Single-photon LiDAR (SPL) sensors offer exceptional temporal resolution and single-photon sensitivity for depth imaging. However, cost-effective SPL arrays typically suffer from extremely low spatial resolutions (e.g. 8×8 pixels), limiting their real-world utility. In this paper, we propose a novel, one-step super-resolution framework, D2ToF, Distilled Diffusion for Time-of-Flight, that bridges the gap between high-fidelity generative modeling and real-time execution. Our approach uses an amortized variational distillation strategy that allows for a measurement-based data consistency term to be incorporated directly into the one-step training objective. This encourages that the synthesized high-resolution depth maps are consistent with the captured low-resolution Time-of-Flight (ToF) histograms. To overcome the severe ill-posedness of upsampling extremely degraded inputs, we introduce a multimodal guidance strategy. By leveraging a foundational monocular depth model to extract structural priors from a high-resolution reflectivity image, our network is guided by a rich triad of the ToF histogram, reflectivity map, and relative depth estimate. Experimental results outperform evaluated baselines, achieving 16× spatial super-resolution (from 8×8×100 histograms to 128×128 depth maps) at an accelerated inference speed of approximately 83 milliseconds.
Figures & tables
Fig. 1: Reconstruction quality (RMSE in mm) vs. time (sec.) of our proposed method in comparison to DELTAR [ 8 ] , DepthAnything-V2 [ 9 ] , DVSR [ 10 ] , and Marigold-DC [ 11 ] evaluated on 12 clean scenes of Middlebury dataset.
Fig. 2: Real 8×8 SPAD measurement and corresponding reconstruction. Right: (top) high-resolution guidance image with the selected field of view, (middle) spatial locations of the selected 3×3 ToF cells within the measured 8×8 array, (bottom) one-step reconstructed depth. Left: photon-arrival waveforms for the highlighted ToF cells, showing the temporal data available.
Fig. 3: Overview of the conditional prior (Stage 1). The three conditioning signals: low-resolution ToF histogram H , high-resolution reflectivity R , and monocular depth prior dDA enter the U-Net asymmetrically: the encoded histogram Henc is injected on the encoder side, while the encoded reflectivity and monocular-depth features Renc are fused into every decoder (synthesis) stage.
Fig. 4: Overview of the distillation stage. A one-step generator qϕ(x0∣y) is trained so that its samples (i) match the learned prior via teacher-student score matching and (ii) enforce estimated histogram Hest to be consistent with the measured histogram H .
Method
Art
Baby
Books
Bowling
Cloth
Dolls
Lampshade
Laundry
Moebius
Monopoly
Plastic
Reindeer
Avg.
SBR = 2.0, PPP = 450 (Clean Reflectivity)
DA (aligned)
41.7/26.7
10.8/ 11.3
25.3/24.6
57.7/ 18.0
13.9/ 15.1
29.3/ 27.1
39.6/19.0
54.0/ 22.2
30.9/23.3
43.5/17.0
22.5/ 12.1
27.1/ 17.7
33.0/ 19.5
DELTAR
73.4/64.0
28.6/51.6
30.7/51.7
63.2/50.5
20.8/45.5
26.0/52.7
52.6/48.4
53.8/54.4
35.0/56.7
74.5/58.3
24.5/41.6
39.6/49.5
43.6/52.1
Marigold-DC
68.9/55.8
20.8/19.8
30.6/37.5
47.3/35.7
8.4/31.4
20.6/40.4
34.1/34.7
53.2/55.1
30.4/38.4
33.8/26.0
16.0/23.1
31.1/41.1
32.9/36.6
DVSR
74.5/51.4
24.9/29.8
39.9/39.9
52.7/36.4
12.5/22.5
24.9/44.7
63.5/39.0
46.3/41.3
39.7/38.3
46.0/39.2
20.7/28.9
42.3/34.0
40.7/37.1
D2ToF (teacher)
24.6 / 23.3
8.9 / 13.1
11.2 / 19.9
16.6 / 16.2
5.9 / 14.6
12.9 /27.8
20.3 / 14.8
17.3 / 22.4
14.8 / 22.9
14.7 / 12.9
6.0 / 11.1
13.1 / 18.0
13.9 / 18.1
TABLE I: Quantitative comparison on Middlebury scenes ( 8×8→128×128 ) at two noise levels with a spacing of 7.65 mm/bin. Each cell reports RMSE in mm / mean surface normal error in degrees (both ↓ ). The clean configuration uses SBR = 2.0, PPP = 450 with clean reflectivity. The noisy configuration uses SBR = 1.0, PPP = 300 with degraded reflectivity (Poisson γ=100 , Gaussian σr=0.02 ). DA: DepthAnything V2-base aligned to the measured histograms, DELTAR [ 8 ] , Marigold-DC [ 11 ] , and DVSR [ 10 ] . Within each scene, bold marks the best (lowest) value and underline the second best, applied to each metric independently. The shaded rightmost column reports the mean over all twelve scenes.
