Is it possible for Pix4D Mapper to create a 3D map based on a serie of images (FHD, 30fps, 90°FOV) which are non-georefenced?
If yes, what is the difference of accuracy/robustness compared to a map created with a serie of geo-referenced images (position & orientation) ?
Thanks for your interest in Pix4D software solutions. Pix4Dmapper can process images without geotags. The project will not be properly scaled and will require implementing scale and perhaps orientation constraints for accurate measurements. Pix4D offers a support article, How to scale a project that explains the steps involved. You can also use Pix4D to extract the still images from the video file and can reference How to use Videos for Processing for more information.
I hope this information is helpful,
Beside the fact of the scaling and the orientation, is the project going to be more noisy or won’t be able to 3D construct some part of the model because of non-geotagged images?
2nd question: In term of consistency of the 3D pointcloud, what is the most important information related to the pictures, position or orientation?
A 4K video file has a resolution of either 3840 x 2160 or 4096 x 2160, which equates to an 8.2mp - 8.8mp image. Typical cellphone cameras can capture “still” images at 12+ mp, your models will typically look better when created from a higher resolution still image vs a frame grab from a video file. As the images are coming from a video file, they will also exhibit linear rolling shutter distortion (How to correct for the Rolling Shutter Effect).
Pix4D looks at the images at a pixel level as it finds keypoints based on unique textures in the image. When keypoints match between images, they become tie points and the basis for data in the 3D point cloud. The resulting 3D point cloud will be better when the images are clear, sharp and have a high degree of overlap.
Thanks for your answer.
What are the improvements of a set of geotagged still images compared to non-geotagged identical images?
The Project will have better/more accurate Scale and orientation based on the geotagged images (relative to the accuracy of the Geotaggs). There won’t really be an “improvement” in how the project “looks”,Geotaggs provides the project with information that allows the output to be more accurately scaled/oriented.