Skip to content

Make videos on the computer you already own. FreeVideo runs MiniMax H3 in as little as 8 GB of VRAM and 16 GB of RAM, and adapts its acceleration path to your hardware.

66.9 KB
Raw Blame
{
  "schema": 1,
  "product_version": "0.4.35",
  "en": {
    "summary": "External Agents can join the best part of every cut from different takes, the record button stays in place, and text and subtitles can be added after the sound has been separated.",
    "changes": [
      "External Agents: when the takes of a shot have the same cuts, an Agent can read each take's cuts, per-part checks and how the takes join at every cut, then join the best part of every cut from different takes. The choice is kept with the film.",
      "External Agents: when a voice is changed with a script, a sentence split by a pause is said once instead of twice, and numbers, hyphenated words and initials stay whole.",
      "Editor: the record button stays in the toolbar and turns grey when the selection cannot take a recording, so the icons after it no longer move.",
      "Editor: audio track names fit in the track head, and text and subtitles can be added after the sound has been separated.",
      "Tutorial and feature tour: new cards and chapters for Describe each part and Reuse settings. The example button reads \"Use example\" in both."
    ]
  },
  "zh": {
    "summary": "外部 Agent 可从不同版本中选取每个切镜最好的一段并拼接,录音按钮位置固定,声音分离后可以添加文字和字幕。",
    "changes": [
      "外部 Agent:同一镜头的各版本切镜相同时,Agent 可读取每一版的切镜、各段检查结果和各切镜处的衔接情况,从不同版本中为每个切镜选取最好的一段并拼接;所选片段随影片保存。",
      "外部 Agent:传入台词换人声时,被停顿切开的一句话只说一次;数字、连字符词和缩写保持完整。",
      "剪辑:录音按钮固定在工具栏中,选区不能录音时变灰,其后的图标不再移动。",
      "剪辑:声音轨名称在轨头内完整显示,声音分离后可以添加文字和字幕。",
      "教程和功能导览:新增“逐段描述”和“复用设置”的卡片和章节;示例按钮统一为“填入示例”。"
    ]
  },
  "history": [
    {
      "product_version": "0.4.34",
      "en": {
        "summary": "Agent can finish the sound of a whole editor project in one job, a line of a clip can be respoken in the original voice, and the Edit tab can save the current frame as a picture.",
        "changes": [
          "Agent: a new job finishes the sound of an editor project in one step. It separates each clip's sound, removes sustained music that came with the clips while keeping effects such as gunshots and crowds, lays one continuous music track across the cut, evens out loudness and fills silent gaps with room tone.",
          "Edit only some of the lines in a clip: those lines are respoken in the original speaker's voice at the original timing, and the rest stay as they were. Dubbed lines keep the original pace and loudness. The dubbing model (about 4.9 GB) is downloaded on first use.",
          "A reference video marked as voice only is now read as an audio reference: only its sound track is used and no picture is encoded. A voice reference used across several shots is no longer encoded again for each one.",
          "Agent: take checks listen for the scripted lines (Chinese and English), notice a single odd frame at the start or end and a picture that stops before the take ends, and do not pick such takes. The Agent skill text also submits storyboards as one batch and resumes a retryable failure once.",
          "Short dramas: long shots submitted right after startup are no longer refused while this computer's generation limit is still being worked out.",
          "The Edit tab can save the frame under the playhead as a picture. It is drawn as the export would draw it and kept under Edit image > Generate with AI > History.",
          "Continued parts in the editor can now use Preview first, and Run second pass replaces the preview in place.",
          "The preview shown while a video is generated plays as a loop of up to sixteen frames instead of a still sheet of four, and no longer flashes black when its stage changes size. Recent videos and Creations fold videos with the same prompt together only when they were made within two hours of each other from the same references.",
          "Reference video encoding now shows progress, and the time estimate counts the reference video from the start.",
          "When a request is refused for lack of GPU memory, the suggested size is now the largest smaller one that fits, instead of one of only two fixed smaller sizes.",
          "Reference-video frames are removed after encoding and stopped requests free their leftovers, so a generation with reference videos no longer leaves over 1 GB per reference in its folder. Changing the prepared model also removes the online LoRA folders that pinned the old model, and unused ones are cleaned up automatically.",
          "Launcher and Settings: the Restart ComfyUI step and the wrong-folder message no longer offer to report a problem, the wrong-folder message points to Install ComfyUI when none is found nearby, and the placement test button is disabled while an installation check or installation is running."
