Textured Mesh Research Articles

3D-PSSIM: Projective Structural Similarity for 3D Mesh Quality Assessment Robust to Topological Irregularities.

Despite acceleration in the use of 3D meshes, it is difficult to find effective mesh quality assessment algorithms that can produce predictions highly correlated with human subjective opinions. Defining mesh quality features is challenging due to the irregular topology of meshes, which are defined on vertices and triangles. To address this, we propose a novel 3D projective structural similarity index ( 3D- PSSIM) for meshes that is robust to differences in mesh topology. We address topological differences between meshes by introducing multi-view and multi-layer projections that can densely represent the mesh textures and geometrical shapes irrespective of mesh topology. It also addresses occlusion problems that occur during projection. We propose visual sensitivity weights that capture the perceptual sensitivity to the degree of mesh surface curvature. 3D- PSSIM computes perceptual quality predictions by aggregating quality-aware features that are computed in multiple projective spaces onto the mesh domain, rather than on 2D spaces. This allows 3D- PSSIM to determine which parts of a mesh surface are distorted by geometric or color impairments. Experimental results show that 3D- PSSIM can predict mesh quality with high correlation against human subjective judgments, across the presence of noise, even when there are large topological differences, outperforming existing mesh quality assessment models.

Generating a virtual acoustic environment to approximate the soundscape of a historical cathedral

This paper focuses on the generation of a virtual acoustic environment of a historical cathedral basilica located in St. Augustine, Florida, USA. The virtual acoustic environment is an effective means to study a cultural heritage's soundscape that restricts visitors from accessing and being quiet. The author used a laser scanner to capture the reality of the cathedral's interior, in order to maximize the visual perception of the worship space, and created a 3D digital textured mesh model using the points cloud generated from laser scanning. The spatial experience recorded by a 360 camera with an ambisonic microphone provided the virtual soundwalk at multiple locations. The background and foreground sound were synthesized in Unity using the same music recordings that the church used during the day. The sound pressure levels were closely matched with the ones measured by a sound level meter. The virtual acoustic environment was simulated by using spatial applications in Unity and optimized through the comparison of the room acoustical measurements performed based on ISO 3382.

Supergene formation of sulfur-rich, tochilinite-bearing serpentinites in the Oman ophiolite

Mass transfer processes between fluids and ultramafic rocks produce subsurface environments encompassing a wide range of redox conditions. A notable locality where an extensive range of redox conditions is observed in one location is Hole BA1B, a ∼ 400 m borehole drilled by the Oman Drilling Project. A sulfur-enriched serpentinite zone, containing up to 0.6 wt% S, occurs between shallow oxidized serpentinites (<30 m) and deep partially serpentinized harzburgite (>150 m). All three alteration zones are predominantly composed of serpentine. However, microanalysis of samples from the sulfur-enriched zone shows that mesh textures after olivine are composed of serpentine, brucite, and tochilinite mixtures, yielding optically black thin-section samples that characterize this sulfidic zone. It is proposed that sulfur accumulates in this zone via a process similar to those found in supergene ore deposits. Reaction-path models show that at shallow conditions open to atmospheric input, sulfur is mobilized via oxidative weathering of serpentinized dunite and harzburgite. Sulfate-bearing fluids percolate deeper and react with host rocks in a system closed to atmospheric input. As fluids become more reduced, dissolved sulfate is precipitated as sulfide minerals yielding rocks with ∼0.4 wt% S, like those observed in Hole BA1B. Despite enrichment of S in the sulfidic zone in Hole BA1B, Ni and Co contents are uniform throughout all three layers in the borehole. This is consistent with model results which show that Ni (and, by analogy, Co) is less mobile than S, and can be hosted in serpentine and NiFe alloys in addition to sulfides. The sulfur enrichment process may occur abiotically. However, sulfide enrichment via microbial reduction of sulfate and other sulfur species can also facilitate the formation of the sulfidic zone. Bioenergetic calculations show that abundant energy is available for sulfur reducing microbes, consistent with previous work demonstrating the presence of active, sulfate-reducing microorganisms in Hole BA1B and other nearby boreholes. This suggests that the observed sulfur enrichment is an ongoing process. Overall, this work shows that variable redox conditions are attained as fluids percolate and react with serpentinized ultramafic rocks at variable extents of interaction between aquifer fluids, host ultramafic rocks, and the atmosphere.

