基于Diffusers框架的电商产品图制作方法研究

基于Diffusers框架与Control Net网络,文章开发并实现了服装产品图的制作流程,将Segment Anything与Grounding DINO结合,实现了自动分割衣服蒙版。将Semantic Segmentation与ViTmatte结合,实现了对皮肤的精细化分割。文章使用大模型的低秩适应实现了对图像风格的微调,利用Ultralytics框架,结合Diffusers提出了一种图像修复方法。将生成图与原图对比,实验结果表明:将Diffusers与Control Net网络结合,能够实现模特替换与背景替换,该方法可以为用户提供衣服上身的参考,降低商家的产品图制作成本。

Experimental Research on Sewage Outfall Diffusers

The influence of diffuser parameters, including the riser spacing, port number in a riser, injection angle, port arrangement, etc., on the surface initial dilution is experimentally investigated. The relative density difference between the effluent and the sea water in the model is the same as that in the prototype, and the effect of the cross current is simulated by an inverse model technique. Based on the result analysis, the arrangement with more ports in a riser and larger riser spacing is suggested to save construction cost. The relationship between the Reynolds number based on the port diameter and velocity, and the surface initial dilution is also explored, and the critical Reynolds number is proposed.

Merging Effluent Discharge Plumes from Multiport Diffusers on a Sloping Beach

Multiport diffusers are the effective engineering devices installed at the marine outfall systems for the steady discharge of effluent streams from the modern coastal plants, such as municipal sewage treatment, power generation and seawater desalination. A far field mathematical model using a two-dimensional advection-diffusion equation is presented for continuous discharges of effluent streams from multiple outfalls on a uniformly sloping beach with a current parallel to the shoreline. The analytical solutions are illustrated graphically to replicate and capture the merging process of effluent plumes in shallow coastal waters, and then asymptotic approximation will be made to the maximum shoreline’s concentration to formulate effluent discharge plume dilution from a multiport diffuser.

Optical information transfer through random unknown diffusers using electronic encoding and diffractive decoding

Free-space optical information transfer through diffusive media is critical in many applications, such as biomedical devices and optical communication, but remains challenging due to random, unknown perturbations in the optical path. We demonstrate an optical diffractive decoder with electronic encoding to accurately transfer the optical information of interest, corresponding to, e.g., any arbitrary input object or message, through unknown random phase diffusers along the optical path. This hybrid electronic-optical model, trained using supervised learning, comprises a convolutional neural network-based electronic encoder and successive passive diffractive layers that are jointly optimized. After their joint training using deep learning,our hybrid model can transfer optical information through unknown phase diffusers, demonstrating generalization to new random diffusers never seen before. The resulting electronic-encoder and optical-decoder model was experimentally validated using a 3D-printed diffractive network that axially spans <70λ, whereλ = 0.75 mm is the illumination wavelength in the terahertz spectrum, carrying the desired optical information through random unknown diffusers. The presented framework can be physically scaled to operate at different parts of the electromagnetic spectrum, without retraining its components, and would offer low-power and compact solutions for optical information transfer in free space through unknown random diffusive media.

ControlNet插件在AI生成图像中的控制应用分析

运用AI绘图工具Stable Diffusion生成图像的过程中,虽然可以用提示词来描述画面,但生成图像的不确定性很大。为了生成满意的图像,可以使用Stable Diffusion的插件ControlNet对图像生成进行控制。本文分析了用ControlNet对图像生成过程的控制方法,并运用该方法在具体案例中实现了在生成图像过程中对画面构图的控制。

AIGC视野下的浙北民居风格研究路径初探

文章以Stable Diffusion为工具,以通过机器学习特定建筑风格图像并输出结果的方式,尝试对浙北民居的建筑风格在AIGC视野下的研究路径进行探索。文章建立了浙北民居与现代住宅图像的数据集并进行训练,得到该语境下的浙北民居风格模型,接着通过批量单独生成与混合生成,对生成的结果进行分类统计。最终发现了该模型在建筑风格模拟方面具有不同于人类直觉判断的抽象提取能力,进而概括了该类工具下生成结果的可能性与局限性。