        ]
      },
      "zh": {
        "summary": "Agent 可以一步完成整个剪辑工程的声音;可以只改一段里的几句台词,用原说话人的声音重说;编辑页可以把当前帧存成图片。",
        "changes": [
          "Agent:新增一个任务,一步完成剪辑工程的全片声音:分离各段音轨,去掉持续的旧配乐(枪声、人群声等效果声保留),铺一条贯穿全片的配乐,统一全片响度,并在静音处补上环境声。",
          "可以只改其中几句台词:改过的句子用原说话人的声音重说,时间位置不变,其余原声不动。配音的语速与响度保持与原句一致。配音模型约 4.9 GB,第一次使用时下载。",
          "标为只用声音的参考视频,现在按音频参考读取:只取其声音,不再编码画面。同一段声音参考用于多个镜头时,不再重复编码。",
          "Agent:镜头质检会听出每条候选是否说出了剧本台词(中英文均可),并发现开头或结尾单帧闪烁和画面提前停住,有这些问题的候选不会被选用。Agent 技能文本也改为把分镜作为一批提交,可重试的失败只重试一次。",
          "短剧:在服务刚启动时提交长镜头,不再因本机的生成能力尚未算完而被拒绝。",
          "编辑页可以把播放头所在的一帧存成图片,画面与导出一致,图片放在“编辑图片 > AI 生成 > 历史”里。",
          "剪辑页的续写段现在可以使用“先出预览”,看过后用“继续二采”原位生成正式版。",
          "生成中的预览改为循环播放最多 16 帧,取代原来 4 帧的静态拼图,画面尺寸变化时也不再闪黑。最近视频和作品库中,提示词相同的视频只有在相隔不超过两小时、且参考内容相同时才折叠在一起。",
          "参考视频编码现在会显示进度,预计时间从一开始就把参考视频算进去。",
          "因显存不足而拒绝请求时,建议的尺寸改为能放行的最大一档,不再只在两个固定的较小尺寸里选。",
          "视频参考的帧文件在编码后即删除,取消或失败的请求也会释放残留文件,带视频参考的生成不再在文件夹里留下每个参考超过 1 GB 的文件。更换模型后会一并清理占住旧模型的在线 LoRA 文件夹,不再使用的也会自动清理。",
          "启动器和设置:“需要重启 ComfyUI”这一步和“这不是 ComfyUI 文件夹”的提示不再显示“提交问题”;后者在附近找不到 ComfyUI 时会提示选择“帮我安装 ComfyUI”;安装检查或安装进行时,放置测试按钮置灰。"
        ]
      }
    },
    {
      "product_version": "0.4.33",
      "en": {
        "summary": "Prompt enhancement works with less free graphics memory, and Setup or Repair now waits for a video that is still generating.",
        "changes": [
          "Prompts no longer receive the added sentence about natural sounds that 0.4.32 introduced.",
          "Audio references: an mp4 file that only contains sound is accepted. An Agent that needs only the voice of a video can send it at a small picture size, which makes generation much faster.",
          "Agent: choosing a take cancels the takes of that shot that are still queued or running.",
          "Agent: when a voice reference is uploaded, the first six seconds are checked for speech, and the Agent is told when there is too little, such as in a clip that opens with music, so another clip can be chosen.",
          "Setup or Repair that is stopped by a video still generating now waits for it and continues by itself, instead of showing a general failure.",
          "Prompt enhancement: rewriting now works with less free graphics memory. When other programs hold the memory it needs, it waits and starts by itself once there is room, and lists those programs when their names can be read.",
          "Audio processing in videos no longer differs depending on when FreeVideo was installed.",
          "With several graphics cards, a new generation waits when its card is still briefly held by the previous request, instead of failing within a few seconds.",
          "The Autopilot run bar stays fully visible in the top bar on narrow windows.",
          "Settings: on narrow screens, the three Getting started buttons no longer break their labels across lines."
        ]
      },
      "zh": {
        "summary": "提示词增强在可用显存较少时也能改写;安装或修复遇到还在生成的视频时会等待并自动继续。",
        "changes": [
          "提示词不再被补写 0.4.32 加入的自然声音描述。",
          "音频参考:只有声音的 mp4 文件也可以使用。Agent 只需要视频里的声音时,可用小画面传入,生成明显更快。",
          "Agent:选定一版后,同一镜头还在排队或生成的其他版本自动撤下。",
          "Agent:上传声音参考时检查前 6 秒的说话时长,说话太少(例如以音乐开头)时告知 Agent,便于换一段。",
          "安装或修复被还在生成的视频拦下时,启动器现在会等视频结束后自动继续,不再显示通用的失败提示。",
          "提示词增强:可用显存较少时也能改写;显存被其他程序占用时会等待,空出后自动改写,能读取到程序名时一并列出。",
          "视频声音的处理结果不再因安装时间不同而有差异。",
          "多张显卡时,新的生成遇到显卡还被上一个请求短暂占用会等待,不再几秒内失败。",
          "AI 全自动运行时,窄窗口顶栏的运行条完整显示。",
          "设置:窄屏上“新手指引”的三个按钮不再折行断字。"
        ]
      }
    },
    {
      "product_version": "0.4.32",
      "en": {
        "summary": "A picture can be generated for a span of imported sound, and the top bar has a permanent Tutorial entry.",
        "changes": [
          "Sound-led video: drag a span over imported sound in the Edit view, describe the picture, and generate a video for that span while the original sound is kept.",
          "Tutorial: the top bar has a permanent Tutorial entry. Three new feature examples cover making a first video, characters and the Edit view. These and some tutorial chapters have \"Take me there\", which highlights the matching control.",
          "Audio descriptions: when a description asks to leave out music or other sounds, that request is removed before generation and, where needed, a sentence is added saying the sound comes from the action and the surroundings, so unwanted background music appears less often. Descriptions that ask for music or silence, and quoted lines, are unchanged.",
          "Two GPUs: a machine with two GPUs now decides whether tasks can run together from the memory each task actually uses. On Linux, two 16 GB cards with 64 GB of RAM can run two tasks at once. Single-GPU machines are unaffected.",
          "Editor: deleting an edit no longer keeps failing after a brief loss of contact with the background service.",
          "Editor import: a video that is cut off or damaged part-way now keeps the part that can be read instead of failing the whole import.",
          "Reference audio: generating with reference audio works again on Mac, and no longer fails on other systems when the model has to be loaded first."
