Abstract: This study aimed to systematically evaluate the applicability of low-quality and unstructured imagery acquired with the smartphone for everyday users. Individual-tree parameters were estimated to input into a low-cost urban tree inventory during dynamic monitoring. Twenty isolated deciduous broadleaf trees with diameter at breast height (DBH) ranging from 8.9 to 58.6 cm were selected on the campus of Northwest A&F University in Yangling, Shaanxi, China, during the winter leaf-off period. Between 71 and 263 images per tree were collected with a smartphone via 360° circumferential photography at distances of 2 to 5 m from the trunk. A graduated rod was placed beside each tree for metric scale recovery. The same image sets were processed using multi-view stereo (MVS), neural radiance fields (NeRF), and 3D Gaussian splatting (3DGS), implemented with COLMAP, Nerfacto, and Splatfacto, respectively. The resulting point clouds were manually segmented in CloudCompare to retain only the target tree, scaled at the known intervals on the graduated rod, and finally corrected for orientation by principal component analysis. Tree height was extracted directly from the vertical extent of each orientation-corrected single-tree point cloud. The DBH was then extracted using quantitative structure models (QSMs) with TreeQSM. A total of 24 parameter combinations were tested, with each combination repeated 5 times to generate 120 QSMs per point cloud. The optimal model was selected according to the minimum mean point-to-model distance. Estimation accuracy was assessed against field measurements using root mean square error (RMSE) and relative root mean square error (RRMSE). Processing time was recorded for MVS dense reconstruction, NeRF and 3DGS training, rather than the structure-from-motion stage. After reconstruction for all 20 trees, the mean final single-tree point counts were 1.055 million for MVS, 0.590 million for NeRF, and 0.024 million for 3DGS. MVS point clouds contained relatively complete trunks, particularly with discontinuities and noise in some smaller branches; NeRF point clouds retained more continuous branches and complete canopies; 3DGS point clouds were relatively sparse and represented fewer fine branches. The average processing times of MVS, NeRF, and 3DGS were 149.3, 127.8, and 20.2 min per tree, respectively. The time required by 3DGS was one-sixth that of NeRF and one-seventh that of MVS. TreeQSM generated QSMs with relatively continuous main-stem cylinders from the point clouds of all trees. Differences occurred mainly in canopy branches: MVS-based QSMs fitted major and some secondary branches, whereas the local discontinuities or abrupt changes in direction; There were no continuous branch connections; Some 3DGS-based QSMs contained locally enlarged cylinder radii or abnormal branch trajectories. In DBH estimation, MVS yielded the lowest errors, with an RMSE of 0.71 cm and an RRMSE of 3.87%. In NeRF and 3DGS, the RMSE values were 2.19 and 4.09 cm, respectively, and the RRMSE values were 12.00% and 22.38%, respectively. Both NeRF and 3DGS underestimated DBH for all 20 trees. In tree-height estimation, the RRMSE values were below 5%. NeRF yielded the lowest errors (RMSE = 0.09 m and RRMSE = 1.47%), followed by 3DGS (0.12 m and 2.10%) and MVS (0.26 m and 4.53%). The differences among tree height estimation were smaller than those in DBH estimation. The findings can provide a technical reference to develop a low-cost and efficient forest observation system with public participation, particularly for fine-scale forest resources.