Photo Registration: HD texturing tips and tricks

1. Place an object in Position A.
2. Make Scan 1
3. Without moving the object take a set of pictures – Photos 1
4. Change the position of the object if needed (e.g. flip it over) – Position B
5. Scan the object in the Position – Scan 2
6. Take pictures of the object in Position B – Photos 2
7. If required, convert photos into an acceptable format (. bmp/ . png/ . jpg/ .tif)
8. Import Photos 1 and Photos 2 as separate folders to Artec Studio 16

9. Align Scan 1 to Scan 2, apply Global Registration to them, and create a Fusion
10. Select registered Scan 1, Fusion, and Photos 1, apply Photo Registration
11. Select registered Scan 2, Fusion, and Photos 2, apply Photo Registration

12. Apply texture to the Fusion using Photos 1 and Photos 2 as a texture source

For shiny and reflective objects the following steps shall be added to the workflow:

- Put an object on/near a texture-rich background

- Capture the photos of a clean object – Clean Photo set

- Scan a clean object – Clean Scan

- Dust the object

- Scan the dusted object again (follows the same requirements above) – Dust Scan

(!) Avoid adding additional dust during scanning!

- Use “Global Registration” by Geometry + Texture while keeping the background for both scans

- Apply Photo Registration for Clean Scan

- Apply Texture to the Fusion using Clean Photos as texture source.

Photo Camera (SLR camera or a smartphone with a good camera. For example, Xiaomi Redmi Note 9/10 or Samsung Note 10 and above (more recent), iPhones

Diffuse Lighting

- A photo box or two lamps is an ideal setup. 

The object should ultimately be lit from all sides. Please avoid a direct light source from one side of the object.

Background:

- If the object has a great variety of texture features, no additional texture features on the background are required

- If the object does not have sufficient texture features, or for small figurines, shiny objects, and objects with reflective surfaces, please add texture features to the background

- when scanning human faces, we recommend adding a textured background or using some textured clothes close to the face

Optional:

- Turntable for photo box

- Tripod

- Before taking photos, make sure you have access to the object from all sides

- Leave the object stationary while scanning and taking photos. Scan and take photos of the object in the same position

(!) If your object is not in a lightbox, do not use a turntable when scanning and taking photos. Similar texture features photographed in different positions in relation to the object and matched during photo registration are likely to cause artifacts on the object’s texture. Make sure you’ve got enough room around your scene and then walk around with the scanner and the camera, instead of rotating the object.

The walls and floor inside the lightbox are monochrome white, as a rule, so it is ok to use a turntable inside a lightbox. If you need to add texture features to the scene because the object itself lacks them, add them to the turntable.

- Keep the object rigid

- Find a proper distance to capture photos from. The object should not be neither too far away nor too close

- Adjust and fix the focal length. 

(!) All photos must have the same focus (“Focal length”)  in order to be registered together. Changing focus or using auto-focus will prevent the photos from being able to register

If there are several groups of photos made with different focus settings, it’s possible to sort them in folders, import as several objects and register them separately.

- Make sure that images are bright enough, adjust white balance.

- Avoid using photo flash as it creates extra glare.

- Use only one orientation of the camera during capturing (i.e. do not rotate the camera)

- Make subsequent photos with 66% or 2/3 overlap while moving the camera.

The idea is to take a series of photos that can be matched to each other. The rule of thumb is that any given point on the object needs to be in at least 3 photos. It’s better to take at least two times more photos than the number of T frames taken by default (with the “Frequency of texture frames” set to “Automatic” in the “Scan” tab).

Photos are registered to both T frames and each other, so more photos can significantly help registration.

• You can perform color corrections or white balance corrections, but it is better to use the same correction for the whole project.
• Strictly avoid:
• any geometrical transformations, including resizing
• distortion corrections
• color space transformation (SRGB is our target colorspace).
• If needed, convert images to JPEG. Better to preserve EXIF information.

- Image pre-processing: the option applies contrast to the images from the scanner. This is needed for Artec Eva and Artec Leo, and rarely for Artec Spider (if the object was captured from a large distance and the texture is dark)

- Describer preset defines the number of texture features on the object, a higher preset corresponds to the higher number of features, the finer the texture patterns are the higher the preset needs to be.

For example, if the user applies photo texture to a knitted object or a sofa with small scale upholstery, the "High" setting is recommended.

- Camera field of view usually can be taken from the metadata of the photos. In this case, the optimal setting in the AS will be applied automatically.

Photos of certain cameras do not contain metadata, so in this case, setting the correct field of view will greatly improve photo registration quality.

The field of view for a certain camera can be found either in the camera manual or on the manufacturer website.

Make sure to change the field of view setting only if metadata cannot be viewed when opening photos with special image-viewing software such as Honeyview

Once Photo Registration is done, you can check the quality of the registration and detect the areas with registration problems by looking at the camera positions.

If the photo viewer shows that many more pictures were taken from the same spot than were actually taken, the registration might have gone wrong. Similarly, if there seem to be uncovered areas despite the fact that photos of these areas were taken, the registration did not work correctly.