What do all of these settings do?
Skanect is a powerful 3D scanning software, but often users see the long list of settings and tools available and they don't know where to start.
In order to truly make the most out of your Skanect experience, it is important to learn what these settings do and what context you would want to apply them. This article aims to easily and clearly break down these settings so you can reference it time and again.
- Bounding Box
- Aspect Ratio
- Config File
- Recording Feedback
- Offline Recording
- Force QVGA
- Track Loss Detection
- Uplink Mode
- Uplink Color Gain
- Log File
- Reset Settings
- Reconstruct (Fusion)
- External Edit
- Fill Holes
- Move & Crop
- Remove Parts
- Remove Colors
- Inpaint Colorless
- Remove Colorless
- Export Model
- Web (Sketchfab)
- 3D Print (Shapeways)
The Scene setting selects the size of the bounding box for the scan. These are more bounding box presets and do not have any effect in how the sensor or software acts. You can see the size of each bounding box in the "Bounding Box" section. Room will position the sensor at the center of the bounding box for a 360-degree option; half-room will position the sensor at one end of the bounding box, allowing for better resolution. A larger bounding box will result in lower resolution.
This gives you finer control over the size of the bounding box. It increases proportionally in X, Y, and Z directions, meaning the box will remain a cube. To alter these dimensions in other ways, you will need to either alter the aspect ratio or input a custom config file.
The Aspect Ratio refers to the bounding box, and gives you two options. "Normal" keeps the bounding box cuboidal; "Height x 2" is meant more for scanning people, and creates a bounding box that is twice the height as the width or length.
Designates the directory into which you would like to save your scans. Scans saved here will be saved under the .skn file format, which can be used to import back into Skanect for further processing and reconstruction.
If the object you are scanning requires a more precise bounding box, you can load a custom config file into Skanect that will allow you to tweak some finer settings, such as bounding box size and camera location in relation to the bounding box. For more information on how to create custom config files, please check out these articles:
If you have a previous .skn file that you would like to load back into Skanect for further processing, you can do that by clicking the box next to "Path" and navigating to the file.
None should be selected only on very low-power computers and should be used in conjunction with "All Frames" in Offline Recording. Selecting none will also disable Feedback Quality (as there is no feedback), so the following setting will have no effect.
CPU will be used if your machine is capable of processing the feedback, but you do not have a CUDA-supported graphics card.
GPU will be struck through unless GPU is available. We recommend using GPU so long as the GPU is faster than the CPU. In certain cases, running Skanect through your CPU may be faster than the available GPU, depending on the specs of the GPU.
Ideally, the user should choose the highest feedback quality setting that their machine can handle (without dropping frames). If your framerate is low (<10 fps), we recommend switching to either medium or low.
High should only really be used with a high-quality processing unit.
Selecting All Frames causes all depth and color frames to be saved to the machine. This is best, especially in scenarios where you want to reconstruct your scan. It is even possible to take the scan on a less-powerful computer with All Frames selected, move over to a more powerful computer, and reconstruct the scan for a higher-quality version.
If you have a slow hard drive, or if you have storage limitations, Key Frames would be a better option to use. Key Frames will only save the frames that are used for colorizing the scan. Meshes cannot be reconstructed if Key Frames is selected.
This is mostly used for those who wish to get a higher frame rate at a reduced quality. We find that most people would need this when they are using dedicated USB3.0 with a Primesense Carmine. In general we recommend that most users set Force QVGA = Off.
Depending on your computer's processor, you may wish to keep Force QVGA on. This will ensure the highest possible framerate and allow for the highest quality scans. If you have feedback quality set to high and you are running through the GPU, it might be wise to set Force QVGA to "off". Additionally, if you have a Primesense Carmine, you are required to use Force QVGA.
The software will additionally switch to QVGA automatically if that is the optimized setting for your configuration.
Track Loss Detection
This feature is essential for informing the software whether you are moving around your object, or if you are scanning using a turntable.
If you are moving around your object, you want to keep Track Loss Detection turned on. This will help alert you to when the sensor has lost tracking (i.e. its position in space in relation to your object).
If you are using a turntable, turn Track Loss Detection off.
In most cases, you should set Track Loss Detection to on to ensure the software can aid you if you are losing tracking.
This determines what sort of information will be transmitted during an Uplink session. If your sensor is not calibrated to your iPad, you will be forced to switch your Uplink mode to Depth, which will only stream the sensor's monochromatic depth information to your computer. Depth is also a good choice if you have limited internet bandwidth and do not care about the color.
Depth & Color can be streamed if the sensor is properly calibrated. This is a lower resolution color and is a good middle ground if you have limited internet bandwidth.
Depth & High-Resolution Color will produce the best colors and textures for your scan, but the resulting files will be larger and the session will require higher bandwidth.
Uplink Color Gain
Gain is the sensor's sensitivity to present lighting conditions in the scene. If you have even lighting, select "Locked on Record". If you have uneven lighting, select "Automatic".
An indispensable tool for troubleshooting. Generally speaking, if you have an issue and must contact support, please also include your log file.
Puts all settings back to their default values.