3D textured mesh scene data fusion method considering multiple scales and resolutions

ABSTRACT The fusion of 3D textured mesh scene data can effectively and seamlessly integrate multiple spatially related 3D geographical scene models into larger and more comprehensive datasets. Existing methods usually have different degrees of texture distortion and unreduced texture problems in the splicing area, and these methods do not consider the current situation of Level of Detail(LoD) data organisation in the 3D textured mesh model. We propose a 3D textured mesh scene data fusion method considering multiple scales and resolutions. First, seamless extraction of splicing areas was achieved based on spatial relationship constraints and orthogonal inverse projection principles. Second, the sparse scene structure concept was employed to generate sparse summary scene images. Finally, a dual simplification operator for both mesh and texture was proposed to construct multi-resolution 3D textured mesh models, simultaneously reconstructing triangular facets and textures within the splicing regions. We used multiple 3D textured mesh data constructed from aerial obliques to conduct data fusion experimental verification. The results indicate that our method exhibits minimal texture distortion during grid reconstruction in splicing areas and model simplification, effectively reducing the amount of simplified texture data. Additionally, our method adeptly addresses multilevel detailed 3D textured mesh models, demonstrating good feasibility and advancement.

Virtual reality in geoeducation: the case of the Lesvos Geopark

ABSTRACT The aim of this study is to investigate the contribution of cartographic results obtained from data acquired with Unmanned Aerial Vehicles to the management procedures of geoheritage monuments, specifically in geoeducation. 2D orthophotomaps and 3D textured meshes of a fossil site in the Lesvos Geopark, Greece, which has been monitored since 2018, were generated. The data recordings were collected using a UAV at different flight altitudes for three different cartographic scales. After acquiring UAV and Global Navigation Satellite Systems data, the resulting cartographic products were used to create educational material on geosites. This material was used to deliver lectures to students in various settings: (a) in situ at the Natural History Museum of the Lesvos Petrified Forest, (b) live in a traditional classroom, (c) remotely via e-class software, and (d) in a collaborative Virtual Reality environment. Feedback was provided by students through customized questionnaires, and their knowledge acquisition was assessed post-lecture with specially designed assignments. The final section of this work compares learning experiences and knowledge acquisition across these different environments. The results contribute to the debate on 2D and 3D cartography in geoeducation, quantifying the value of new materials in understanding spatial and geographic features and improving geoheritage management.

Surface‐aware Mesh Texture Synthesis with Pre‐trained 2D CNNs

AbstractMesh texture synthesis is a key component in the automatic generation of 3D content. Existing learning‐based methods have drawbacks—either by disregarding the shape manifold during texture generation or by requiring a large number of different views to mitigate occlusion‐related inconsistencies. In this paper, we present a novel surface‐aware approach for mesh texture synthesis that overcomes these drawbacks by leveraging the pre‐trained weights of 2D Convolutional Neural Networks (CNNs) with the same architecture, but with convolutions designed for 3D meshes. Our proposed network keeps track of the oriented patches surrounding each texel, enabling seamless texture synthesis and retaining local similarity to classical 2D convolutions with square kernels. Our approach allows us to synthesize textures that account for the geometric content of mesh surfaces, eliminating discontinuities and achieving comparable quality to 2D image synthesis algorithms. We compare our approach with state‐of‐the‐art methods where, through qualitative and quantitative evaluations, we demonstrate that our approach is more effective for a variety of meshes and styles, while also producing visually appealing and consistent textures on meshes.

Evaluation of multi-camera images in different SfM-MVS based photogrammetric software and comparison of digital products in generating 3D city models

High-resolution 3D city modeling is generated using the oblique and vertical (nadir) imaging features of multi-camera systems. There are several SfM-MVS based photogrammetric software available. They enable high-resolution image processing and matching and acquire successful outcomes. The orientation of large photogrammetric blocks consisting of multiple sets of images and the accuracy of the 3D city models to be obtained vary according to the ability of the software using these approaches to solve large image matrix problems. In this study, a 3D city model was produced with vertical and oblique images obtained with UltraCam Osprey Mark 3 Premium camera in a selected application area in Kütahya city center. The main purpose of the study is to evaluate the point cloud, textured mesh model, digital elevation model, true orthophoto products and vectorial roof detail drawings via utilizing the vertical + oblique images with different SfM-MVS based photogrammetric software and to compare the results.

