This dialog allows you to reconstruct “full tomograms”, i. e. volumes located at the center of the tilt series with the Unbinned tomogram dimensions set previously in the Input settings section. These tomograms can be used for visual inspection and template matching. Important: Unlike other pipelines, where the reconstruction of a giant, unbinned tomogram is required to extract sub-tomograms at unbinned pixel size from it, Warp’s workflow expects you to use the dedicated sub-tomogram reconstruction dialog to obtain sub-tomograms with an unbinned/very small pixel size. Reconstructing a very large tomographic volume in Warp will take very long and has a good chance of crashing.

The reconstructed tomograms will be located in the reconstruction subfolder and the file names will contain the pixel size as a suffix. Tomogram reconstruction will be skipped for items that already have volumes saved at that pixel size. If you would like to replace them, please delete the volumes manually from the reconstruction subfolder.

Pixel size: The target pixel size to which the tilt series data will be scaled prior to reconstruction. It can’t be smaller than the size dictated by the Bin parameter in the Input settings section. The maximum resolution achievable with the data will be twice the pixel size.

Invert contrast: Multiply the input data by -1 to invert the contrast. Following standard conventions, black particles from cryo data will become white if this is enabled.

Normalize input images: The scaled tilt images will be normalized to mean = 0, standard deviation = 1 prior to reconstruction.

Also produce deconvolved version: In addition to the normal reconstruction, a deconvolved version will be made and stored in the reconstruction/deconv subfolder. The visual effect is depicted here. Please also refer to that page for parameter description. The defaults are often fine.

Separate odd/even tilts for denoising: In addition to the normal reconstruction, a pair of reconstructions from all odd and even tilts will be made and stored in the reconstruction/odd and reconstruction/even subfolders. These pairs can be used to train a 3D denoiser. If Also produce deconvolved version is checked, both tomograms will have the same deconvolution filter applied to them.

Keep only fully covered voxels: If enabled, voxels in the reconstruction that are not covered by all of the tilts due to excessive shifts or rotations will be zeroed out. Do not enable this if you intend to perform template matching on the tomograms – it will lead to artifacts. The template matching procedure itself will apply a similar mask to the result.