Skip to content

Releases: Yaro-lu/LingJingAPI

LingJingAPI 2.1.0

Choose a tag to compare

@Yaro-lu Yaro-lu released this 09 Oct 08:27

中文

  • 产品名称统一为 LingJingAPI,更新程序、安装包、桌面入口、托盘及示例页图标。
  • 新安装默认使用 LingJingAPI 目录;升级沿用原安装位置,保留模型、生成结果和用户配置。
  • 新增可选的 DeepSeek、火山方舟 Seedream / Seedance 和即梦 CLI 接口,可在“模型与环境”中配置并通过统一任务队列调用。
  • 将旧版待修复的高清修复工作流替换为 FLUX.2 轻量高清修复,复用已安装的 FLUX.2 Klein 4B 模型。
  • 更新中英文项目介绍、示例页和使用教学,并保留旧文件名的读取兼容。

English

  • Unified the product name as LingJingAPI and updated icons across the app, installer, desktop shortcuts, system tray, and example page.
  • Fresh installations use the LingJingAPI directory. Upgrades retain the existing location, models, outputs, and user configuration.
  • Added optional DeepSeek, Volcengine Ark Seedream / Seedance, and Dreamina CLI integrations, configurable in Models & Environment and served through the shared task queue.
  • Replaced the legacy pending-repair restoration workflow with a lightweight FLUX.2 workflow that reuses installed FLUX.2 Klein 4B models.
  • Updated the bilingual project introduction, example page, and user guide while retaining compatibility with legacy filenames.

2.0.7

Choose a tag to compare

@Yaro-lu Yaro-lu released this 29 Sep 10:17

中文

  • 启动时的自动修复提示现在只检查运行环境和 Qwen3.5 文字模型;仅缺少图片或视频模型时不再弹窗。
  • 打开“模型与环境”或手动重新检查时,仍会检查完整模型状态。一键修复仍可按所选档位补齐图文视频模型。

English

  • The startup repair prompt now checks only the runtime and Qwen3.5 text model. Missing image or video models alone no longer open the dialog.
  • Full model status remains available in Models & Environment and through manual recheck. One-click repair can still install the selected text, image, and video models.

2.0.6

Choose a tag to compare

@Yaro-lu Yaro-lu released this 29 Sep 09:25

中文

  • 修复 FLUX.2 Klein 9B 默认下载源需要登录的问题,改用经大小和 SHA256 校验的公开来源。
  • 大模型改为分段下载;连接中断时保留断点,持续低速时自动切换来源,并显示下载速度。
  • 改进首次环境修复时的自动重试和错误提示。

English

  • Replaced the sign-in-restricted FLUX.2 Klein 9B source with public sources verified by size and SHA256.
  • Large models now download in ranges, resume after interruptions, switch sources on sustained slowdowns, and show transfer speed.
  • Improved automatic retries and error messages in first-run environment repair.

2.0.4

Choose a tag to compare

@Yaro-lu Yaro-lu released this 29 Sep 06:22

中文

  • 移除桌面端的本地/远程模式切换及平台账号登录,不再自动向平台同步连接信息和任务状态。
  • 保留本机、局域网、公网 URL + API Key 调用及环境一键修复。
  • 更新首次安装的截图与快速入门说明。

English

  • Removed desktop local/remote modes and platform account sign-in; connection and task status are no longer synced to a platform.
  • Kept local, LAN and public URL + API Key access and one-click environment repair.
  • Updated the first-time setup screenshot and quick-start guide.

2.0.3

Choose a tag to compare

@Yaro-lu Yaro-lu released this 29 Sep 04:38

中文

  • 新增统一任务队列:按提交时间执行,显示排队位置,支持取消等待任务和强制取消运行中的任务。
  • 首次安装或环境异常时提供一键修复,按显存档位准备文字、图片和视频模型。
  • 界面名称统一为“灵境”;卸载时保留模型和生成结果,并提示保存位置。

English

  • Added a shared FIFO queue with queue positions and cancellation for waiting or running tasks.
  • Added one-click environment and model repair on first install or when checks fail.
  • Updated the interface branding to LingJing; uninstall now preserves models and generated outputs and shows their locations.

