jimp-dev/jimp

Efficient clone + crop (or contain)

Open

#847 建立於 2020年2月24日

在 GitHub 查看
 (0 留言) (0 反應) (0 負責人)JavaScript (13,218 star) (785 fork)batch import
enhancementhelp wanted

描述

Expected Behavior

Working on a Video Stream at 15fps I need to crop a part of each frame then resize it into a canvas using crop() or contain() that use crop underlaying

Current Behavior

It works but it use about 100ms to process a 720P video.

Failure Information (for bugs)

Not sure if it's a bug or a feature.

Actually I have to perform img.clone().crop() whereas it might be more efficient to add an option to crop() into a new image (the code would gather only the relevant pixels) ?

貢獻者指南

技術棧
javascriptnodejs
領域
backend
議題類型
feature
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
3
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
1-3 hours
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
stale
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
clear
前置要求
Understanding of Jimp's image data structureFamiliarity with Buffer operations in Node.js
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
65
研究方向
Examine the current crop() implementation in Jimp (likely in packages/core/src/image/crop.js). The goal is to add an option to crop() that directly creates a new image containing only the cropped pixels, avoiding the memory and time overhead of cloning the entire image first. Study how Jimp stores pixel data (RGBA buffer) and how the clone() method works to understand the current bottleneck. Then implement a modification that, when an optional parameter (e.g., 'clone: false') is provided, copies only the relevant pixel region into a new Jimp instance. Test performance with a sample 720p video frame.