Mixed exposures: why a VSF loss beats L1

15 Nov 2026 · Fei · checker toy, 3DGS, 3 seeds per number

Weight each view by its own noise.

The signal grows in proportion to the exposure time t, but the noise does not: read noise stays the same and shot noise grows as √t. After dividing each frame by t, a t = 1/4 view is 7–16× noisier than a t = 4 view. L1 weights all views equally; VSF weights each view by 1/σi.

With 8 views at t = 4 and 8 at t = 1/4: L1 31.1 dB → VSF 33.1 dB (CVVDP 8.81 → 9.34); R1 33.2 dB with one setting. Adding the noisy views makes L1 worse than using the good views alone (32.4 dB).

Loss

Frame i is divided by its exposure ti, yi = rawi / ti, so every view has the same mean but noise

σi²(v) = a·v / ti + b² / ti²

VSF compares render r and observation yi after the variance-stabilising transform of that view:

L = Σi | φi(r) − φi(yi) |, φi(v) = ∫0v du / σi(u), ∂L/∂r = sign(r − yi) / σi(r)

So each view is weighted by its own 1/σi. L1 is the same loss with σi ≡ 1.

Toy

Exposure-ladder checkerboard with Dark3R noise; the 16 views alternate t = 4 and t = 1/4. Thresholds are swept and the best is reported; R1 has a single setting.

inputs
Figure 1. Clean reference and one view of each exposure, divided by t.

Results

methodfitting loss (r = render, rawi = frame i, ti = its exposure)best thresholdsRGB PSNR (3 seeds)CVVDP#Gaussians
L1Σi | r − rawi / ti |τ 1.25e-531.13 ± 0.698.816.8k
VSFΣi | φi(r) − φi(rawi / ti) |τ 0.0133.12 ± 0.139.3410.0k
R1VSF loss; densify by the noise test (no threshold)none33.21 ± 0.369.012.8k

Using only the 8 t = 4 views: L1 32.45 dB.

threshold sweeps
Figure 2. Quality vs number of Gaussians over each threshold sweep; R1 is one point. Seed means, bars min–max. PDF
qualitative
Figure 3. Dark crop, bright crop, |error| ×4 (dark). Median-PSNR seed of each method's best setting. PDF

Why

With an L1-type loss each pixel converges to a weighted median over the views, and the best weights are 1/σi. Monte Carlo of one pixel with this toy's noise, giving the error reduction relative to L1:

pixel valuedrop the noisy viewsVSF
0.0002+3.14 dB+3.27 dB
0.0016 (dark cell)+3.03 dB+3.17 dB
0.0456 (bright cell)+2.02 dB+2.37 dB

Equal weights let the noisy views pull every pixel, and, with one view per training step, they also dominate the update noise and the densification score.