Unreal Open Day 2017 UE4 for Mobile: The Future of High Quality Mobile GamesEpic Games China
This document summarizes a presentation about Unreal Engine 4 for mobile game development. It discusses UE4's mobile rendering pipeline and features for high-end graphics on mobile, including OpenGL ES 3.1, Vulkan and Metal. It provides an overview of the state of the mobile game market and examples of AAA open-world games made with UE4. It also outlines UE4's feature levels for mobile, describes the components of the mobile rendering pipeline, and highlights specific rendering techniques like HDR encoding.
The Real-time Volumetric Cloudscapes of Horizon Zero DawnGuerrilla
The document provides details on the cloud system developed for the video game Horizon Zero Dawn. Key points:
- Clouds were procedurally generated using noise textures and gradients to define shapes rather than pre-made assets, allowing them to evolve over time as part of the weather system.
- Cloud modeling involved sampling noises to build density and add detail, with different noises used to define cloud types and add turbulence.
- A weather system controlled cloud coverage, type, and precipitation to drive cloud evolution based on location. It was overridden for cutscenes with custom textures.
- Lighting approximated key effects like scattering and cloud edges to give a realistic look within performance constraints.
The document discusses advanced rendering techniques for achieving anime-style rendering in Unity. It covers topics such as advanced rendering features for mobile like bloom and dynamic particles. It also discusses stylized scene lighting, specialized shaders for materials like silk and hair, character shading techniques including multi-ramp shading and facial expression blending, and effects like volume lights, depth of field, and image-space outlines. The overall goal is to achieve high quality anime-style rendering on mobile and PC platforms.
https://coloso.co.kr/game/gamegraphic_yijuyeong 섭디의 모든 것은 여기서.
베이크는 경험과 시행착오로 하는 것이 아닙니다. 정해진 법칙이 있고, 그 법칙을 따르면 쉽게 성공할 수 있습니다.
오타
43번 슬라이드 왼쪽 아래 사진의 캡션은 'PolyBevel 직후'가 아니라 'PolyBevel 전' 입니다.
The Real-time Volumetric Cloudscapes of Horizon Zero DawnGuerrilla
The document provides details on the cloud system developed for the video game Horizon Zero Dawn. Key points:
- Clouds were procedurally generated using noise textures and gradients to define shapes rather than pre-made assets, allowing them to evolve over time as part of the weather system.
- Cloud modeling involved sampling noises to build density and add detail, with different noises used to define cloud types and add turbulence.
- A weather system controlled cloud coverage, type, and precipitation to drive cloud evolution based on location. It was overridden for cutscenes with custom textures.
- Lighting approximated key effects like scattering and cloud edges to give a realistic look within performance constraints.
The document discusses advanced rendering techniques for achieving anime-style rendering in Unity. It covers topics such as advanced rendering features for mobile like bloom and dynamic particles. It also discusses stylized scene lighting, specialized shaders for materials like silk and hair, character shading techniques including multi-ramp shading and facial expression blending, and effects like volume lights, depth of field, and image-space outlines. The overall goal is to achieve high quality anime-style rendering on mobile and PC platforms.
https://coloso.co.kr/game/gamegraphic_yijuyeong 섭디의 모든 것은 여기서.
베이크는 경험과 시행착오로 하는 것이 아닙니다. 정해진 법칙이 있고, 그 법칙을 따르면 쉽게 성공할 수 있습니다.
오타
43번 슬라이드 왼쪽 아래 사진의 캡션은 'PolyBevel 직후'가 아니라 'PolyBevel 전' 입니다.
論文紹介:「Amodal Completion via Progressive Mixed Context Diffusion」「Amodal Insta...Toru Tamaki
Katherine Xu, Lingzhi Zhang, Jianbo Shi; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, "Amodal Completion via Progressive Mixed Context Diffusion"CVPR2024
https://openaccess.thecvf.com/content/CVPR2024/html/Xu_Amodal_Completion_via_Progressive_Mixed_Context_Diffusion_CVPR_2024_paper.html
Minh Tran, Khoa Vo, Tri Nguyen, and Ngan Le,"Amodal Instance Segmentation with Diffusion Shape Prior Estimation"ACCV 2024
https://uark-aicv.github.io/AISDiff/
This study aims to develop an interactive idea-generation support system that enables users to consider the potential side effects of realizing new ideas.
In idea generation, confirmation bias often leads to an excessive focus on ``convenience,'' which can result in the oversight of unintended consequences, referred to as the ``side effects of convenience.''
To address this, we explored methods to alleviate user biases and expand perspectives through system-supported dialogue, facilitating a broader consideration of potential side effects.
The proposed system employs a stepwise idea-generation process supported by large language models (LLMs), enabling users to refine their ideas interactively.
By dividing the ideation process into distinct stages, the system mitigates biases at each stage while promoting ideas' concretization and identifying side effects through visually supported dialogues.
Preliminary evaluation suggests that engaging with the proposed system fosters awareness of diverse perspectives on potential side effects and facilitates the generation of ideas that proactively address these issues.