing the person whose face you want to change. Start with an MP4 or AVI clip with a clearly visible face. For a long project, prepare a short sample in a video editor before importing it so you can test the setup before the full run.
Inspect PREVIEW after loading the inputs. Review FACE SELECTOR MODE and the REFERENCE FACE area to check the intended person in the target video. The supplied screenshot shows reference mode. For a group scene, check the result where people cross or overlap rather than judging only one clear frame.
Use OUTPUT PATH to choose a writable destination on your computer. Keep the output separate from the original footage and source portrait so the files are easy to compare. The Documents path in the screenshot is an example, not a folder name you need to copy.
Review PREVIEW MODE and PREVIEW RESOLUTION when checking the preview. The screenshot's 1024x1024 value is a preview setting, not a guarantee of the final video's dimensions. Run the sample using the processing controls in your installed workflow and wait for the completed result.
Open the saved result from your output folder and watch it at normal speed. Check head turns, speech, face boundaries, duration, and audio synchronization. Once the sample is satisfactory, apply the chosen inputs to the full clip, save a separate result, and review the finished video throughout.
Use the online processing workflow when uploading the selected clip suits the project. Cloud jobs use a usage-based model; they are separate from the offline license.
Use Offline FaceSwap Master when you want to process the video on a Windows PC and keep source media and results local. Processing time depends on the work and your computer.
Both options start from the GiliSoft Face Swap trial download. For a fuller comparison, see choosing cloud or offline face swapping. The offline purchase page lists the local licenses; buying one is not an unlimited cloud-processing allowance.
Play the result at normal speed first. Then pause around difficult moments to see whether a problem lasts across several frames or appears only during a particular movement.
Check the eyes, nose, mouth, and face proportions. The result should remain recognizable through changes in expression rather than looking correct only when the person faces the camera.
Inspect the cheeks, jawline, hairline, and visible skin tone. Look for an abrupt edge or a face that appears to shift independently of the head.
Review speech, turns, hands crossing the face, and the start of each new shot. Check that the replacement remains on the intended person when other people appear.
Open the saved result outside the processing preview. Check duration, sound, synchronization, picture size, and playback in the application where you plan to use the clip.
| What you see | First adjustment |
|---|---|
| The replacement does not resemble the portrait | Try a clearer reference with fewer filters and an angle closer to the target face. |
| The face looks good when still but flickers during motion | Retest the exact head turn or blurred section with a stronger reference or cleaner target footage. |
| The wrong person changes in a group scene | Recheck the intended face selection and review the moment people overlap or cross. |
| The processing time is too high for repeated tests | Use a shorter prepared sample before processing the complete file. Avoid changing several inputs at once. |
| The saved file behaves differently from the preview | Check the completed output, its available export settings, and playback in another application before rerunning the whole video. |
For detailed diagnosis, continue with fixing flicker and face drift in a video face swap or improving an unnatural-looking face swap.
Keep the original footage and edited export. Use portraits and video you have permission to edit, and make the face-swap context clear when viewers could otherwise mistake the clip for an unedited recording.
Yes. Add the replacement portrait in SOURCE and the clip containing the person to change in TARGET. Review the preview and face selection, choose an output path, and process a short sample. The portrait supplies the replacement face; the video supplies the scene and movement.
The GiliSoft product page lists JPEG, PNG, and BMP photo sources and MP4 and AVI video sources. Use a clear portrait and a target clip that opens correctly in the selected workflow. Check the installed mode's available output choices before preparing a large batch.
Review FACE SELECTOR MODE and the REFERENCE FACE area for the intended person in the target video. The supplied interface shows reference mode. Check moments where people cross, overlap, or leave and re-enter the shot. A short single-person test is a useful starting point before attempting a more complex group scene.
Offline FaceSwap Master supports local photo and video processing on a Windows PC without uploading the source media for cloud processing. Choose the local workflow when you want the footage and results to stay on your computer. The offline license and usage-based cloud processing are separate options.
A free trial download is available. Offline FaceSwap Master is a paid local-processing product, while cloud jobs use a usage-based model. Check the activation and export information in the selected trial mode for its current limits; a trial download does not mean unlimited free production or cloud usage.
There is no useful single processing time for every clip. Duration, resolution, scene complexity, processing mode, and local hardware affect the job. Run a representative sample first to judge both the result and the processing time for your own project.
Motion blur, rapid head turns, partly covered facial features, and a weak reference portrait can make a video replacement unstable. Compare the frames around the exact failure, try a clearer or better-matched portrait, and retest that section. A higher export resolution alone is not a substitute for clearer inputs.
Use the successful sample to choose your reference portrait and target footage before starting the full clip. Check the selected mode's import and export limits, allow sufficient processing time and storage, and review the completed video throughout rather than assuming every scene will match the sample.
Choose a reference portrait, try a short video, and compare the saved result with GiliSoft Face Swap on Windows.
