TransNormal-2: Geometry-Grounded Rectified Flow with Edge-Aware Decoding for Precise Normal Estimation

Mingwei Li; Yi Yang; Hehe Fan

Fig. 1 : In-the-wild qualitative results on surface normal estimation. TransNormal-2 (ours) recovers accurate surface normals on transparent objects. Scene 1 : prism under extreme lighting. Scene 2 : double-walled glass. Scene 3 : balls viewed through a glass plate. (§ V-B )

Fig. 2 : VAE reconstruction degradation. (a) Mean angular error (MAE, ∘ ) from encoding and decoding ground-truth normals with the frozen VAE; the Ratio column reports Edge/Global MAE. (b) The encode

\to

decode reconstruction and its full-frame angular error map (bright = high error) show that errors concentrate at geometric boundaries, consistent with boundary detail being blurred by the

8{\times}

latent bottleneck. (§ III-A )

Fig. 3 : Overview of the TransNormal-2 framework. (a) Encoding : the frozen FLUX.2 VAE encoder

\mathcal{E}_{\text{vae}}

extracts and patchifies the RGB latent

\bm{z}_{\text{rgb}}

into image tokens

\bm{h}_{\text{img}}

; (b) Single-Step Direct Prediction : the fine-tuned FLUX.2[klein] DiT

f_{{\bm{\theta}}}

directly predicts the normal latent

\hat{\bm{z}}_{\text{n}}

from

\bm{z}_{\text{rgb}}

at a fixed timestep

t

under empty-prompt conditioning, processing the tokens through the LoRA-adapted double-stream and single-stream blocks; (c) Decoding & Refinement : the DiT output passes through a Local Continuity Module (LCM) in latent space, then the frozen FLUX.2 VAE decoder

\mathcal{D}_{\text{vae}}

reconstructs the coarse geometry map, which is further refined by the Geometric Refinement Module (GRM) using the input RGB’s edge information. Training combines latent normal MSE (

\mathcal{L}_{\text{normal}}

), wavelet edge-aware regularization (

\mathcal{L}_{\text{wavelet}}

), von Mises-Fisher angular loss (

\mathcal{L}_{\text{vmf}}

), and inverse rendering self-consistency loss (

\mathcal{L}_{\text{render}}

). (§ IV )

Fig. 4 : Geometry-aware pixel-space losses. (Left) The vMF loss acts on decoded unit normals on

\mathbb{S}^{2}

: maximizing

\hat{\bm{n}}^{\top}\bm{n}^{*}=\cos\theta

shrinks the geodesic angle

\theta

between prediction and ground truth, a quantity that the latent-space MSE on VAE codes never sees. The Euclidean chord is shown for reference; for unit vectors it is a monotone function of

\theta

. (Right) Inverse rendering self-consistency estimates a gradient-detached lighting vector

\hat{\bm{\ell}}_{\text{det}}

, renders Lambertian shading from

\hat{\bm{N}}

, and minimizes

\mathcal{L}_{\text{render}}=1-\mathrm{NCC}(\hat{S},I_{\text{gray}})

, with panel (d) showing standardized local NCC evidence. Together, the losses enforce spherical normal geometry and diffuse image formation in pixel space. (§ IV-C )

Method	Venue	Data $\downarrow$	NYUv2		ScanNet		iBims		Sintel		Avg.
	Venue	Data $\downarrow$	Mean $\downarrow$	$11.25^{\circ}\uparrow$	Mean $\downarrow$	$11.25^{\circ}\uparrow$	Mean $\downarrow$	$11.25^{\circ}\uparrow$	Mean $\downarrow$	$11.25^{\circ}\uparrow$	Rank
Omnidata V2 † [ 84 ]	CVPR 22	12.2M	17.2	55.5	16.2	60.2	18.2	63.9	40.5	14.7	12.7
DSINE [ 17 ]	CVPR 24	160K	16.4	59.6	16.2	61.0	17.1	67.4	34.9	21.5	8.9
GeoWizard ⋆ [ 23 ]	ECCV 24	280K	18.9	50.7	17.4	53.8	19.3	63.0	40.3	12.3	14.5
StableNormal ⋆ [ 24 ]	SA 24	250K	18.6	53.5	17.1	57.4	18.2	65.0	36.7	14.1	13.0
E2E-FT ⋆ [ 4 ]	WACV 25	74K	16.5	60.4	14.7	66.1	16.1	69.7	33.5	22.3	6.0

