Pinhole SPECT provides superior trade-off between resolution and detection efficiency as compared to parallel-hole collimator for imaging small field-of-view. Previously we proposed an adjustable multi-pinhole (MPH) collimator for clinical myocardial perfusion imaging (MPI) and small animal imaging (SAI) of rats, based on a clinical SPECT/CT scanner. This study aims to optimize and evaluate this MPH collimator design. For MPI and SAI, the target resolutions were set to 1 cm and 1 mm, the FOVs were 20 cm and 5 cm, and the radii-of-rotation were 25 cm and 5 cm, respectively. The maximum detector-to-detector distance for the scanner was 76 cm. Based on the work from Rentmeester et al., Nillius et al. and the system constraints, we optimized the detection efficiency and calculated the design parameters as follows: (i) MPI: 18 knife-edge pinholes with 2.61 mm aperture size, 13 cm collimator length; (ii) SAI: 4 knife-edge pinholes with aperture size of 0.41 mm, 20.8 cm collimator length. All pinholes were focused to the center of FOV. The proposed collimator was composed of two parts: a base cone with a height of 13 cm for MPI and an auxiliary cone with a height of 7.8 cm that can be combined with the base cone for SAI. The pinholes were placed in a way to utilize the whole detector, minimize projection truncation and keep the multiplexing <25%. Tc-99m point source simulations, using GATE (Geant4 Application for Emission Tomography) were performed for different imaging distances to compare the detection efficiency and resolution between low-energy-high-resolution (LEHR) parallel-hole and MPH collimator. For MPI, the measured FWHMs were 11.70 mm and 9.02 mm for LEHR and MPH respectively, while the detection efficiency of MPH had 17.5 % improvement as compared to LEHR. For SAI, the FWHMs for LEHR and MPH were 4.92 mm and 1.01 mm respectively. The new collimator has improved the imaging performance as compared to our previous design. It provides a significant improvement in resolution and detection efficiency trade-off as compared to current MPI using LEHR with extra capability for SAI.