Revisiting an Old Perspective Projection for Monocular 3D Morphable Models Regression

Toby Chong; Ryota Nakajima

cs.CVMar 5, 2026

Revisiting an Old Perspective Projection for Monocular 3D Morphable Models Regression

Authors: Toby Chong, Ryota Nakajima

Organizations: TOEI Company Tokyo, Japan

Abstract

We introduce a novel camera model for monocular 3D Morphable Model (3DMM) regression methods that effectively captures the perspective distortion effect commonly seen in close-up facial images. Fitting 3D morphable models to video is a key technique in content creation. In particular, regression-based approaches have produced fast and accurate results by matching the rendered output of the morphable model to the target image. These methods typically achieve stable performance with orthographic projection, which eliminates the ambiguity between focal length and object distance. However, this simplification makes them unsuitable for close-up footage, such as that captured with head-mounted cameras. We extend orthographic projection with a new shrinkage parameter, incorporating a pseudo-perspective effect while preserving the stability of the original projection. We present several techniques that allow finetuning of existing models, and demonstrate the effectiveness of our modification through both quantitative and qualitative comparisons using a custom dataset recorded with head-mounted cameras.

Figures & tables

Figure 2: Visualization of the newly introduced shrinkage parameter $\rho$ . We estimate the 3DMM and camera parameters using SMIRK [ 19 ] , which employs orthogonal projection ( $\rho=0$ ). We vary $\rho$ from 0.0 to 5.0 while holding all other parameters constant. Unlike perspective projection, which relies on the combination of $f$ and $t_{z}$ to control the shrinkage effect, $\rho$ isolates this effect and can therefore be incorporated into existing 3DMM regression methods via fine-tuning.

Landmark Region	Method	MEAD	HMC1M
	$smirk_{p}$	7.941	5.861
Jaw line	$smirk_{r}$	7.004	5.065
	Ours	10.16	4.646
	$smirk_{p}$	2.763	3.236
Facial	$smirk_{r}$	2.058	1.737
	Ours	1.565	1.655

Table 1: Landmark reconstruction loss on the MEAD and HMC1M datasets. We report the reconstruction loss for the jaw line and the rest of the facial landmarks separately. All numbers are multiplied by

10^{4}

for readability.

Figure 3: We compare our method with the pretrained SMIRK model, on the images used in the SMIRK paper. They provide similar reconstruction quality visibly.

Reconstruction Loss	NoW Neutral	NoW Selfie
$smirk_{p}$	1.2563	1.2718
Ours	1.2320	1.2143
MICA [ 26 ]	1.1162	1.1238

Table 2: Reconstruction loss on the NoW dataset. We report the reconstruction loss for the NoW neutral and selfie subset.

Figure 4: We compare our method with the pretrained and retrained versions of the SMIRK model, on the images from the HMC1M dataset.

	HMC1M	MEAD	CelebA	FFHQ
$\rho$	$2.95\pm 0.96$	$1.35\pm 0.42$	$0.69\pm 0.31$	$0.68\pm 0.29$

Table 3: Estimated

\rho

values for all datasets used. We sampled and computed the average

\rho

from 10k images from each of the datasets, under our camera model. HMC1M experiencing the highest distortion, and CelebA and FFHQ exhibiting minimal distortion.

Figure 5: We finetuned the baseline SMIRK model with full perspective projection. The network would fail to properly capture the perspective distortion effect, and the results remain mostly orthogonal (large $f$ value). For comparsion, this converts to $\rho\approx 0.78$ in our camera model, while our method predicts a more visually similar result with $\rho=1.94$ .

3D Face Reconstruction

3D face reconstruction estimates facial shape and appearance from images or video to create usable models and avatars.

See all 54 papers

3D Human Face Reconstruction with 3DMM face model from RGB image
May 5, 2026Zhangnan Jiang, Zichen Yang3D Morphable Models 3D Mesh Reconstruction
Tandon School of Engineering New York University
Profile-Specific 3DMM Regression from a Single Lateral Face Image
May 3, 2026Taiki Kanaya, Hideo Saito3D Morphable Models 3D Mesh Reconstruction
Keio University
CAM3R: Camera-Agnostic Model for 3D Reconstruction
Mar 23, 2026Namitha Guruprasad, Abhay Yadav, Cheng Peng +13D Reconstruction Multi-View 3D Reconstruction
Johns Hopkins University, Baltimore, MD 21218, USA · University of Virginia, Charlottesville, VA 22904, USA

May 5, 2026J/K move · Enter open · S save

3D Human Face Reconstruction with 3DMM face model from RGB image

Zhangnan Jiang, Zichen Yang

Tandon School of Engineering New York University

3D Morphable Models 3D Mesh Reconstruction 3D Human Reconstruction 3D Face Reconstruction Single-View 3D Reconstruction Feed-Forward 3D Reconstruction

Open paper AskPDF arXiv

Nowadays as convolution neural networks demonstrate its powerful problem-solving ability in the area of image processing, efforts have been made to reconstruct detailed face shapes from 2D face images or videos. However, to make the full use of CNN, a large number of labeled data is required to train the network. Coarse morphable face model has been used to synthesize labeled data. However, it is hard for coarse morphable face models to generate photo-realistic data with detail such as wrinkles. In this project, we present a pipeline that reconstructs a human face 3D model from a single RGB image. The pipeline includes face detection, landmark detection, regression of 3DMM model parameters, and soft rendering. Mentor: Zhipeng Fan (Email: [email protected]) Code Repository: https://github.com/SeVEnMY/3d-face- reconstruction Code Reference: https://github.com/sicxu/Deep3DFaceRecon pytorch

Abstract

3D Face Reconstruction

Explore similar work

3D Human Face Reconstruction with 3DMM face model from RGB image

Profile-Specific 3DMM Regression from a Single Lateral Face Image

CAM3R: Camera-Agnostic Model for 3D Reconstruction