Three-Dimensional Digitization of Documentation and Perpetual Preservation of Cultural Heritage Buildings at Risk of Liquidation and Loss—The Methodology and Case Study of St Adalbert’s Church in Chicago

This paper presents a three-dimensional (3D) digitization methodology for documenting the appearance and geometry of cultural heritage buildings using modern tools for data acquisition and processing. This work presents a method combining laser scanning of building dimensions with photographic texture acquisition techniques in order to develop accurate, photo-realistic 3D models. This work also presents a method for digital reconstruction of the elements of interiors. This case study presents the process and result of scanning the church of St Adalbert in Chicago along with its inner finishing elements, achieved during the interventional 3D scanning of a historical building interior. The obtained colorized point cloud has 3.5 billion points and a volume of 65 GB. Its creation took nearly 3.5 h on a high-end computer. The generated textured mesh model has 1.6 billion triangles and a volume of 60 GB. The analysis of the results showed that the elements of the church furnishings were reproduced with very high accuracy. The developed 3D model of the interior, appearance and dimensions of the church is its perpetual documentation. It can be used for various purposes, such as popularizing the appearance on the Internet, scientific research on interior artefacts, creating files for VR and shooting high-resolution films.

Assessing the utility of 3D modeling with photogrammetry in assigned sex estimation from the greater sciatic notch

Assigned sex estimation via the greater sciatic notch (GSN) is traditionally performed via physical/visual examination and ordinal scoring; however, this relies on the subjective assessment of morphology for typological classification which may not be reflective of human variation. Three-dimensional (3D) photogrammetry may offer a technologically advanced, low cost, and more objective alternative to assess the complex curvature of anatomical landmarks. This research explores the accuracy of photogrammetry derived 3D models by comparing digital measurements to those obtained from the skeletal elements and to streamline the application of curvature analysis for the estimation of assigned sex from the GSN. This study utilizes the left and right os coxae from 15 skeletal individuals (5 females, 10 males) from the Boston University Chobanian & Avedisian School of Medicine. A Fujifilm X-Pro2 and Fujifilm 35 mm prime lens captured 123 images per element, which were processed in Meshroom by AliceVision® to create a 3D textured mesh. The mesh was exported into Blender for cleanup, scaling, measurement, and curvature analysis. The measurements were between 96.54 % and 99.94 % consistent across methods and observations. The consistency between digital metric observations increased by an average of 0.07 % when compared to the consistency of the dry bone measurements. Additionally, curvature analysis of the GSN correctly estimated the assigned sex of all os coxae in the sample. This study demonstrates that photogrammetry is an accurate and reliable method for the digitization of remains that enables analytical techniques to better capture skeletal variation compared to traditional methods.

3D Semantic Segmentation of Aerial Photogrammetry Models Based on Orthographic Projection

Semantic segmentation of 3D scenes is one of the most important tasks in the field of computer vision and has attracted much attention. In this paper, we propose a novel framework for 3D semantic segmentation of aerial photogrammetry models, which uses orthographic projection to improve efficiency while still ensuring high precision, and can also be applied to multiple types of models (i.e., textured mesh or colored point cloud). In our pipeline, we first obtain RGB images and elevation images from the 3D scene through orthographic projection, then use the image semantic segmentation network to segment these images to obtain pixel-wise semantic predictions, and finally back-project the segmentation results to the 3D model for fusion. Specifically, for the image semantic segmentation model, we design a cross-modality feature aggregation module and a context guidance module based on category features, which assist the network in learning more discriminative features between different objects. For the 2D-3D semantic fusion, we combine the segmentation results of the 2D images with the geometric consistency of the 3D models for joint optimization, which further improves the accuracy of the 3D semantic segmentation. Extensive experiments on two large-scale urban scenes demonstrate the efficiency and feasibility of our algorithm and surpass the current mainstream 3D deep learning methods.

STAR-TM: STructure Aware Reconstruction of Textured Mesh From Single Image.

We present a novel method for single-view 3D reconstruction of textured meshes, with a focus to address the primary challenge surrounding texture inference and transfer. Our key observation is that learning textured reconstruction in a structure-aware and globally consistent manner is effective in handling the severe ill-posedness of the texturing problem and significant variations in object pose and texture details. Specifically, we perform structured mesh reconstruction, via a retrieval-and-assembly approach, to produce a set of genus-zero parts parameterized by deformable boxes and endowed with semantic information. For texturing, we first transfer visible colors from the input image onto the unified UV texture space of the deformable boxes. Then we combine a learned transformer model for per-part texture completion with a global consistency loss to optimize inter-part texture consistency. Our texture completion model operates in a VQ-VAE embedding space and is trained end-to-end, with the transformer training enhanced with retrieved texture instances to improve texture completion performance amid significant occlusion. Extensive experiments demonstrate higher-quality textured mesh reconstruction obtained by our method over state-of-the-art alternatives, both quantitatively and qualitatively, as reflected by a better recovery of texture coherence and details.