        ]
      },
      "zh": {
        "summary": "可以为导入声音的一段生成画面;顶栏有常驻的教程入口。",
        "changes": [
          "声音带动画面:在剪辑视图里,在导入的声音上拖出一段,写上画面描述,为这一段生成视频,原声保留。",
          "教程:顶栏有常驻的“教程”入口。功能示例新增三张:做出第一条视频、角色、剪辑视图。这三张和部分教程章节有“带我去”,点后高亮对应的位置。",
          "声音描述:写了不要音乐或某种声音时,生成前会去掉这类说法,必要时补一句说明声音来自动作和环境,减少无意带出的背景音乐。要求音乐或静音的描述、引号里的台词不改。",
          "双显卡:有两张显卡的机器按每个任务的实际内存用量判断能否同时运行。在 Linux 上,两张 16 GB 显卡配 64 GB 内存可以同时运行两个任务。单卡机器不受影响。",
          "剪辑:与后台的通信短暂中断后,删除剪辑不再连续失败。",
          "剪辑导入:中途被截断或损坏的视频,现在会保留可读取的部分,不再整个导入失败。",
          "参考音频:使用参考音频生成在 Mac 上恢复可用;其他系统需要先载入模型时也不再失败。"
        ]
      }
    },
    {
      "product_version": "0.4.31",
      "en": {
        "summary": "The Edit view can now record audio against the picture, a sound clip can be separated into dialogue, music and ambience tracks, and timed lines and cuts written in a description are placed at those times in the generated video.",
        "changes": [
          "Recording: the Edit view can now record audio against the picture. The recording is placed on a sound track, aligned with the picture.",
          "Separate tracks: a sound clip can be split into dialogue, music and ambience tracks, each adjusted and exported on its own. Separated music can be replaced from the generation bar with \"Replace music\": new music is made at the original length and lowered where people speak.",
          "Timed lines and cuts: write lines such as \"0:02 Jay: ...\" or \"0:05 Cut: ...\" in the description, and the generated video places dialogue and shot changes at those times. A time beyond the video length or an empty cut is flagged before generation.",
          "Preview first: the editor's generation bar can make a preview first. A selected preview part can run its second pass with \"Run second pass\" and is replaced in place at full quality. \"Reuse settings\" also fills in the original seed.",
          "Ripple delete: deleting a clip in the editor also closes the gap it leaves.",
          "Recent videos and Creations fold the videos of one task into a group. Click the cover to open it.",
          "Progress estimate: the estimated total time of a request with a long reference video is no longer far above the real time. Speed and results are unchanged.",
          "Start-up: the launcher now opens when the bridge folder is redirected. A failed model upgrade keeps its cause in the diagnostic report and points to \"Try again\" or \"Report a problem\". The launch progress names the GPU detection, acceleration test and workflow-node connection steps, a slow full log export shows how many files have been packed, and a failed start in the terminal points to the diagnose command.",
          "Install: on some Linux systems, setup no longer stops while building GPU acceleration.",
          "Mac: a fault while encoding a reference video is corrected or reported within seconds instead of at the end of encoding. When a Light or Medium text-to-video generation without references or continuation is interrupted during the first pass, the automatic retry continues from the last completed step instead of starting the pass over.",
          "Speed: on 6 GB-class cards on Windows, each sampling step of a High generation with references is faster.",
          "Audio cards: \"Use as is\" now stays clickable when Preview first is on or a video is being continued. Choosing it turns the preview off with a one-line note, and a continuation says why it cannot be used and what to remove."
        ]
      },
      "zh": {
        "summary": "剪辑视图可以对着画面录音;一段声音可以分成人声、音乐和环境声三条轨道;在描述里按时间写的台词和切镜,生成的视频会按写定的时间安排。",
        "changes": [
          "录音:在剪辑视图里对着画面录一段声音,录好后放到声音轨上,与画面对齐。",
          "分离音轨:把一段声音分成人声、音乐和环境声三条轨道,每条都可单独调整和导出。分离出的音乐可在生成条点“替换配乐”,按原长度重新配乐,并在说话处压低。",
          "按时间写台词和切镜:在描述里写“0:02 阿杰:……”或“0:05 切:……”,生成的视频会按写定的时间安排台词和镜头切换;时间超出视频长度或“切:”后面没有内容时,生成前给出提示。",
          "先出预览:剪辑视图生成条可选“先出预览”;选中预览片段可点“继续二采”,在原位置换成完整画质。点“复用设置”时,原来的种子也一并填入。",
          "波纹删除:在剪辑视图删除片段时,同时合上它留下的空隙。",
          "最近作品和作品库将同一任务的多条视频折叠成一组,点封面展开。",
          "进度估计:带较长参考视频的请求,估计的总时间不再远高于实际用时。速度和结果不变。",
          "启动:启动器在桥接目录被重定向时也能打开。模型升级失败时,诊断报告保留失败原因,并提示可点“重试”或“提交问题”。启动进度会写出检测显卡、测试加速和连接工作流节点这几步;导出完整日志较慢时显示已打包的文件数;在终端里启动失败时,会提示运行诊断命令。",
          "安装:部分 Linux 系统上,安装不再停在构建 GPU 加速这一步。",
          "Mac:参考视频编码出错时,几秒内即修正或报告,不再等到编码结束。轻量或标准档、不带参考、不续接的文生视频在第一遍被中断后,自动重试会从最后完成的步骤继续,不再从头开始这一遍。",
          "速度:Windows 上的 6 GB 级显卡,带参考的精细档生成每一步采样更快。",
          "音频卡:“先出预览”开启或接着生成时,“原样使用”不再变灰。选择它会关闭预览并用一行说明;接着生成时说明原因和需要移除的内容。"
        ]
      }
    },
    {
      "product_version": "0.4.30",
      "en": {
        "summary": "Image generation on Windows machines with little memory no longer stops while the model loads, large requests are steadier in the second pass, and the editor, the launcher and phone access now explain what stopped and what to do next.",
        "changes": [
          "Image generation: on Windows machines with little free memory, loading the image model no longer crashes. If loading is interrupted anyway, FreeVideo retries the request once automatically.",
          "Generation: large requests on Windows are less likely to run out of memory in the second pass, and when they do, the retry is aimed at the memory that actually ran short instead of falling back to much slower settings. A generation also no longer stops when a memory copy is refused during sampling.",
          "Creation panel: a note appears when other programs are using enough GPU memory or RAM to slow a generation down, and a generation that fails because of that shortage says so.",
          "Prompt enhancement: a prompt over the 12,000-character limit now shows its current length, counting the descriptions of characters mentioned with @, before anything is sent. When the prompt or references change while a rewrite is running, a plain note says so instead of a red error. Time points later than the end of the video no longer cause an error; they are fitted within the video's length.",
          "Editor: playback stops with a notice at a clip that cannot be played, instead of running on over a black picture. After FreeVideo restarts, interrupted exports, multi-part generations and subtitle jobs no longer show as in progress, and can be started again or continued.",
          "Phone access: when phone access cannot be opened, the page says why and offers \"Try again\", instead of showing \"Opening...\" indefinitely.",
          "Start-up: the launcher's failure cards now explain the cause and the next step. An access-denied error is titled \"Access denied\" and shows where access was denied; a port in use offers \"Change address\"; a refused setup step, an expired diagnostic report and a failed report export also show the next step. When Python fails to install during setup, the card shows the installer's own error message instead of only an exit code. Exporting a report can replace an existing file.",
          "Install: while a downloaded image model file is being verified, its card says \"Checking\". On Windows, entry folders left in the shared FV folder by removed installations are cleaned up. If the setup tool has gone missing before an install step, it is downloaded again instead of the step failing.",
          "Multi-GPU machines: requests from the Creation panel and the Agent that include images, continue a clip or are longer now run on a free card alongside other generations. A single very large task runs on its own instead of waiting in the queue indefinitely.",
          "GPU memory: very long second passes, such as ones with large reference videos, need less GPU memory on smaller cards. When other programs take GPU memory during a run, decoding adapts instead of failing and starting over. Results are unchanged."
