Fig. 1. Overview of methods for joint reconstruction of dynamic and static neural radiation fields

Fig. 2. Human body decomposition diagram. (a) Original image; (b) decomposition results of SAM; (c) decomposition results of Mask-RCNN

Fig. 3. Alignment optimization example. (a) Original image; (b) alignment image by ROMP algorithm; (c) optimized alignment image

Fig. 4. Video data collection scenarios human 1‒6

Fig. 5. Background reconstruction. (a) Video capture scene; (b) static background reconstruction results

Fig. 6. Human body reconstruction results. (a) Da-pose new perspective map; (b) free viewing angle view

Fig. 7. Background and human body joint reconstruction of new pose rendering results. (a) Rollover; (b) walk; (c) roll; (d) balance

Fig. 8. Comparison of qualitative analysis results. (a) NSFF method; (b) Neuman method; (c) proposed method; (d) ground truth

Fig. 9. Ablation experiment. (a) No GC module; (b) no L_SMPL module; (c) full module