From coastal geomorphometry to virtual environments

Communicating environmental change and mitigation scenarios to stakeholders and decision-makers can be challenging. Immersive environments offer an innovative approach for knowledge transfer, allowing science-based scenarios to be discussed interactively. The use of such environments is particularly helpful for the analysis of large, multi-component geospatial datasets, as commonly employed in the classification of ecosystems. Virtual environments can play an important role in conveying and discussing the findings gathered from these geomorphometric datasets. However, textured meshes and point clouds are not always well suited for direct import to a virtual reality or the creation of a truly immersive environment, and often result in geometrical artifacts, which can be misinterpreted during the import to a game engine. Such technical hurdles may lead to viewers rejecting the experience altogether, failing to achieve a higher educational purpose. In this study, we apply an asset-based approach to create an immersive virtual representation of a coastal environment. The focus hereby is on the coastal vegetation and changes in species distribution, which could potentially be triggered by the impact of climate change. We present an easy-to-use blueprint for the game engine EPIC Unreal Engine 5. In contrast to traditional virtual reality environments, which use static textured mesh data derived from photogrammetry, this asset-based approach enables the use of dynamic and physical properties (e.g. vegetation moving due to wind or waves), which makes the virtual environment more immersive. This will help to stimulate understanding and discussion amongst different stakeholders, and will also help to foster inclusion in earth- and environmental science education.

Deep learning based multi-view stereo matching and 3D scene reconstruction from oblique aerial images

In this paper, we propose a practical three-dimensional (3D) real-scene reconstruction framework named Deep3D, which is paired with a deep learning based multi-view stereo (MVS) matching model named the adaptive multi-view aggregation matching (Ada-MVS) model, to obtain a 3D textured mesh model from multi-view oblique aerial images. Deep3D is the first deep learning based framework for 3D scene reconstruction, in which aerial triangulation and view selection are first performed on the input images, and the depth map of each image is then inferred using the pretrained Ada-MVS model. All the inferred depth maps are then fused into a dense point cloud after filtering the outliers. Finally, the 3D textured mesh is extracted from the dense 3D points as the final product. In the Ada-MVS model, a novel adaptive inter-view aggregation module is specially proposed to address the inconsistent information among oblique views and to fuse the multi-view costs into a robust cost volume. A lightweight recurrent regularization module is also designed for high-efficiency processing of high-capacity aerial images with large depth variations. Moreover, as oblique aerial image datasets are currently lacking, we built a large-scale synthetic multi-view oblique aerial image dataset (WHU-OMVS dataset) for deep learning based model training and methodology evaluation for the task of 3D scene reconstruction. The experimental results show that, firstly, the proposed Ada-MVS model has obvious advantages when used with high-capacity oblique aerial images, compared with several relevant learning-based MVS methods. Secondly, through a comprehensive comparison with popular commercial software packages and open-source solutions, it is shown that the proposed Deep3D framework outperforms all the other solutions in terms of reconstruction quality, and outperforms all the open-source solutions and some of the software packages in terms of efficiency on the WHU-OMVS dataset. Thirdly, the Deep3D framework shows a stable generalization ability and excellent performance when applied to other oblique or nadir aerial images, without any further fine-tuning. The dataset and code will be available at http://gpcv.whu.edu.cn/data.

Sat-Mesh: Learning Neural Implicit Surfaces for Multi-View Satellite Reconstruction

Automatic reconstruction of surfaces from satellite imagery is a hot topic in computer vision and photogrammetry. State-of-the-art reconstruction methods typically produce 2.5D elevation data. In contrast, we propose a one-stage method directly generating a 3D mesh model from multi-view satellite imagery. We introduce a novel Sat-Mesh approach for satellite implicit surface reconstruction: We represent the scene as a continuous signed distance function (SDF) and leverage a volume rendering framework to learn the SDF values. To address the challenges posed by lighting variations and inconsistent appearances in satellite imagery, we incorporate a latent vector in the network architecture to encode image appearances. Furthermore, we introduce a multi-view stereo constraint to enhance surface quality. This constraint minimizes the similarity between image patches to optimize the position and orientation of the SDF surface. Experimental results demonstrate that our method achieves superior visual quality and quantitative accuracy in generating mesh models. Moreover, our approach can learn seasonal variations in satellite imagery, resulting in texture mesh models with different and consistent seasonal appearances.