基于Stable Diffusion的虚拟人形象预设计的应用与研究

对当前AIGC在虚拟形象预设计方面的现状及影响进行了分析和探讨。以Stable Diffusion为例,详细介绍了工程构建和实现,对相关模块的作用、运行环境、使用方法及其指令等多个方面进行了综合叙述、分析和探讨,针对使用不同采样方法、不同采样参数及不同训练模型生成图片效果的优劣进行了说明及展示。随后,通过项目实例,完整地展示了人物形象预设计的过程。最后,对AIGC等新技术可能带来的社会影响进行了预测和总结。

Microstructure and thermal properties of dissimilar M300–CuCr1Zr alloys by multi-material laser-based powder bed fusion

Multi-material laser-based powder bed fusion (PBF-LB) allows manufacturing of parts with 3-dimensional gradient and additional functionality in a single step. This research focuses on the combination of thermally-conductive CuCr1Zr with hard M300 tool steel.Two interface configurations of M300 on CuCr1Zr and CuCr1Zr on M300 were investigated. Ultra-fine grains form at the interface due to the low mutual solubility of Cu and steel. The material mixing zone size is dependent on the configurations and tunable in the range of0.1–0.3 mm by introducing a separate set of parameters for the interface layers. Microcracks and pores mainly occur in the transition zone.Regardless of these defects, the thermal diffusivity of bimetallic parts with 50vol% of CuCr1Zr significantly increases by 70%–150%compared to pure M300. The thermal diffusivity of CuCr1Zr and the hardness of M300 steel can be enhanced simultaneously by applying the aging heat treatment.

从特征辨识到图像生成:基于AIGC范式的苗族服饰设计

苗族服饰是中国民族服饰体系中重要的组成部分,在商品化发展中开辟了广阔的市场,但产品同质化现象将阻碍民族文化的进一步发展,人工智能生成内容(AIGC)的发展为传统服饰设计提供了新的可能。文章基于苗族服饰现代化的需要,通过搭建多模型组合流程,为苗族服饰成衣化设计提供灵感样本。实验结果表明,以服饰辨识性为基础,采用Lora训练方式,仅需少量样本即可实现苗族服饰特征在Stable Diffusion生成内容中的迁移。将深度学习引入非遗服饰现代化的进程中,旨在助力非遗服饰朝着成衣化、多样化、大众化方向发展,以数字化方式推动民族服饰转型升级。

Computational imaging without a computer:seeing through random diffusers at the speed of light

Imaging through diffusers presents a challenging problem with various digital image reconstruction solutions demonstrated to date using computers.Here,we present a computer-free,all-optical image reconstruction method to see through random diffusers at the speed of light.Using deep learning,a set of transmissive diffractive surfaces are trained to all-optically reconstruct images of arbitrary objects that are completely covered by unknown,random phase diffusers.After the training stage,which is a one-time effort,the resulting diffractive surfaces are fabricated and form a passive optical network that is physically positioned between the unknown object and the image plane to all-optically reconstruct the object pattern through an unknown,new phase diffuser.We experimentally demonstrated this concept using coherent THz illumination and all-optically reconstructed objects distorted by unknown,random diffusers,never used during training.Unlike digital methods,all-optical diffractive reconstructions do not require power except for the illumination light.This diffractive solution to see through diffusers can be extended to other wavelengths,and might fuel various applications in biomedical imaging,astronomy,atmospheric sciences,oceanography,security,robotics,autonomous vehicles,among many others.

