Images

Use mouse wheel to zoom in/out, click and drag to pan. Press keys [1], [2], ... to switch between individual images.

16 spp

64 spp

256 spp

1024 spp

Color buffers

Structural Dissimilarity

rMSE Heatmap

1: Input

2: NLM

3: NLM-MS

4: RHF

5: RDFC

6: LBF

7: WLR

8: WLR-PF

9: NFOR (ours)

0: Reference

Input

NLM

NLM-MS

RHF

RDFC

LBF

WLR

WLR-PF

NFOR (ours)

Reference

Scene by Jay-Artist, downloaded from blendswap.com.

Charts

Relative improvement, obtained by dividing the filtered error by the input error. For each method, we show, from left to right, the improvement at 16, 64, 256 and 1024 samples. For MSE and relative MSE, lower values are better; for PSNR and SSIM, higher values are better.

MSE

rMSE

PSNR

SSIM

Error Metrics

MSE
SPP	Input	NLM	NLM-MS	RHF	RDFC	LBF	WLR	WLR-PF	NFOR (ours)
16 spp	0.8929	0.1012	0.08592	0.1103	0.08701	0.1415	0.2948	0.2740	0.1602
64 spp	0.2277	0.03480	0.03099	0.03655	0.03091	0.06386	0.06741	0.04997	0.04070
256 spp	0.05636	0.01060	0.009876	0.01735	0.009395	0.02110	0.009833	0.009132	0.009602
1024 spp	0.01416	0.002664	0.002626	0.01356	0.002819	0.009619	0.003479	0.002555	0.002486

rMSE
SPP	Input	NLM	NLM-MS	RHF	RDFC	LBF	WLR	WLR-PF	NFOR (ours)
16 spp	2.887	0.06287	0.05424	0.1733	0.03710	0.05841	0.1305	0.1081	0.07159
64 spp	0.7269	0.02141	0.02158	0.05540	0.01248	0.03469	0.03095	0.01848	0.01063
256 spp	0.1817	0.006464	0.006855	0.01332	0.004185	0.01026	0.005381	0.005033	0.003044
1024 spp	0.04564	0.002220	0.002225	0.003711	0.001602	0.003074	0.002604	0.001853	0.001208

PSNR
SPP	Input	NLM	NLM-MS	RHF	RDFC	LBF	WLR	WLR-PF	NFOR (ours)
16 spp	8.355	24.76	26.64	22.33	26.97	27.59	19.95	20.24	22.44
64 spp	13.39	28.65	29.86	25.96	31.65	29.89	27.40	28.86	31.55
256 spp	19.45	33.47	34.04	30.88	35.70	32.91	34.44	34.55	36.77
1024 spp	25.29	37.94	38.07	35.86	39.51	36.87	38.09	38.91	40.77

SSIM
SPP	Input	NLM	NLM-MS	RHF	RDFC	LBF	WLR	WLR-PF	NFOR (ours)
16 spp	0.04769	0.6917	0.7394	0.8045	0.8531	0.8980	0.7202	0.7489	0.8915
64 spp	0.09695	0.8066	0.8313	0.8623	0.9245	0.9217	0.8323	0.8804	0.9493
256 spp	0.2327	0.9085	0.9176	0.9162	0.9595	0.9445	0.9504	0.9510	0.9735
1024 spp	0.4650	0.9597	0.9621	0.9573	0.9775	0.9660	0.9661	0.9755	0.9840