2.0.2

Choose a tag to compare

@Yaro-lu Yaro-lu released this 28 Sep 10:02

中文

  • 新增“性价比H3”和“高质量H3”,完善模型依赖、视频分类及参数配置。
  • 模型下载支持自动测速选源、断线重试、换源续传和完整性校验。
  • 支持查找模型来源、选择同名版本、填写地址及复用已确认来源。
  • 图生视频默认跟随参考图比例,保留手动尺寸选择。
  • 修复网页资源加载,并让首次安装和升级均可选择安装路径。

English

  • Added Cost-effective H3 and High-quality H3 workflows with model dependencies, video classification, and parameter mappings.
  • Improved model downloads with source speed checks, retries, resumable failover, and integrity verification.
  • Added model source discovery, version selection, manual URLs, and reuse of verified sources.
  • Image-to-video now defaults to the reference image aspect ratio while preserving manual sizing.
  • Fixed bundled web page loading and enabled installation path selection for both fresh installs and upgrades.

2.0.1

Choose a tag to compare

@Yaro-lu Yaro-lu released this 16 Sep 03:20

更新说明

  • 根据工作流输入和输出识别分类,修复 H3 图生视频被标为文生图的问题。
  • 修复点击“退出并完成安装”及关闭按钮无响应的问题,确保确认后继续环境安装。
  • 修复运行环境安装确认窗口高度不足、退出按钮被遮挡的问题;窗口按内容调整大小,并解除确认阶段的模态锁定,避免多个弹窗叠加时无法操作。
  • 修复 H3 视频工作流尺寸识别,按首帧比例和工作流分辨率设置自动计算尺寸,支持手动指定宽高。
  • 环境包下载自动检测可用线路,失败时切换线路,支持断点续传与完整性校验。
  • 自动下载失败后提供可点击的 123 云盘和百度网盘入口,显示并支持复制提取码,下载后可选择本地环境包安装。

Release notes

  • Classify workflows by their input/output contract, fixing H3 image-to-video workflows incorrectly labeled as text-to-image.
  • Fixed unresponsive exit and close actions in the runtime installation confirmation, allowing installation to continue after confirmation.
  • Fixed the runtime installation confirmation dialog clipping its exit button. The dialog now sizes to its content and releases the modal grab during confirmation to prevent overlapping dialogs from blocking interaction.
  • Fixed dimension detection for H3 video workflows. Output dimensions follow the first-frame aspect ratio and workflow resolution settings, with manual width and height overrides.
  • Added automatic runtime download route selection, failover, resumable downloads, and integrity verification.
  • Added clickable 123 Cloud Drive and Baidu Netdisk fallback links with visible, copyable extraction codes and local runtime package installation.

灵境造片厂 2.0.0 · 局域网/远程调用本地算力

Choose a tag to compare

@Yaro-lu Yaro-lu released this 15 Sep 08:11

灵境造片厂 2.0.0

远程调用算力,生成文字、图片和视频。

  • 租用远程算力:服务器部署服务,自己的电脑和手机通过 URL + API Key 调用。
  • 公司共享算力:一台机器部署,同事通过局域网在各自的电脑或手机上调用。

下载

程序和环境安装在算力端,不包含模型。当前发行包适用于 Windows x64 + NVIDIA 显卡,配套 CUDA 13 环境。

三步使用

  1. 在算力服务器或公司机器上安装程序、环境与模型,启动服务。
  2. 将公网或局域网 URL 和 Key 填入客户端页面,或通过电脑、手机浏览器直接访问。
  3. 输入提示词或上传参考图,调用远程算力生成作品。

客户端创作页面

算力端控制台

工作流与模型管理

后两张为界面演示,连接信息和状态数值为演示数据。

本次更新

  • 电脑、手机直接通过 URL 打开创作页,记住上次连接。
  • 工作流与模型管理合并,支持下载模型和映射已有模型。
  • 项目管理、历史作品分批加载、图片放大与视频首帧预览。
  • 按工作流提供接口文档,完善文字提示词及视频时长、帧率设置。

已知问题:H3 尺寸仍采用工作流内置的 9:16,后续修复。当前按单用户使用;安装包未做代码签名。

项目介绍与更多截图 · Apache License 2.0

灵境造片厂 1.0.3 / LingJing AI Studio 1.0.3

Choose a tag to compare

@Yaro-lu Yaro-lu released this 26 Aug 08:26

中文

灵境造片厂 1.0.3 完善了本机调用体验,并加入客户端更新提示。

本次更新

  • 控制台在“公网 URL”下方新增“本地 API”。同一台电脑上的程序可直接调用回环地址,不依赖公网隧道,更稳定。
  • 公网 URL 与本地 API 上下排列,合计高度与 API Key 卡片对齐。
  • 本地 API 与公网 URL 均可一键复制,并共用控制台显示的同一个 API Key。
  • 左下角版本区域新增更新提示:发现更高版本时显示绿点,悬停可查看提示,点击可打开最新 GitHub Release。
  • GitHub 仓库、下载链接和内置使用教学已更新为 LingJingAPI / 1.0.3。
  • 本地 API 地址只接受 HTTP 回环地址及有效端口,避免复制到无效地址。