基于美学梯度法的人工智能风格化绘画系统

当前Stable diffusion等人工智能绘画模型在绘画时难以直接控制图像风格,同时风格模型训练仅针对单种风格。针对该问题,提出了一种基于美学梯度法的人工智能风格化绘画系统,以实现多种图像风格的控制和融合,并提供更加便捷的图像创作体验。收集并分析网络用户数据,结合问卷得到用户对图像风格的感性需求;根据感性需求收集各风格图像数据得到对应的风格图像训练集。使用梯度下降算法计算风格化文本编码器的权重,实现生成图像风格化的效果。通过可用性测试对比用户对该系统与传统人工智能绘画系统产出图像的风格满意程度,结果表明:人工智能风格化绘画系统的平均满意度相较传统人工智能绘画系统提升23%,表明人工智能风格化绘画系统在图像风格生成上具有更好的效果,可满足用户对图像风格的需求。该人工智能风格化绘画系统可以更便捷地实现图像风格调整,允许用户直观选择不同风格的权重,便捷使用一种或多种风格,能够有效满足用户对图像风格设计的需求。

Applications of Additively Manufactured Adjustable Vaned Diffusers in Centrifugal Compressor

As a variable-condition adjustment technology,the adjustable vaned diffusers(AVDs)can expand the working flow range of the compressor in the compressed air energy storage(CAES)system and improve its aerodynamic performance.In order to investigate the regulatory mechanism of AVDs and capture the details of vane loading distribution for the diffuser design optimization,additively manufactured AVDs for testing in a centrifugal compressor closed test facility are designed and implemented.Firstly,the regulation law of AVDs was summarized by numerical analysis and experimental support,and the corresponding vane loading data was extracted for the distribution law.Then,based on the distribution characteristics,3D diffuser models were designed suitably for the adjustable components.Then,the laser selective melting(SLM)technology and die steel material 1.2709 were selected for metal printing according to the actual operating environment.Finally,performance testing and accuracy detection were performed on the finished test pieces,almost all inlet hole’s deviations were within the 0.3 mm tolerance.The research results indicated that additive manufacturing can significantly improve the accessibility of the internal flow channels of the diffuser,and derive the load of the blade on the pressure surface and suction surface in detail,also provide adjustable functions for variable operating conditions.It can not only break through the traditional processing bottleneck of the complicated internal flow channels of AVDs but also improve the design matching degree with adjustable components;simultaneously,it ensures high performance with high precision and effectively shortens the long lead time.

Spatial sensitivity to absorption changes for various near-infrared spectroscopy methods:A compendium review

This compendium review focuses on the spatial distribution of sensitivity to localized absorption changes in optically diffuse media,particularly for measurements relevant to near-infrared spectroscopy.The three temporal domains,continuous wave,frequency domain,and time domain,each obtain different optical data types whose changes may be related to effective homogeneous changes in the absorption coefficient.Sensitivity is the relationship between a localized perturbation and the recovered effective homogeneous absorption change.Therefore,spatial sensitivity maps representing the perturbation location can be generated for the numerous optical data types in the three temporal domains.The review first presents a history of the past 30 years of work investigating this sensitivity in optically diffuse media.These works are experimental and theoretical,presenting one-,two-,and three-dimensional sensitivity maps for different Near-Infrared Spectroscopy methods,domains,and data types.Following this history,we present a compendium of sensitivity maps organized by temporal domain and then data type.This compendium provides a valuable tool to compare the spatial sensitivity of various measurement methods and parameters in one document.Methods for one to generate these maps are provided in Appendix A,including the code.This historical review and comprehensive sensitivity map compendium provides a single source researchers may use to visualize,investigate,compare,and generate sensitivity to localized absorption change maps.

