home
Phantom Desktop is an application for forensic digital image processing.
- Phantom Desktop has useful tools for forensic image processing.
- Image Viewer with support for EXIF metadata and image properties.
- Image Deblur tool using Lucy-Richardson deconvolution.
- Visual Perspective correction tool.
- Face detection and clustering with an intuitive user friendly interface.
- Support for saving and loading projects.
To get started, download the latest release from the releases page. Unzip the file and run it.
The first time you start Phantom Desktop the application will download the AI models required for face detection and clustering. This may take a few minutes.
When you start Phantom Desktop a new project is created. You can add images to the project by using the Add images button in the toolbar to add images from your computer. You can also use the Add folder button to add all the images in a folder and its subfolders.
Once you add images to the project Phantom Desktop will automatically detect faces in each image. The progress of the face detection can be seen in the status bar at the bottom of the screen. Images with faces detected will have a small 
person icon in the top left corner of the image.
Phantom Desktop has a face detection and clasification tool that let you classify thousands of images of different people in a matter of minutes. To begin, add images to the project and wait for the face detection to finish. Once the face detection is finished, click on the Group Similar Faces button in the toolbar. Phantom Desktop will show the results of the face clustering in a new window.
In the face clustering window you can see the results of the face clustering. The images are grouped by similarity. You can click on any group to see the faces in that group.
Info: Results will be automatically saved when you save the project.
- Change group name: Groups can be renamed by double clicking on the picture of the person or by selecting the group and clicking on the Edit name button in group details on the right side of the screen.
- Combining groups: Select one group and click on the Combine group button in the group details on the right side of the screen, then select the group you want to combine with the selected group. All the faces in the selected group will be added to the selected group.
- Remove face from group: Right click on the face you want to remove from the group and select Remove from group.
- Move face to another group: Right click on the face you want to move to another group and select Move to group. Then select the group you want to move the face to or pick New group to create a new group for that face.
- Use as main face: Right click on the face you want to use as the main face for the group and select Use as main face. The main face will be the one that is shown in the group details.
When Phantom Desktop detects that there are similar faces but it is not confident enough to merge them, it will show a face merging wizard to ask you some questions about the faces.
You can answer the questions to improve the results of the face clustering:
- Same person: Select this option if you think the faces are of the same person. The groups will be merged.
- Different people: Select this option if you think the faces are of different people. The groups will not be merged and you will not be asked about these faces again.
- I'm not sure: Select this option if you are not sure if the faces are of the same person or not. Phantom Desktop will ask you again about these faces later.
The perspective correction tool lets you correct the perspective of an image. For example, if you have an image of a document that is not perfectly flat, you can use the perspective correction tool to create a flat image of the document.
Other use cases are correcting the perspective of images taken from a drone, car plate images, paper documents laying on a table, billboards and other images where the perspective is not perfectly flat.
To use the perspective correction tool, select the image you want to correct in the project and click on the Perspective correction button in the toolbar. Phantom Desktop will show the image in a new window.
This window let you adjust the perspective of the image.
- Start by clicking on 4 points in the image to define the corners of the image. For example, in the image above, the corners of the billboard on the left are selected. You can use the zoom preview to help you select the corners. After you have defined the points, you can drag them and reposition them to adjust the perspective.
- Then select the aspect ratio. You can pick one of the predefined aspect ratios or select Custom to define your own aspect ratio.
- Pick the output width and height. The output image will have the aspect ratio you selected and the width and height you picked.
- You can rotate the image by clicking on the Rotate buttons. The Smart rotation button will try to rotate the image to make it more readable automatically.
- Optionally you can select the interpolation method. The default method is Linear.
- That's it! Click on the Export Image button to save the image to your computer or click the Export and add to project button to save the image to your computer and add it to the project.
The deblur tool lets you remove blur from an image. This tool uses Lucy-Richardson deconvolution to remove blur from an image.
To use the deblur tool, select the image you want to deblur in the project and click on the Deblur button in the toolbar. Phantom Desktop will show the image in a new window.
This window let you adjust the deblur parameters.
-
Blur type: The first step is to select the blur type. The available blur types are:
-
Gaussian blur Usually images out of focus or that have artificial blur applied to them.
-
Motion blur Usually images that have been taken while moving the camera.
-
Box blur Usually images that have been artificially blurred with a box blur filter.
-
Disk blur Usually images that are out of focus or that have been artificially blurred with a disk blur filter.
-
Custom Use this option if you know the exact PSF (Point Spread Function) that was used to blur the image in the first place.
The following image illustrates the difference between some of the blur types:
| Type of blur | Parameters | Example PSF kernel |
|---|---|---|
| Gaussian blur | - Sigma: The standard deviation of the Gaussian blur. |  |
| Motion blur | - Angle: The angle of the motion blur. - Length: The length of the motion blur. - Width: The width of the motion blur. |
 |
| Box blur | - Size: The size of the box blur. |  |
| Disk blur | - Radius: The size of the disk blur. |  |
| Custom | - PSF kernel: The PSF kernel to use for the deconvolution. You can choose any grayscale image to use as a PSF kernel. - Scale: The scale of the PSF kernel. |
 |
- Number of iterations: The number of iterations to use in the deconvolution process. The default value is 10 to create a fast preview, but you should use a higher value (100 to 500 are good values) as soon as you have selected the blur and configured the other parameters.
- Progressive preview: Check this option to see a progressive preview of the deconvolution process. This is useful when you are using a high number of iterations to see the progress of the deconvolution process. Uncheck this option if you are getting artifacts in the preview.
Once you are happy with the results, click on the Export Image button to save the image to your computer or click the Export and add to project button to save the image to your computer and add it to the project.