Tribological properties of high-speed steel surface with texture and vertical fibers

Inadequate lubrication of the two touching surfaces during friction can lead to severe wear, especially in metal cutting. Therefore, a surface with synergistic anti-friction effect of texture and solid lubricant was proposed to improve lubrication. A mesh texture with excellent wettability was prepared on the high-speed steel (HSS) surface by laser, and then nylon fibers were vertically implanted into the grooves of the texture using the electrostatic flocking technology. The friction and wear state of different surfaces (smooth, textured, flocking) under dry/oil-lubricated were studied by a linear reciprocating wear tester. The coefficient of friction (COF) under different working conditions was used to analyze the anti-friction properties, and the wear rate was used to evaluate the wear resistance of the surface. The results showed that the tribological properties of flocking surfaces were better than those of the other two surfaces. This is because the addition of nylon fibers eases shear at the edges of the texture. The broken fibers form a solid lubricating film on the specimen surface, which prevents the surface from being scratched by debris. In addition, it is found that COF decreases with increasing load. Finally, the rapid wettability of the oil droplets on the flocking surface shows the great potential of the surface for lubrication and anti-friction.

Progressive carbonation and Ca-metasomatism of serpentinized ultramafic rocks: insights from natural occurrences and hydrothermal experiments

Hydration, carbonation, and related metasomatism of mantle peridotite play a significant role in the global geochemical cycle. In this study, we combined an analysis of carbonated serpentinite with hydrothermal experiments on carbonation and Ca-metasomatism for samples from the Manlay ophiolite, southern Mongolia to investigate that carbonation mechanism of the serpentinite body after serpentinization. Samples show that the serpentinite was either transected by calcite and dolomite veins or was completely replaced by carbonates (calcite with minor dolomite) and quartz, in which the original mesh texture of serpentinite was preserved. Carbonation occurred after low-temperature serpentinization (lizardite/chrysotile), suggesting that carbonation occurred at temperatures lower than 300 ˚C. Calcite in the serpentinite showed δ13 CVPDB values ranging from -8.83 to -5.11 ‰ and δ18 OVSMOW from + 20.1 to + 24.4 ‰, suggesting that CO2 in the fluids could be derived from the degradation of organic material or methanotrophic processes rather than the origin of seafloor limestone. Three batch-type experiments, i.e., single step experiments (1) Olivine + NaHCO3,aq + CaCl2,aq and (2) Chrysotile + NaHCO3,aq + wollastonite (Ca source), and two steps experiment (3) Olivine carbonation and Ca-metasomatism, were conducted at 275 °C and 5.7 MPa to constrain the mechanism of calcite replacement of serpentinite. We found that calcite precipitated from the solution directly in the first two experiments, but replacement of serpentinite by calcite was not observed. In contrast, the third experiment caused the initial carbonation to form magnesite and then changed to calcite by later alteration. The natural occurrences and experiments revealed the possibility that the carbonation of olivine followed by Ca-rich fluid infiltration produced calcite in the carbonated serpentinite. Such Ca-metasomatism of Mg carbonates could easily occur in the ultramafic bodies and significantly affect the global carbon cycle.

Crafting the MPEG metrics for objective and perceptual quality assessment of volumetric videos.

Efficient objective and perceptual metrics are valuable tools to evaluate the visual impact of compression artifacts on the visual quality of volumetric videos (VVs). In this paper, we present some of the MPEG group efforts to create, benchmark and calibrate objective quality assessment metrics for volumetric videos represented as textured meshes. We created a challenging dataset of 176 volumetric videos impaired with various distortions and conducted a subjective experiment to gather human opinions (more than 5896 subjective scores were collected). We adapted two state-of-the-art model-based metrics for point cloud evaluation to our context of textured mesh evaluation by selecting efficient sampling methods. We also present a new image-based metric for the evaluation of such VVs whose purpose is to reduce the cumbersome computation times inherent to the point-based metrics due to their use of multiple kd-tree searches. Each metric presented above is calibrated (i.e., selection of best values for parameters such as the number of views or grid sampling density) and evaluated on our new ground-truth subjective dataset. For each metric, the optimal selection and combination of features is determined by logistic regression through cross-validation. This performance analysis, combined with MPEG experts' requirements, lead to the validation of two selected metrics and recommendations on the features of most importance through learned feature weights.