Record - Delay and Limit
If you are using a turntable or must set up your scan without the luxury of being able to hit "Record", it can be useful to delay the amount of time between the "Record" button being pressed and the scan beginning. In this case, change the length of time by dragging the red bar to the left or right. This is not used in Uplink sessions, when Record is tapped on the iPad.
In situations where you have more automatic set ups (say a turntable that moves at a speed of 1 rpm), you may want to automatically stop recording at a set amount of time. Limit will allow you to set how long your scan will last, from 0 to 3 minutes.
Reconstruct - Fusion
Using the data collected throughout the scan, Skanect will attempt to normalize and recreate the scan. If your scan was not negatively affected by random artifacts or tracking loss, it would be better to skip this step. Below is a good scan, along with reconstructions on various levels of fidelity:
|original scan||low fidelity reconstruction||medium fidelity reconstruction||very high fidelity reconstruction|
Note that these scans were all processed on the same machine as the original scan.
If you have set Offline Recording to All Frames and Feedback Quality to high, you can take the scan from a less powerful machine and load it into a more powerful machine.
Further, there are additional controls available on the right side:
The top four controls adjust the position of the bounding box around the object.
The third row controls the distance of the bounding box (left for further away, right for closer).
The fourth row controls the bounding box size (left for smaller, right for larger).
Resets the mesh back to its original state.
Algorithmically attempts to create a watertight mesh--that is a mesh with a single, closed surface. This is important in 3D printing. Four options of smoothing exist, from very low to high. The higher the smoother, the better chance the software will have of creating a 100% watertight mesh, but the more detail will be lost.
There are a few caveats here: the file created here must remain in the same folder and have the same name before and after editing. If you edit and change the name, you will be unable to load the file back into Skanect.
Please follow this workflow:
- Click "View mesh in explorer". A new folder will be created in the directory in which you have been working called "external". Your mesh will be named "edit_me.ply".
- Import this mesh into your favorite post-processor.
- Edit your mesh as you normally would.
- When you are done, save the mesh as the same "edit_me.ply".
- Return to Skanect and click "Reload edited mesh from disk". Provide that you haven't changed the name of your file, Skanect will automatically load the edited mesh back into the application.
Reduces the number of faces of the scan (and also the size of the file). Click and drag the bar to the desired number of faces.
Skanect will automagically comb through the mesh, looking for places where holes exist and attempt to fill them with the best fit.
There are three strategies available here: Watertight, Opened Hull, and Closed Hull.
Watertight and Closed Hull are best for objects, where you are trying to fills as many holes in the mesh as possible to prepare it for 3D printing. Watertight and Closed Hull will close bigger holes.
Opened Hull is a good option for creating room scans. It will still attempt to fill holes, but larger holes may not be filled.
Smoothing the mesh will reduce the number of triangles in the final mesh. The higher the smoothing, the more reduction of triangles. Low or Medium are recommended.
Limit specifies the maximum size hole that the software will attempt to fill.
Move & Crop
Rotates and translates the mesh.
For example if there are parts of your mesh you wish to remove, you may wish to use the "Crop to ground" function. If the only thing you need to remove is the ground plane, you only need to translate the mesh downward (the Ty option). To lower the mesh, drag the slider to the right or click the + or -0.1 for finer control. Once the unwanted portion of the mesh is below the bounding box, click "Crop to ground". This will remove that portion of the mesh.
To remove other parts of the mesh, simply rotate the mesh until the unwanted section is nearest the ground plane, lower the mesh, and crop.
Engraving can be specified to have some text carved into where the ground plane cropped the scan.
Skanect analyzes the mesh and removes any parts that are smaller than a percentage of the biggest model part. The smaller the percentage, the smaller this filter becomes, removing smaller parts. This is useful when removing random or floating artifacts.
If using a Structure Core (Color) or Structure Sensor (Mark II) via Uplink, Skanect will overlay the color captured from the color camera on top of the monochromatic depth mesh.
You have a few options in Colorize--resolution, which sharpens the color, which can look great in some cases and poor in others, Inpaint Colorless (see below), Highest Quality (turn on if file size is not an issue and turn off if file size is must remain smaller), and Prioritize First Frame, where the very first color keyframe captured is the progeniture of all other color frames. Prioritize First Frame also takes a full picture of the first and will not change it.
If the mesh is colorized, running Remove Colors will strip the mesh back to its original monochromatic form.
Any polygons that did not receive a color frame can be algorithmically colored by running Inpaint Colorless. The artificial colors that are generated are created by propagating neighboring colors.
Any polygons that did not receive a color frame will be removed.
Saves the scan locally as a .skn folder, which can then be reloaded into Skanect later for further use and processing.
Exports the model as a number of different formats (.ply, .obj, .stl, and .vrml), applies different coloring models, allows the user to choose the number of exported faces, the scale associated with the file, and the color space.
Allows you to quickly upload your scan directly to your Sketchfab account, directly from Skanect. Title your model and set whether you wish for your model to be listed on public galleries or to remain private.
Get your API Key here: https://sketchfab.com/settings/password
3D Print (Shapeways)
Allows you to upload your model directly to Shapeways for sale. Set the model's name, scaling, price, and price type.