从Make-A-Video到Sora:AI视频生成技术的进步与挑战

随着人工智能技术的飞速发展,AI视频生成技术已成为研究和应用的热点。从Meta的MakeA-Video, Runway AI的Runway Gen-2,Stability AI的Stable Video Diffusion,到Google的Lumiere,再到OpenAI的Sora,每一个模型的推出都不仅代表了AI视频生成技术的进步,也带来了新的挑战。回顾了这些关键AI视频模型的原理和特点,并对比它们之间的优势和不足,探讨了AI视频生成技术面临的主要挑战,展望了未来的发展方向。

Unraveling engineering disturbance effects on deformation in red-bed mudstone railway cuttings:incorporating crack-facilitated moisture diffusion

Red-bed mudstone, prevalent in southwest China, poses a formidable challenge due to its hydrophilic clay minerals, resulting in expansion, deformation, and cracking upon exposure to moisture. This study addresses uplift deformation disasters in high-speed railways by employing a moisture diffusion-deformation-fracture coupling model based on the finite-discrete element method(FDEM). The model integrates the influence of cracks on moisture diffusion. The investigation into various excavation depths reveals a direct correlation between depth and the formation of tensile cracks at the bottom of the railway cutting. These cracks expedite moisture migration, significantly impacting the temporal and spatial evolution of the moisture field. Additionally, crack expansion dominates hygroscopic deformation, with the lateral coordinate of the crack zone determining peak vertical displacement. Furthermore, key factors influencing deformation in railway cuttings, including the swelling factor and initial moisture content at the bottom of the cutting, are explored. The number of tensile and shear cracks increases with greater excavation depth, particularly concerning shear cracks. Higher swelling factors and initial moisture contents result in an increased total number of cracks, predominantly shear cracks. Numerical calculations provide valuable insights, offering a scientific foundation and directional guidance for the precise prevention, control, prediction, and comprehensive treatment of mudstone-related issues in high-speed railways.

3D robust anisotropic diffusion filtering algorithm for sparse view neutron computed tomography 3D image reconstruction

The most critical part of a neutron computed tomography(NCT) system is the image processing algorithm,which directly affects the quality and speed of the reconstructed images.Various types of noise in the system can degrade the quality of the reconstructed images.Therefore,to improve the quality of the reconstructed images of NCT systems,efficient image processing algorithms must be used.The anisotropic diffusion filtering(ADF) algorithm can not only effectively suppress the noise in the projection data,but also preserve the image edge structure information by reducing the diffusion at the image edges.Therefore,we propose the application of the ADF algorithm for NCT image reconstruction.To compare the performance of different algorithms in NCT systems,we reconstructed images using the ordered subset simultaneous algebraic reconstruction technique(OS-SART) algorithm with different regular terms as image processing algorithms.In the iterative reconstruction,we selected two image processing algorithms,the Total Variation and split Bregman solved total variation algorithms,for comparison with the performance of the ADF algorithm.Additionally,the filtered back-projection algorithm was used for comparison with an iterative algorithm.By reconstructing the projection data of the numerical and clock models,we compared and analyzed the effects of each algorithm applied in the NCT system.Based on the reconstruction results,OS-SART-ADF outperformed the other algorithms in terms of denoising,preserving the edge structure,and suppressing artifacts.For example,when the 3D Shepp–Logan was reconstructed at 25 views,the root mean square error of OS-SART-ADF was the smallest among the four iterative algorithms,at only 0.0292.The universal quality index,mean structural similarity,and correlation coefficient of the reconstructed image were the largest among all algorithms,with values of 0.9877,0.9878,and 0.9887,respectively.

Diffusion and reaction mechanism of limestone and quartz in fluxed iron ore pellet roasting process