TABLE I : Quantitative comparison on general scene normal estimation . We evaluate on NYUv2, ScanNet, iBims, and Sintel datasets. Metrics: mean angular error (Mean

\downarrow

) and percentage of pixels within

11.25^{\circ}

(

\uparrow

). TransNormal-2 matches or exceeds MoGe-2 on all eight general-scene metrics while using only 122K training samples (

1.4\%

of MoGe-2’s 8.9M). The best , second best , and third best results are highlighted. ⋆ : diffusion-based; † : transformer-based; ‡ : trained on Sintel (not zero-shot). SA: SIGGRAPH Asia. (§ V-C )

Method	Venue	Data $\downarrow$	ClearGrasp (Synthetic)			TN-Syn			ClearPose (Real-World)			Avg.
	Venue	Data $\downarrow$	Mean $\downarrow$	$11.25^{\circ}\uparrow$	$30^{\circ}\uparrow$	Mean $\downarrow$	$11.25^{\circ}\uparrow$	$30^{\circ}\uparrow$	Mean $\downarrow$	$11.25^{\circ}\uparrow$	$30^{\circ}\uparrow$	Rank
Omnidata [ 16 ]	ICCV 21	12.2M	36.9	15.1	49.1	11.3	80.9	89.3	48.3	10.8	33.8	17.1
Omnidata V2 † [ 84 ]	CVPR 22	12.2M	33.8	18.3	55.9	8.2	87.0	92.6	51.7	13.8	33.2	15.6
DSINE [ 17 ]	CVPR 24	160K	25.7	26.4	68.6	13.2	70.3	90.7	40.2	15.9	46.3	14.2
Marigold ⋆ [ 2 ]	CVPR 24	74K	27.6	31.0	65.3	6.2	90.4	96.3	33.0	25.5	57.5	10.8
GeoWizard ⋆ [ 23 ]	ECCV 24	280K	31.3	20.8	59.5	9.4	78.9	95.0	36.8	14.2	49.7	14.8

TABLE II : Quantitative comparison on transparent object normal estimation . We evaluate on ClearGrasp, our proposed TN-Syn, and ClearPose datasets. Metrics: mean angular error (Mean

\downarrow

, lower is better) and percentage of pixels within

11.25^{\circ}

and

30^{\circ}

thresholds (

\uparrow

, higher is better). Data

\downarrow

: number of labeled training samples for the geometric task. TransNormal-2 achieves the best result on every reported metric using only 122K samples; gains are largest on ClearGrasp and ClearPose, while TN-Syn is near-saturated. See Table I for general scene results. The best , second best , and third best results are highlighted. ⋆ : diffusion-based; † : transformer-based; ‡ : trained on Sintel (not zero-shot). SA: SIGGRAPH Asia. (§ V-C )

Fig. 5 : Qualitative results on unlabeled in-the-wild general objects. We compare TransNormal-2 with representative baselines on examples from in-the-wild images. TransNormal-2 produces coherent object-scale geometry while preserving fine structures on curved ceramic surfaces and mechanical parts. Please zoom in for details. (§ V-B )

