Image downscale

There are more possibilities to change the quality of the resulting model using the image downscale parameters. There is Image downscale factor in Alignment settings, Downscale for depth maps in selected inputs and Image downscale in Reconstruction settings.

Image downscale factor in Alignment settings

Image downscale factor is a multiple by which the application reduces the size of an image before feature detection. In order to get the best precision, use full-resolution features (downscale factor 1). When you use lower number, then you get lower number of founded features on the images.

This factor has the highest influence on the resulted quality of preview model. It is because there was set Use sparse point cloud for Preview model mesh calculation. Models computed in Normal or High quality have similar triangle counts when were used different Image downscale values.

On the next images it is possible to see the count of founded and used features on the test image and created model in Preview quality.

Image downscale: 1 (used 9000 features), preview model created with 46000 tris, normal detail 1.3 M tris, high detail 5.4 M tris

Image downscale: 2 (used 6600 features), preview model created with 30000 tris, normal detail 1.3 M tris, high detail 5.3 M tris

Image downscale: 4 (used 1800 features), preview model created with 6000 tris, normal detail 1.3 M tris, high detail 5.4 M tris

Image downscale: 6 (used 600 features), preview model created with 2500 tris, normal detail 1.4 M tris, high detail 5.4 M tris

Downscale for depth maps in Selected inputs

Downscale for depth maps set the image downscale factor for the depth map computation. The higher the integer, the faster the computation, but it results in a lower detail. Using integer 1 means no change in the image scale (100% resolution), using integer 2 means each side of the image will be twice as small (25% of the original image resolution), etc.

Image downscale in Reconstruction settings

This value relates to the wanted quality of reconstructed model. The Preview value is originally set to 4, Normal detail 2 and High detail 1.

Images scaled down by this factor are used for the depth map computation instead of the original resolution of images. The higher the number, the faster the computation, but it results in a lower detail. The images resolution is changing similar as in previous case.

Please note: The final downscale for depth map computation is calculated as follows.

Downscale for depth maps (defined in the 1Ds Selected input(s) panel for selected pictures) x Image downscale (defined in the Reconstruction settings).

It means that if you change downscale for depth maps for specific image(s) in the Selected input table to e.g. 2, then you use Normal detail reconstruction with the downscale of images set to 2 (defined in the Reconstruction settings), the final downscale is the product of these two numbers, so 2*2 = 4.

Examples:

1. example case

If you use HIGH reconstruction mode so all images are used with downscale of 1x

If you set a particular camera and set downscale 2x so it will use the 2x downscale

2. example case

If you use NORMAL that is 2x downscale

If you set particular camera and set downscale 2x so it will use the 4x downscale

If you set particular camera and set downscale 3x so it will use the 6x downscale

If you set particular camera and set downscale 1x so it will use the 2x downscale

According to these examples, to the test model were applied Downscale parameters in Reconstruction settings. If the Downscale parameter was set to 2, created model in Preview quality had 46 000 tris, in Normal 336 000 tris and in High 1.3M tris. When the Downscale parameter was set to 4, the resulted models had 3 200 tris for Preview, 133 000 tris for Normal and 518 000 tris for High detail.