DAMMIF is a revised implementation of the ab-initio shape-determination program DAMMIN for small-angle scattering (SAS) data. It has been fully rewritten and optimized for speed, allowing it to run 25–40 times faster than DAMMIN on a single CPU. DAMMIF also supports multiple CPUs and provides options for symmetry and anisotropy constraints. The program uses an unlimited, growing search volume to avoid artifacts caused by limited search space. It is available in binary form for major platforms. The DAMMIN algorithm uses a search volume filled with dummy atoms to model the particle. The scattering intensity is calculated using spherical harmonics, which is computationally more efficient than the standard Debye formula. The algorithm minimizes a function that includes discrepancies between experimental and calculated data, as well as penalties for model features. DAMMIF improves upon DAMMIN by implementing an unlimited search volume, allowing for more accurate reconstructions, especially for anisometric particles. It also introduces new penalties, such as center and Rg penalties, to guide the annealing process. DAMMIF uses OpenMP for parallelization and a prefetch strategy to speed up the algorithm. The quality of reconstruction is assessed using the R factor, which indicates the goodness of fit. DAMMIF's improved speed allows for faster analysis of multiple runs, enhancing the reliability of the results. The program is available for major platforms and is supported by the EU FP6 Design study SAXIER grant. DAMMIF offers significant improvements in speed and accuracy over DAMMIN, making it a valuable tool for ab-initio shape determination in SAS data analysis. It is particularly effective for reconstructing the shapes of particles with high anisometry, as it avoids border effects. The program is available in binary form and is suitable for use in both research and practical applications.DAMMIF is a revised implementation of the ab-initio shape-determination program DAMMIN for small-angle scattering (SAS) data. It has been fully rewritten and optimized for speed, allowing it to run 25–40 times faster than DAMMIN on a single CPU. DAMMIF also supports multiple CPUs and provides options for symmetry and anisotropy constraints. The program uses an unlimited, growing search volume to avoid artifacts caused by limited search space. It is available in binary form for major platforms. The DAMMIN algorithm uses a search volume filled with dummy atoms to model the particle. The scattering intensity is calculated using spherical harmonics, which is computationally more efficient than the standard Debye formula. The algorithm minimizes a function that includes discrepancies between experimental and calculated data, as well as penalties for model features. DAMMIF improves upon DAMMIN by implementing an unlimited search volume, allowing for more accurate reconstructions, especially for anisometric particles. It also introduces new penalties, such as center and Rg penalties, to guide the annealing process. DAMMIF uses OpenMP for parallelization and a prefetch strategy to speed up the algorithm. The quality of reconstruction is assessed using the R factor, which indicates the goodness of fit. DAMMIF's improved speed allows for faster analysis of multiple runs, enhancing the reliability of the results. The program is available for major platforms and is supported by the EU FP6 Design study SAXIER grant. DAMMIF offers significant improvements in speed and accuracy over DAMMIN, making it a valuable tool for ab-initio shape determination in SAS data analysis. It is particularly effective for reconstructing the shapes of particles with high anisometry, as it avoids border effects. The program is available in binary form and is suitable for use in both research and practical applications.