Textured Mesh Quality Assessment: Large-scale Dataset and Deep Learning-based Quality Metric

Over the past decade, three-dimensional (3D) graphics have become highly detailed to mimic the real world, exploding their size and complexity. Certain applications and device constraints necessitate their simplification and/or lossy compression, which can degrade their visual quality. Thus, to ensure the best Quality of Experience, it is important to evaluate the visual quality to accurately drive the compression and find the right compromise between visual quality and data size. In this work, we focus on subjective and objective quality assessment of textured 3D meshes. We first establish a large-scale dataset, which includes 55 source models quantitatively characterized in terms of geometric, color, and semantic complexity, and corrupted by combinations of five types of compression-based distortions applied on the geometry, texture mapping, and texture image of the meshes. This dataset contains over 343k distorted stimuli. We propose an approach to select a challenging subset of 3,000 stimuli for which we collected 148,929 quality judgments from over 4,500 participants in a large-scale crowdsourced subjective experiment. Leveraging our subject-rated dataset, a learning-based quality metric for 3D graphics was proposed. Our metric demonstrates state-of-the-art results on our dataset of textured meshes and on a dataset of distorted meshes with vertex colors. Finally, we present an application of our metric and dataset to explore the influence of distortion interactions and content characteristics on the perceived quality of compressed textured meshes.

Multiview Textured Mesh Recovery by Differentiable Rendering

Although having achieved the promising results on shape and color recovery through self-supervision, the multi-layer perceptrons-based methods usually suffer from heavy computational cost on learning the deep implicit surface representation. Since rendering each pixel requires a forward network inference, it is very computational intensive to synthesize a whole image. To tackle these challenges, we propose an effective coarse-to-fine approach to recover the textured mesh from multi-views in this paper. Specifically, a differentiable Poisson Solver is employed to represent the object's shape, which is able to produce topology-agnostic and watertight surfaces. To account for depth information, we optimize the shape geometry by minimizing the differences between the rendered mesh and the predicted depth from multi-view stereo. In contrast to the implicit neural representation on shape and color, we introduce a physically based inverse rendering scheme to jointly estimate the environment lighting and object's reflectance, which is able to render the high resolution image at real-time. The texture of the reconstructed mesh is interpolated from a learnable dense texture grid. We have conducted the extensive experiments on several multi-view stereo datasets, whose promising results demonstrate the efficacy of our proposed approach. The code is available at https://github.com/l1346792580123/diff.

Incompatibility between serpentinization and epidote formation in the lower oceanic crust: Evidence from the Oman Drilling Project

AbstractIt is a general tendency that epidote, which is a typical greenschist facies mineral, is scarce in the lower oceanic crust, in spite of the widespread occurrence of the other minerals indicative of similar temperature conditions such as chlorite, actinolite, prehnite and serpentine. To find the cause of this, we carried out petrological analyses of lower crustal rocks of the Oman ophiolite sampled by the Oman Drilling Project of the International Continental Scientific Drilling Program (ICDP). Petrographic observations revealed the tendency, as expected, that the amount of epidote formed by static alteration of plagioclase decreases with depth. Because mineral assemblages indicative of a wide range of temperature conditions from amphibolite to subgreenschist facies occur throughout the cores without systematic variations of abundance, the decrease of epidote amount cannot be explained by the difference of temperature condition of alteration. Petrographic observations also revealed that epidote is absent or rare in rocks containing serpentinized olivine in contrast to prehnite showing a close association with serpentinization of olivine. In an exceptional sample containing both epidote and serpentinized olivine, epidote occurs with chlorite that cuts or replaces plagioclase, mantles adjacent olivine and is connected with chlorite + lizardite veins cutting mesh‐forming serpentine veins. The distribution and mode of occurrence of epidote suggest decoupling of its formation with the main stage of serpentinization. Serpentine veins cutting olivine to form mesh texture are typically lizardite with magnetite ribbons at vein centres and have compositions of lizardite–cronstedtite solid solution at vein margins or in magnetite‐free veins, suggesting a chemical condition with low silica and low oxygen potentials at an early stage of serpentinization. Thermodynamic modelling for olivine and plagioclase alteration at greenschist facies conditions indicates that silica potential for plagioclase alteration to form prehnite + chlorite and epidote + chlorite could be higher than for olivine serpentinization. On the other hand, oxygen potential for the prehnite + chlorite formation is lower than for the epidote + chlorite formation and is comparable with that for olivine serpentinization. From the observations and analyses, it is concluded that epidote formation is inhibited by olivine serpentinization, which maintains a reducing condition for alteration in the lower oceanic crust.