The increase to the proportion of fluxed pellets in the blast furnace burden is a useful way to reduce the carbon emissions in the ironmaking process.In this study,the interaction between calcium carbonate and iron ore powder and the mineralization mechanism of fluxed iron ore pellet in the roasting process were investigated through diffusion couple experiments.Scanning electron microscopy with energy dispersive spectroscopy was used to study the elements’diffusion and phase transformation during the roasting process.The results indicated that limestone decomposed into calcium oxide,and magnetite was oxidized to hematite at the early stage of preheating.With the increase in roasting temperature,the diffusion rate of Fe and Ca was obviously accelerated,while the diffusion rate of Si was relatively slow.The order of magnitude of interdiffusion coefficient of Fe_(2)O_(3)-CaO diffusion couple was 10^(−10) m^(2)·s^(−1) at a roasting temperature of 1200℃for 9 h.Ca_(2)Fe_(2)O_(5) was the initial product in the Fe_(2)O_(3)-CaO-SiO_(2) diffusion interface,and then Ca_(2)Fe_(2)O_(5) continued to react with Fe_(2)O_(3) to form CaFe_(2)O_(4).With the expansion of the diffusion region,the sillico-ferrite of calcium liquid phase was produced due to the melting of SiO_(2) into CaFe_(2)O_(4),which can strengthen the consolidation of fluxed pellets.Furthermore,andradite would be formed around a small part of quartz particles,which is also conducive to the consolidation of fluxed pellets.In addition,the principle diagram of limestone and quartz diffusion reaction in the process of fluxed pellet roasting was discussed.

Modeling of Leachate Propagation in a Municipal Solid Waste Landfill Foundation

The study deals with the numerical modeling of leachate distribution in the porous medium located under a municipal solid waste disposal landfill(MSWLF).The considered three-layer system is based on geological data obtained from field measurements.For simplicity,the problem is investigated by assuming a two-component approach.Nevertheless,the heat produced by landfills due to biological and chemical processes and the thermal diffusion mechanism contributing to pollution transport are taken into account.The numerical modeling of the propagation of leachate in the considered layered porous medium is implemented for parameters corresponding to natural soil and for the case where disruption of the porous medium structure is due to cracks formation or erosion.The latter is accompanied by an increase in rock permeability.In this case,the emergence of multi-vortex flow in the layer of high permeability is observed to substantially influence the bottom layers and cause intensification of the leachate propagation.These effects should be taken into account when estimating the parameters and properties of materials required for the construction of solid waste disposal facilities.

Impact of Sky Conditions on Net Ecosystem Productivity over a “Floating Blanket” Wetland in Southwest China

Based on eddy covariance(EC) measurements during 2016–20, the effects of sky conditions on the net ecosystem productivity(NEP) over a subtropical “floating blanket ” wetland were investigated. Sky conditions were divided into overcast, cloudy, and sunny conditions. On the half-hourly timescale, the daytime NEP responded more rapidly to the changes in the total photosynthetic active radiation(PARt) under overcast and cloudy skies than that under sunny skies. The increase in the apparent quantum yield under overcast and cloudy conditions was the greatest in spring and the least in summer. Additionally, lower atmospheric vapor pressure deficit(VPD) and moderate air temperature were more conducive to enhancing the apparent quantum yield under cloudy skies. On the daily timescale, NEP and the gross primary production(GPP) were higher under cloudy or sunny conditions than those under overcast conditions across seasons. The daily NEP and GPP during the wet season peaked under cloudy skies. The daily ecosystem light use efficiency(LUE) and water use efficiency(WUE) during the wet season also changed with sky conditions and reached their maximum under overcast and cloudy skies, respectively. The diffuse photosynthetic active radiation(PAR_d) and air temperature were primarily responsible for the variation of daily NEP from half-hourly to monthly timescales, and the direct photosynthetic active radiation(PAR_b) had a secondary effect on NEP. Under sunny conditions, PAR_b and air temperature were the dominant factors controlling daily NEP. While daily NEP was mainly controlled by PAR_d under cloudy and overcast conditions.

Tuning the diffusion constant to optimize the readout of positional information of spatial concentration patterns

Positional information encoded in spatial concentration patterns is crucial for the development of multicellular organisms.However,it is still unclear how such information is affected by the physically dissipative diffusion process.Here we study one-dimensional patterning systems with analytical derivation and numerical simulations.We find that the diffusion constant of the patterning molecules exhibits a nonmonotonic effect on the readout of the positional information from the concentration patterns.Specifically,there exists an optimal diffusion constant that maximizes the positional information.Moreover,we find that the energy dissipation due to the physical diffusion imposes a fundamental upper limit on the positional information.