        ]
      },
      "zh": {
        "summary": "内存较小的 Windows 机器生成图片时,不再在加载模型时中断;大画面请求的第二遍更稳定;剪辑视图、启动器和手机访问出问题时,会说明原因和下一步。",
        "changes": [
          "图像生成:可用内存较小的 Windows 机器加载图像模型时不再崩溃。加载仍意外中断时,FreeVideo 会自动重试一次。",
          "生成:Windows 上的大画面请求在第二遍更不容易出现内存不足;出现时,重试会针对实际不足的那部分内存,不再退到慢得多的设置。采样中途内存拷贝被拒绝时,生成也不再中止。",
          "创作面板:其他程序占用的显存或内存会使生成变慢时,面板会提示;因此类不足导致生成失败时,失败卡片同样说明这一点。",
          "提示词增强:提示词超过 12,000 字符上限时,发送前会显示当前字数,展开的 @角色描述一并计入。改写过程中原文或素材有改动时,改为普通提示,不再显示红色错误。提示词里晚于视频结尾的时间点不再报错,会调整到视频时长以内。",
          "剪辑视图:播放到无法播放的素材时会停下并提示,不再黑屏空走。FreeVideo 重启后,中断的导出、分段生成和字幕任务不再显示为进行中,可以重新开始或继续。",
          "手机访问:手机连接无法打开时,网页会说明原因并提供“重试”,不再一直显示“正在打开…”。",
          "启动:启动器的失败卡片会说明原因和下一步。拒绝访问类错误的标题为“访问被拒绝”,并写出被拒绝的位置;端口被占用时提供“更改地址”;安装步骤被拒绝、诊断报告过期和导出报告失败时,也会给出下一步。安装过程中 Python 安装失败时,失败卡片会显示安装器的报错,而不只是退出码。导出报告时可以覆盖已有文件。",
          "安装:下载完成的图像模型文件在校验时,卡片显示“正在校验”。Windows 上,已删除的安装在共用 FV 目录里留下的入口目录会被清理。安装步骤开始前如果安装工具缺失,会重新下载,安装步骤不再因此失败。",
          "多显卡机器:创作面板和 Agent 发出的带图、续接或较长的请求,现在可以在空闲的显卡上与其他生成同时运行。单个很大的任务改为单独运行,不再一直排队。",
          "显存:较长的第二遍(例如参考视频很大的请求)在小显存显卡上占用的显存更少。运行中其他程序占用显存时,解码会自行调整,不再失败后重新开始。结果不变。"
        ]
      }
    },
    {
      "product_version": "0.4.29",
      "en": {
        "summary": "Music can be generated on the second audio track from the generation bar (not on Mac), scenes can be added to the character library, long videos can be extended by typing a length in seconds, and Autopilot chat continues a failed or stopped video from where it left off.",
        "changes": [
          "Editor: music can be made from the generation bar by dragging a range on the second audio track, which new edits now show by default. Choose a style and a mood and optionally describe the music; if the music model is not installed yet, it is downloaded first. A selected generated piece can be regenerated in place, and one undo restores the previous piece. Generated music is lowered while someone speaks; this can be turned off with Lower during speech in the clip properties. Not available on Mac.",
          "Editor: the timeline scrolls vertically when the second audio track, text or subtitle rows do not fit, with the ruler kept in place.",
          "Character library: scenes can be added from your own pictures or from several AI candidates, and mentioned with @ in the prompt; the time of day, weather, light and tone written in the prompt are kept in the generated video.",
          "Long videos: when an ungenerated draft part or a range on the picture track is selected, the generation bar shows a Duration field in whole seconds. A typed length follows the same rules as dragging the right edge; a value out of range is clamped with a short note, and one undo restores the previous length. When a video is extended, the last continuation part, including a single added part, can be as short as 2 s; parts in the middle still last at least 5 s.",
          "Long videos: joining parts no longer fails at the last step when one part differs from the first in size or encoding; such a part is now matched to the first part before the parts are joined, and when that is not possible, the message names the part and the property that differs. A video that already has the most parts a continuous shot can have now says so as soon as it is chosen to be extended, and the limit is checked before generation starts.",
          "Autopilot: Continue generating on a failed or stopped video picks up the same request and keeps the parts already made, also when the Agent is offline. A failed picture is submitted again with the original request. The latest message can be edited and sent again; unfinished tasks after it are stopped first.",
          "Agent: when the Connect an agent card cannot get an address, it explains the reason, for example when the page was opened from another device or by machine name; the feedback report includes the status code and error code.",
          "Prompts: when the Agent is asked to improve the prompt before generating, the generation uses the improved text exactly, and is no longer treated as a different prompt because quotation marks were changed while the text was copied. Asking for no background music in the scene text now works when the prompt still has the starter prompt's music section: the unchanged starter score is replaced with N/A, and a rewritten prompt ends with N/A. A music section written by the user is kept.",
          "Speed: the background compile also prepares portrait videos and videos with a reference picture, so their first request no longer compiles GPU programs at the first step of each pass. The one-time background compile takes about 80 seconds longer; results are unchanged.",
          "Install: a damaged offline package is now reported by name with a request to download it again, instead of a raw decompression error, and downloads over an unstable connection continue from the part already downloaded instead of stopping after the first round.",
          "Install: on Windows, when another program, such as security software, holds FreeVideo's Python file open, setup now waits and creates the Python environment again instead of failing at the first attempt. When Windows refuses to run that file, the failure message says so in any system language and points to the security software.",
          "Start-up: a downloaded source update is verified against the release signature again every time the launcher starts, not only when it is downloaded, and the diagnostic log times each start-up step of the video worker."