Fig. 6 : Qualitative comparison on transparent object normal estimation. We compare our method against state-of-the-art approaches across TN-Syn, ClearGrasp, and ClearPose datasets. For each dataset, the top row shows predicted normals and the bottom row shows angular error maps (blue: low, red: high). Even on ClearPose, an extremely challenging real-world dataset with diverse transparent objects under cluttered scenes, our method produces lower-error zero-shot predictions than the compared approaches. Existing methods often produce blurry or incorrect normals on transparent regions due to refraction, while our method preserves more coherent surface geometry. Please zoom in for details. (§ V-B )

Component	TransNormal	TransNormal-2
Backbone	SD 2.0 (U-Net, 865M)	FLUX.2[klein] (DiT, 9B)
Adaptation	Full fine-tune	LoRA ( $r{=}256$ )
Inference	Single-step prediction	Single-step prediction
Training loss	MSE + Wavelet	MSE + Wavelet + vMF + Rendering
Decode fix	None	GRM ( $\sim$ 0.4M params)
Training data	CG + TN-Syn + HS + VK	CG + TN-Syn + HS + VK

TABLE III : TransNormal vs. TransNormal-2: Key differences and head-to-head mean angular error ( ∘ ,

\downarrow

). TransNormal denotes our prior model [ 1 ] ; both methods are evaluated under identical protocols. (§ V-C )

	Configuration	NYUv2		ScanNet
	Configuration	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$
Group A: Training recipe (cumulative)
(a)	MSE only	16.9	58.8	14.6	64.9
(b)	+ Wavelet + vMF	16.6	58.5	14.3	65.8
(c)	+ Rendering loss	16.0	61.1	13.7	68.2
Group B: Decode refinement (from row c)

TABLE IV : Ablation on key design choices. Group A: cumulative training-recipe ablation; rows (a)–(b) are 20K-step from-scratch controls, and row (c) uses the full staged Phase-1 schedule. Group B: decode refinement. Row (f) is full TransNormal-2. (§ V-D )

Fig. 7 : Qualitative ablation on in-the-wild objects. Starting from the MSE-only baseline, wavelet and vMF supervision recover more coherent local geometry, inverse-rendering self-consistency further improves structure, and the full model with the GRM reduces residual boundary artifacts. Panel letters (b)–(e) correspond to rows (a), (b), (c), and (f) of Table IV , respectively. (§ V-D )

Configuration	#Params	NYUv2		ScanNet		iBims		Sintel		CG-Syn		TN-Syn		ClearPose
Configuration	#Params	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$	Mean $\downarrow$	$11.25^{\circ}\!\uparrow$
$K{=}2$ , hidden $=$ 64	234K	14.7	62.5	12.7	69.2	14.8	70.6	29.2	27.6	11.2	65.1	3.6	95.8	19.1	52.3
$K{=}8$ , hidden $=$ 64	716K	14.7	62.5	12.7	69.1	14.8	70.5	29.2	27.4	11.3	65.0	3.6	95.8	19.1	52.3
$K{=}4$ , hidden $=$ 128	1.47M	14.8	62.3	12.8	68.9	14.7	70.9	29.1	27.7	11.3	64.9	3.5	95.8	19.1	52.2
$K{=}4$ , hidden $=$ 64, w/o RGB	356K	15.7	61.5	13.4	68.6	14.6	72.1	28.8	28.8	11.2	65.2	3.5	95.8	19.1	52.3
$K{=}4$ , hidden $=$ 64 (default)	$\sim$ 0.4M	14.7	62.5	12.7	69.2	14.7	70.6	29.2	27.5	11.3	65.0	3.6	95.8	19.1	52.3

TABLE V : GRM design ablation. We ablate the depth (

K

), width (hidden channels), and input design of the Geometric Refinement Module while keeping the final core predictor, GRM training losses, two-stage GRM training protocol, and inference settings fixed. Metrics: mean angular error (Mean

\downarrow

) and percentage of pixels within

11.25^{\circ}

(

\uparrow

). The RGB-guided depth and width variants are highly stable, with average mean angular error varying by only