Fig. 5: (Top) Comparison with different methods when both histograms (SBR=2.0, PPP=450.0) and reflectivity maps are in clean scenario. Our method achieves superior results compared to other reconstruction methods. (Bottom) Qualitative comparison of different methods under noisy conditions. The inputs are degraded by both reflectivity noise (Poisson γ=100 , Gaussian σr=0.02 ) and ToF histogram noise (SBR = 1.0, PPP = 300). Our proposed approach significantly outperforms existing baselines under such noisy scenarios.
Fig. 6: Comparison with different methods for real data collected using an inexpensive STMicroelectronics sensor in 8×8 mode. The reflectivity maps ( 128×128 ) are provided by a Kinect sensor connected adjacent to the SPAD.
Histogram
Reflectivity
Setting
SBR
PPP
γ
σr
Clean
2.0
450
–
–
Noise 1
2.0
450
100
0.02
Noise 2
1.0
300
100
0.02
Noise 3
1.0
300
40
0.05
TABLE II: Operating points used throughout the ablation study.
Fig. 7: Effect of reflectivity map R and DA output dDA when either of them is used as a condition on the network decoder of D2ToF.
Configuration
Clean
Noise 1
Noise 2
Monocular ( dDA only)
33.0
61.2
61.1
with R only ( H+R )
18.19 ± 0.14
25.83 ± 0.08
26.69 ± 0.05
with dDA only ( H+dDA )
12.59 ± 0.06
21.69 ± 0.07
23.11 ± 0.07
Full ( H+R+dDA )
11.71 ± 0.06
20.16 ± 0.05
21.75 ± 0.05
TABLE III: Ablation of the reflectivity ( R ) and DA-output ( dDA ) conditioning. The highlighted row is the full model.
Config.
Clean
Noise 1
Noise 2
Noise 3
w/o LC
12.16 ± 0.13
21.26 ± 0.09
22.42 ± 0.08
25.98 ± 0.08
with LC
11.71 ± 0.06
20.16 ± 0.05
21.75 ± 0.05
25.01 ± 0.06
TABLE IV: Ablation of the consistency loss LC under noise levels defined in Table II .
DELTAR
Marigold DC
DVSR
D2ToF (Teacher)
D2ToF (w/o Con. Loss)
D2ToF
GT
Simulated Middlebury (12-scene mean)
37.04
13.04
25.28
12.91
8.62
6.74
4.75
Real data (4-scene mean)
53.41
18.00
30.02
21.04
19.70
16.12
–
TABLE V: Histogram reconstruction error ( ∥Hest−H∥22 ; lower is better), averaged over 12 clean simulated Middlebury scenes and four real scenes (well-lit scenes from Figure 6 ). GT is unavailable for the real scenes.
Fig. 8: Histogram consistency for one low-resolution pixel from the Bowling (top) and Laundry (bottom) scenes. The gray bars show the measured ToF histogram, while the colored curves show the histograms re-rendered from the depth estimates produced by each method. We generate 50 samples for each method to generate the mean (solid line) and the standard deviation (shaded area). The proposed methods (D2ToF-Teacher and D2ToF) reproduce the measured temporal structure more closely than the compared methods.
Fig. 9: We show the uncertainty estimation (standard deviation) of our teacher and student model for the Art scene from Middlebury in a clean setting. D2ToF (T) denotes our teacher model.