        ]
      },
      "zh": {
        "summary": "可以在第二条声音轨上用生成栏生成配乐(暂不支持 Mac);角色库可以新建场景;长视频可以直接填写秒数来接着生成;AI 全自动对话中失败或停止的视频可以从中断处继续生成。",
        "changes": [
          "剪辑视图:在第二条声音轨上拖出一段,即可用生成栏生成配乐;新剪辑默认显示这条声音轨。可选择风格和情绪,也可另写描述;配乐模型未安装时会先下载。选中已生成的配乐可以原位重新生成,一次撤销即可恢复原来那段。生成的配乐在有人说话处自动降低音量,可在片段属性中取消“对白时降低音量”。暂不支持 Mac。",
          "剪辑视图:第二条声音轨、文字或字幕行放不下时,时间线可以纵向滚动,标尺保持固定。",
          "角色库:可以用自己的图片或从 AI 生成的多张候选中新建场景,并在提示词中用 @ 引用;提示词里写的时间、天气、光线和色调会保留到生成的视频中。",
          "长视频:选中未生成的草稿段或画面轨上的一段范围时,生成栏显示以整秒为单位的“时长”输入框。填写的时长与拖动右边缘规则相同;超出范围时自动取到边界并给出简短提示,一次撤销即可恢复原来的时长。接着生成时,最后一个接续段(包括只补一段的情况)最短可为 2 秒,中间的接续段仍至少 5 秒。",
          "长视频:某一段的尺寸或编码与第 1 段不同时,不再在最后一步接合失败:接合前会先把这样的段转换为与第 1 段一致;无法转换时,提示会指明是哪一段、哪一项不同。已达到连续镜头最多段数的视频,选中接着生成时就会说明原因,生成前也会检查。",
          "AI 全自动:视频失败或被停止后,点“继续生成”会按原请求接着做,保留已完成的段,Agent 不在线时同样可用。图片失败时按原请求重新提交。最后一条消息可以修改后重新发送,其后未完成的任务会先停止。",
          "Agent:“连接 Agent”卡片取不到地址时,会说明原因(例如从其他设备或用机器名打开页面);反馈报告附带状态码和错误码。",
          "提示词:让 Agent 先润色提示词再生成时,生成使用的就是润色后的原文,不会因抄写时引号被改动而被当作另一段提示词。在场景文字里要求不要背景音乐,而提示词仍带着默认示例的配乐段时,这一要求现在会生效:没改过的默认配乐会被换成 N/A,提示词重写的结果也以 N/A 结尾。用户自己写的配乐段保持不变。",
          "速度:后台预编译也准备竖向视频和带参考图的视频,它们的第一次请求不再在每遍的第一步现场编译 GPU 程序。一次性的后台编译多约 80 秒,结果不变。",
          "安装:离线包损坏时,现在会指出文件名并提示重新下载,不再显示解压缩的原始错误;网络不稳定时,下载会接着已下载的部分继续,不再在第一轮失败后直接停止。",
          "安装:在 Windows 上,FreeVideo 的 Python 文件被其他程序(例如安全软件)占用时,安装程序会等待并重新创建 Python 环境,不再第一次失败就停止;Windows 拒绝运行这个文件时,失败提示会说明原因并指向安全软件,不受系统语言影响。",
          "启动:已下载的源码更新在启动器每次启动时都会再按发布签名校验,不只在下载时校验;诊断日志也为视频 worker 的启动各步计时。"
        ]
      }
    },
    {
      "product_version": "0.4.28",
      "en": {
        "summary": "Image generation, editing and character generation accept LoRAs, the image model stays on the card longer for people who mostly make pictures, and Repair replaces Python packages whose files are missing.",
        "changes": [
          "Image generation: image generation, image editing and character generation each have a LoRA control. Up to 8 LoRAs can be added, each with a file, a strength and a trigger word that is added to the prompt. Video and image LoRAs are told apart and a wrong choice is explained in one sentence. Without a LoRA the output is unchanged.",
          "Speed: for people who mostly make pictures, the image model stays loaded for up to 10 minutes when it fits on the card, so the next picture starts sooner. It gives the card back at once when a video or another program needs it.",
          "Speed: the usage habit that decides which model parts stay in memory (only the kind of each operation and its count, nothing from prompts or images) is now kept across restarts. When prompts change often, the text encoder is loaded while the engine is idle, as far as free video and system memory allow, and stops when a new job starts.",
          "Speed: after the engine starts, the model files for the first request are read ahead as far as free memory allows: the video model for people who mostly generate again with the same prompt and images, and the text encoder otherwise. A new request or a busy queue stops the reading, and a machine with little free memory reads less.",
          "Install: Repair now reinstalls Python packages whose files are missing or cut short, for example a numpy library that has gone missing, which setup did not notice before when the package was already at the required version; the repair review counts them in the download. When the hardware check before a request fails to load such a package, the message names the missing file and offers Repair.",
          "Windows: on machines with little memory, a long video whose second pass stops for lack of system memory now retries in a fresh process, keeping the finished first pass, and can finish instead of failing.",
          "Guide: the tutorial book covers the additions of 0.4.25 and 0.4.26 and has a new chapter on replacing a person, with a matching guide card. In the Chinese book, quoted interface names of up to eight characters and numbers with their units are no longer split across lines."