0.03^{\circ}

across

K{=}2

,

K{=}4

,

K{=}8

, and hidden

=

128. We use

K{=}4

, hidden

=

64 as a conservative accuracy-parameter-stability tradeoff. Removing RGB guidance mainly hurts NYUv2 and ScanNet, while remaining competitive on synthetic and transparent benchmarks. (§ V-E )

Rank	#Params	General				Transparent			Avg $\downarrow$
Rank	#Params	NYUv2	ScanNet	iBims	Sintel	CG-Syn	TN-Syn	ClearPose	Avg $\downarrow$
64	$\sim$ 175M	16.6	14.6	15.9	30.6	12.0	4.6	21.1	16.5
128	$\sim$ 350M	16.0	14.1	15.7	30.6	12.3	5.4	21.0	16.5
256	$\sim$ 700M	16.0	13.7	14.8	28.9	11.4	3.9	19.3	15.4
512	$\sim$ 1.4B	16.8	14.6	16.0	31.2	12.6	4.0	20.6	16.5

TABLE VI : LoRA rank ablation. We vary the LoRA rank

r

(with

\alpha{=}r

) and evaluate the core predictor with GRM refinement disabled, isolating the capacity of the DiT LoRA adapter. Rank 256 gives the best average error and matches or leads the best rounded value on every dataset, with no further gains at rank 512. (§ V-F )

Method	NYUv2	ScanNet	CG-Syn	TN-Syn	Avg.
No refinement	16.0	13.7	11.4	3.9	11.3
Guided Filter ( $r_{\text{GF}}{=}4$ )	14.8	12.8	12.4	4.7	11.2
Guided Filter ( $r_{\text{GF}}{=}8$ )	15.0	12.9	14.5	5.6	12.0
Guided Filter ( $r_{\text{GF}}{=}16$ )	16.3	13.8	18.8	7.1	14.0
Joint Bilateral	15.0	13.0	11.7	4.2	11.0
Bilateral	15.4	13.4	11.4	3.9	11.0

TABLE VII : Learned GRM vs. classical image-guided filters. All rows refine the same coarse normal predictions and are evaluated with the harness of the main tables, so “No refinement” matches the core-predictor results reported there. Mean angular error ( ∘ ,

\downarrow

); the GRM is best on every dataset and on average. (§ V-G )

	Edge ( ∘ )		Non-Edge ( ∘ )		$\Delta$ ( ∘ )
Dataset	w/o	w/	w/o	w/	Edge	Non-E
NYUv2	26.1	23.5	13.7	12.9	$-2.6$	$-0.8$
ScanNet	26.0	23.4	12.5	11.9	$-2.6$	$-0.6$

TABLE VIII : Edge vs. non-edge MAE analysis. Pixels are separated into edge (Canny on RGB, dilated

3{\times}3

) and non-edge regions. We compare the final core predictor with and without the GRM on general-scene benchmarks, where RGB edges provide a reliable proxy for geometric discontinuities. Negative

\Delta

means lower MAE after GRM. (§ V-H )

Fig. 8 : Frequency-wise ratio of angular-error PSD after and before GRM refinement. Values below

1

indicate lower spatial error energy after refinement. The shaded region marks the highest third of radial-frequency bins; integrated energy in this band decreases by

53.2\%

on NYUv2 and

42.9\%

on ScanNet. (§ V-I )

TransNormal-2: Geometry-Grounded Rectified Flow with Edge-Aware Decoding for Precise Normal Estimation

Abstract

Figures & tables

Surface Normal Estimation

GeoStereo: A Unified Stereo Geometry Estimation Framework for Disparity and Surface Normal

Towards Consistent Video Geometry Estimation

Any Resolution Any Geometry: From Multi-View To Multi-Patch

Abstract

Surface Normal Estimation

Explore similar work

GeoStereo: A Unified Stereo Geometry Estimation Framework for Disparity and Surface Normal

Towards Consistent Video Geometry Estimation

Any Resolution Any Geometry: From Multi-View To Multi-Patch