安装与升级

  1. 下载 LingJingAI-Setup-1.0.3-win-x64.exe。
  2. 同机调用优先复制“本地 API”;其他设备调用复制“公网 URL”。
  3. 两种地址继续使用控制台显示的同一个 API Key。
  4. 升级轻量客户端不会要求重新下载运行环境、模型或生成文件。

更新提示功能从 1.0.3 开始提供,因此安装 1.0.3 后,会从后续更高版本开始显示更新绿点。

安装包当前未进行 Authenticode 代码签名,Windows 可能显示 SmartScreen 提示。请只从本 Release 下载,并核对 SHA-256。

安装包 SHA-256:

f88d662ac4770048ae8961c05e21771af79a6da6d3f520a4d66a508bc6d56960

English

LingJing AI Studio 1.0.3 improves local access and adds an in-app update notice.

What's new

  • Added a Local API card below the Public URL. Apps on the same PC can use the loopback address without relying on the public tunnel.
  • The Public URL and Local API cards now align with the API Key card.
  • Both URLs can be copied with one click and use the same API key.
  • A green update dot appears beside the desktop version when a newer release is available. Hover for details and click to open the latest GitHub Release.
  • Repository links, downloads, and the bundled user guide now point to LingJingAPI 1.0.3.
  • Local API validation is restricted to valid HTTP loopback origins.

The update indicator is introduced in 1.0.3 and will notify users about later versions. The installer is currently unsigned, so Windows SmartScreen may appear; download only from this Release and verify the SHA-256 above.

灵境造片厂 1.0.2 / LingJing AI Studio 1.0.2

Choose a tag to compare

@Yaro-lu Yaro-lu released this 24 Jul 06:33

中文

灵境造片厂把本机 ComfyUI 工作流整理成统一的 URL + API Key 调用方式,可供随附示例页、网页或其他程序调用。

v1.0.2 更新

  • 修复 API、公网连接短暂波动后状态无法自动恢复的问题。
  • 改进运行环境安装与更新流程,提供明确的退出反馈、失败恢复和隐藏后台处理。
  • 更新使用教学、发行校验和相关稳定性测试。

使用

  1. 安装 LingJingAI-Setup-1.0.2-win-x64.exe 并启动灵境造片厂。
  2. 首次使用时,在“模型与环境”中修复/安装运行环境,再按提示下载所需模型。
  3. 在控制台复制 URL 与 API Key;可直接打开随客户端安装的示例页进行文字、图片和视频调用。

接口

  • GET /v1/models:查看可用模型
  • GET /v1/workflows:查看可用工作流
  • POST /v1/workflows/run/{workflow_id}:提交任务
  • GET /v1/tasks/{task_id}:查询进度与结果
  • 业务接口使用 Authorization: Bearer <API_KEY>

安装包 SHA-256:

ac522e67a3b5ce0582155823746489aff3ed06612a436d8dd9677d099f532b08

English

LingJing AI Studio exposes local ComfyUI workflows through a consistent URL + API Key interface for the bundled example page, web apps, or other software.

v1.0.2 changes

  • Fixed status recovery after temporary API or public-tunnel interruptions.
  • Improved runtime installation and update handoff with clear exit feedback, failure recovery, and hidden background processing.
  • Updated the user guide, release validation, and reliability tests.

Usage

  1. Install LingJingAI-Setup-1.0.2-win-x64.exe and launch LingJing AI Studio.
  2. On first use, repair/install the runtime under “Models & Runtime”, then download the models required by your workflow.
  3. Copy the URL and API Key from the dashboard. The bundled example page supports text, image, and video calls.

API

  • GET /v1/models: list models
  • GET /v1/workflows: list workflows
  • POST /v1/workflows/run/{workflow_id}: submit a task
  • GET /v1/tasks/{task_id}: query progress and results
  • Use Authorization: Bearer <API_KEY> for protected endpoints.

Installer SHA-256:

ac522e67a3b5ce0582155823746489aff3ed06612a436d8dd9677d099f532b08