        ]
      },
      "zh": {
        "summary": "图片生成、改图和角色生成可以使用 LoRA;以做图为主时,图片模型在显存中保留更久;“修复”会重新安装文件缺失的 Python 包。",
        "changes": [
          "图片生成:图片生成、改图和角色生成各增加了 LoRA 控件。最多可添加 8 个 LoRA,每个可选择文件、调整强度,触发词会加入提示词。视频和图片的 LoRA 分开识别,选错时用一句话说明。不使用 LoRA 时输出不变。",
          "速度:以做图为主的用户,图片模型在能完整放进显存时最多保留 10 分钟,下一张图开始更快;视频或其他程序需要显存时立即让出。",
          "速度:决定哪些模型部分驻留内存的操作习惯(只记录每次操作的类型和次数,不含提示词和图片的任何内容)现在会跨重启保存。经常更换提示词时,引擎空闲期间会预先载入文本编码器,只载入空闲显存和内存放得下的部分,新任务开始即停止。",
          "速度:引擎启动后,按空闲内存的允许范围预先读入第一次请求要用的模型文件:以相同提示词和图片反复生成为主时读视频模型,其余情况读文本编码器。新请求到来或队列忙碌时停止读取,空闲内存较少的机器读得更少。",
          "安装:“修复”现在会重新安装文件缺失或被截短的 Python 包(例如 numpy 的某个库文件丢失);此前包已是所需版本时,这类问题安装流程检测不到。修复审阅页会把它们计入下载量。请求前的硬件检测因此类包无法加载而失败时,失败信息会点名缺失的文件并提供“修复”。",
          "Windows:内存较小的机器生成长视频时,第二遍因系统内存不足中断后,会保留已完成的第一遍,改在新进程里重试,可以继续完成,而不是整次失败。",
          "教程:教程书补充了 0.4.25 和 0.4.26 新增功能的说明,并新增“替换人物”一章和对应的导览卡。中文教程书中,八个字以内的带引号界面名称以及数字与单位,不再被拆在两行。"
        ]
      }
    },
    {
      "product_version": "0.4.27",
      "en": {
        "summary": "The Agent can make music for a project, an update started during a generation waits for it to finish, and a preview can keep the guide clip's sound.",
        "changes": [
          "Agent: a new music tool makes a soundtrack for a project and places it on an audio track. The music model is downloaded on first use (about 20 GB); cards with little video memory use a built-in 8-bit form. If two attempts fail, the Agent synthesizes the music itself.",
          "Updates: when \"Update now\" is pressed while a video is generating, the page shows a waiting screen instead of a failure card, and the update starts when the job ends. The wait also covers a job on a server that another program started.",
          "Audio: a kept original audio track is written as lossless audio that browsers can play. A preview made with a guide clip keeps the clip's sound when that was requested.",
          "Long videos: when seams cannot be joined, the notes name the seams that failed first, with the step and the format. The how-to for a long video points to \"Extend\" below the finished video.",
          "Phone access: the password page is shown in Chinese when the browser's preferred language is Chinese.",
          "Speed: the encoded text of a prompt is kept per encoder file and prompt, so a repeated prompt starts sooner.",
          "Precompile: the status keeps the order of its writes, stopping responds quickly, and the diagnostic report includes the status.",
          "Decoding and reports: a CUDA error while decoding after sampling has finished no longer fails the video; the saved result is decoded once more in a fresh process. A process that keeps running after a CUDA error is ended, so the failure is reported instead of waiting. Diagnostic reports record the polling of Agent generations.",
          "Errors: a refused offline package says why, and a slow ComfyUI page is no longer blamed on the browser.",
          "Install: on Windows the portable Git is used when the installed Git cannot load libcurl, and the Windows certificate store is used for HTTPS checks, skipping a certificate that cannot be read. A rejected token or address stays explained under its field until it is accepted.",
          "Linux: on Arch, the setup names the pacman command for missing basic tools.",
          "Mac: after two Metal command buffer failures during sampling, a third attempt runs with Torch attention."
        ]
      },
      "zh": {
        "summary": "Agent 可以为作品制作配乐;生成中点击更新会等任务结束后再更新;预览也可以保留引导片段的声音。",
        "changes": [
          "Agent:新增配乐工具,为作品制作配乐并放到一条音频轨上。配乐模型首次使用时下载(约 20 GB);显存较小的显卡使用内置的 8 位形式。两次尝试都失败时,由 Agent 自己合成。",
          "更新:有视频正在生成时点击“立即更新”,页面显示等待画面,不再出现失败卡片,任务结束后开始更新。由其他程序启动的服务上的任务也会等待。",
          "音频:保留的原始音轨以浏览器可播放的无损音频写入。使用引导片段的预览在要求保留时,也保留该片段的声音。",
          "长视频:接缝无法衔接时,说明先列出失败的接缝及其步骤和格式。长视频的使用指引改为指向成片下方的“接着生成”。",
          "手机访问:浏览器的首选语言为中文时,密码页显示中文。",
          "速度:提示词的文本编码结果按编码器文件和提示词保存,重复使用同一提示词时更快开始。",
          "预编译:状态按写入顺序保存,停止时响应更快,诊断报告包含该状态。",
          "解码与诊断报告:采样完成后解码时出现 CUDA 错误,不再整条失败,已保存的结果会在新进程中重新解码一次;出现 CUDA 错误后仍不退出的进程会被结束,直接报告失败,不再一直等待。诊断报告记录 Agent 生成任务的轮询情况。",
          "错误提示:离线包被拒绝时说明原因;ComfyUI 页面加载较慢时不再归咎于浏览器。",
          "安装:Windows 上已安装的 Git 无法载入 libcurl 时改用便携版 Git;HTTPS 检查使用 Windows 证书库,遇到无法读取的证书时跳过。令牌或地址被拒绝后,说明一直显示在对应输入框下,直到输入被接受。",
          "Linux:在 Arch 上缺少基础工具时,安装脚本给出对应的 pacman 命令。",
          "Mac:采样时连续两次 Metal 命令缓冲区失败后,用 Torch 注意力再试第三次。"
        ]
      }
    },
    {
      "product_version": "0.4.26",
      "en": {
        "summary": "Recent videos open at every window width, when every download source fails the launcher shows the reason for each source, and replacing a person runs on 16 GB Macs.",
        "changes": [
          "Creation panel: \"Recent videos\" can be opened at every window width; in a narrow window the list opens over the preview. The @ character menu shows each character's face close-up instead of the full-length picture. When a part length is refused, the length panel says whether it is too short or too long instead of asking to try again. When the connection to FreeVideo drops briefly, the queue no longer reports browser extension interference; a loop that cannot submit its next job stops with the error instead of retrying.",
          "Install: when every source for a download fails, the launcher's error card lists each source with its reason and time, and the copied details and the diagnostic report include the same list. Git errors name the folder of the dependency. Account names, passwords, tokens and local paths are removed from the listed addresses. The video text encoder can also be downloaded from ModelScope, so a first install no longer depends on Hugging Face alone for this file; a download that fails the integrity check falls back to Hugging Face.",
          "Prompt enhancement: when the connection to FreeVideo is lost, a connection-lost message is shown instead of the browser's raw error, and a failed progress check is retried before the rewrite stops. After a rewrite, characters mentioned with @ are still shown by name in the prompt box. When the prompt or references change during a rewrite, a note asks for a new rewrite instead of an error with a report button.",
          "Prompt enhancement: the rewrite tab shows the scene description followed by a separate \"Sound:\" section instead of the model's field names. With several shots, each shot is its own paragraph labelled with its number and start time. A rewrite that is not edited is submitted exactly as the model wrote it. The default prompt in a prompt box that has not been edited uses the same layout and is also submitted unchanged. A prompt filled in with \"Reuse settings\" or opened from a workflow saved by an earlier version also uses this layout and, when not edited, is submitted unchanged.",
          "Editor: when the last edit can no longer be found, a blank edit opens with a message saying so; clicking the edit name at the top opens another edit.",
          "Settings: a new AI Agent section chooses whether an agent asks before starting a film and before a long reshoot, and how many minutes make a reshoot long. Both ask by default. When an agent's white-model scene fails, the agent receives the reason instead of a generic connection error.",
          "Install: when the NVIDIA driver is installed but not running, for example right after it was installed, the launcher's error card asks for a restart instead of showing the raw detection error. When the NVIDIA driver is missing, the card adds one line naming the graphics card Windows lists; when Windows lists only AMD or Intel graphics, the card says FreeVideo supports NVIDIA graphics cards only.",
          "Video generation: when a generation falls back to a freshly started worker, the result is the same as in the worker that was already running. After a driver or FreeVideo update, or on another graphics card in the same computer, the first generation no longer fails because an attention method the card cannot run was chosen.",
          "Image generation: when Windows runs out of virtual memory while loading the image model, the failure card says so and suggests closing other apps or increasing virtual memory.",
          "Linux: when no file manager can open a folder, the launcher and the image history show a message, and the image history also shows the images folder. The default save location follows the desktop's Documents folder. Curly apostrophes in English text are drawn at normal width instead of full width.",
          "Mac: encoding a reference video (replace person, extending a video) needs about half the unified memory per tile, so it runs on 16 GB Macs where it stopped with a unified-memory error, including at the lowest resolution. Results are unchanged.",
          "Mac: before installing, the launcher checks whether the disk of the chosen folder supports the file operations the Python environment needs. Without hard links the installation copies files instead; when links or renaming are not supported, a card asks for a folder on the Mac's built-in disk before anything is installed."
        ]
      },
      "zh": {
        "summary": "任何窗口宽度下都能打开“最近作品”,依赖下载在所有源都失败时,启动器列出每个源的原因;16 GB 的 Mac 上也能替换人物。",
        "changes": [
          "创作面板:任何窗口宽度下都能打开“最近作品”;窗口较窄时,列表盖在预览上打开。输入 @ 时,角色头像改为脸部特写,不再是全身图。每段时长被拒绝时,长度面板会说明是过短还是过长,不再提示重试。与 FreeVideo 的连接短暂中断时,队列不再提示浏览器扩展干扰;循环生成提交下一个任务失败时,循环停止并显示错误,不再重试。",
          "安装:依赖下载在所有源都失败时,启动器错误卡片逐个列出每个源的失败原因和耗时,复制的详情和诊断报告中也附有这份列表。Git 报错会写明对应依赖的文件夹。列出的地址会去掉账号、密码、令牌和本地路径。视频文本编码器增加魔搭下载源,首次安装下载这个文件时不再只依赖 Hugging Face;下载的文件校验不通过时改从 Hugging Face 下载。",
          "提示词增强:与 FreeVideo 的连接中断时,显示连接断开的提示,而不是浏览器的原始错误;改写进度读取失败时先重试,不再立即中止。改写后,人物仍以“@人物名”显示在提示词框中。改写过程中原文或素材有改动时,显示一条提醒,请按新内容重新改写,不再显示错误和报告按钮。",
          "提示词增强:“改写”页签只显示画面描述,下面单独列出“声音:”段,不再显示模型的字段名;有多个镜头时,每个镜头单独成段,标出镜头编号和开始时间。未修改的改写结果按模型原文提交。提示词框里未改动的默认提示词也按同样方式显示,提交时仍用原文。用“复用设置”填入的提示词和旧版本保存的工作流里的提示词同样按这种方式显示,未修改时按原文提交。",
          "剪辑视图:上次的剪辑已找不到时,打开空白剪辑并给出说明;点击顶部的剪辑名可打开其他剪辑。",
          "设置:新增“AI Agent”区块,可选择 Agent 在开拍整部片前、以及耗时较长的重拍前是否先询问,并设置重拍达到多少分钟算较长。两项默认都询问。Agent 搭建的白模场景失败时,Agent 收到具体原因,不再是笼统的连接错误。",
          "安装:NVIDIA 显卡驱动已安装但未运行时(例如刚装好还未重启),启动器错误卡片提示重启电脑,不再显示原始的检测失败信息。未找到 NVIDIA 显卡驱动时,提示卡新增一行,写出系统列出的显卡;系统只列出 AMD 或 Intel 显卡时,卡片说明 FreeVideo 只支持 NVIDIA 显卡。",
          "视频生成:生成改由新启动的进程完成时,结果与原先运行中的进程相同。驱动或 FreeVideo 更新后,或换用同一台电脑里的另一张显卡时,首次生成不再因选用这张显卡无法运行的注意力方式而失败。",
          "图像生成:Windows 虚拟内存不足、无法载入图像模型时,失败卡片会说明原因,并建议关闭其他程序或调大虚拟内存。",
          "Linux:没有可用的文件管理器打开文件夹时,启动器和图片历史给出提示,图片历史同时显示图片文件夹的路径;默认保存位置跟随桌面设置的文档文件夹。英文中的弯撇号按西文宽度显示,不再显示为全角。",
          "Mac:编码参考视频(替换人物、接着生成)时,每个瓦片所需的统一内存约减半,在 16 GB 的 Mac 上不再因统一内存不足而中止,最低分辨率下也是如此;结果不变。",
          "Mac:安装开始前,启动器先检查所选文件夹所在磁盘是否支持 Python 运行环境所需的文件操作。不支持硬链接时改用复制继续安装;不支持链接或改名时,在安装任何内容之前提示选择 Mac 内置硬盘上的文件夹。"
        ]
      }
    },
    {
      "product_version": "0.4.25",
      "en": {
        "summary": "An audio clip can be used as is, so the video carries that clip as its sound; a tutorial and feature examples explain each feature with an example that can be filled in directly; after a video finishes, the model stays loaded and the next video starts without reloading.",
        "changes": [
          "Studio: audio cards offer \"Reference\" or \"Use as is\". With \"Use as is\", the clip is written into the video unchanged as its sound, the picture follows it, and the duration follows the clip and is locked. \"Use as is\" is not available with \"Preview first\" or when extending a video. Replace person downloads its model when it is missing.",
          "Guides: \"Tutorial\" in Settings has chapters with pictures for each feature, and \"Try it\" fills a chapter's example into the creation panel. \"Feature examples\" shows each feature as a looping preview of the input and the result, with \"Use this example\". The examples are shown once on first use, and after an update only the new ones are shown.",
          "Memory and speed: after a video finishes, the model stays on the graphics card and in memory with no time limit, so the next video starts without reloading. It is released when memory runs low, when another program starts using more video memory, or before the picture or prompt model starts; when memory is tight, as much of it is kept as still leaves room for the next request.",
          "Speed: the first video of a new length or size starts sooner, because three of the model's kernels are compiled once for every length, size and head split. Requests with LoRA adapters run faster, because each adapter's input projection is computed once per layer instead of once for every group of attention heads. A slowdown switched on automatically after the app was closed or stopped unexpectedly now lifts after one completed generation instead of three; a slowdown caused by a reported GPU error still needs three.",
          "Generation: pressing Generate with an empty prompt shows \"The prompt is empty.\" next to the button, with \"Generate anyway\". In the editor, broken characters left in a description by a cut or a paste are removed before the part is sent. On a computer with several graphics cards, queued videos run at the same time, one on each card.",
          "Workflows: a saved workflow opens ready to run when a value on its FreeVideo node is no longer accepted; that value returns to its default and the rest of the workflow is unchanged.",
          "Creations: a video can be marked as a favorite with the heart button, and \"All\" and \"Favorites\" at the top switch between every video and favorites only; the choice is kept for the next time Creations opens.",
          "Autopilot: a chat can be renamed with \"Rename\" in its menu. The name is kept after a restart, and clearing it shows the first message as the title again.",
          "Install: on Linux, setup from the launcher downloads Git and a C++ compiler privately when they are missing instead of asking for sudo; only basic tools such as curl still come from the system, and the launcher shows the command to install them. A graphics driver older than 580 gets its notice with the detected card even when nvidia-smi cannot answer the setup query. When the system is older than 2022 or lacks the graphics libraries its window needs, the Linux desktop app shows the reason and what to do instead of closing without a message.",
          "Updates: on every platform, update downloads can use the mirror in China, and a slow source is switched automatically, continuing the download.",
          "Diagnostics: on Mac, a generation refused at the start for lack of free memory has its own row in the error panel, and the report records the memory at that moment. Feedback for a video in Creations that cannot be opened includes the player's error code and the file's response. Diagnostic reports record which graphics card the run used."
        ]
      },
      "zh": {
        "summary": "音频可以原样使用,视频直接以这段音频为声音;新增教程和功能示例,逐项介绍功能,示例可以直接填入;视频完成后模型保留,下一条视频无需重新加载。",
        "changes": [
          "工作室:音频卡片可选“参考”或“原样使用”。选择“原样使用”时,这段音频原样写入视频作为声音,画面随之生成,时长跟随音频并锁定。开启“先出预览”或接着生成时,不能选择“原样使用”。替换人物缺少模型时会自动下载。",
          "指引:设置中的“教程”按功能分章,配有图片,点“试一试”可把该章的示例填入创作面板。“功能示例”逐项以循环预览展示输入和成片,并可点“用这个示例”。首次使用时会依次展示一遍,更新后只展示新增的示例。",
          "内存与速度:视频完成后,模型会一直保留在显卡和内存中,不再有时间限制,下一条视频无需重新加载。内存不足、其他程序占用的显存增加,或图片和提示词模型启动前,模型会被释放;内存紧张时,在保证下一次请求放得下的前提下尽量多保留。",
          "速度:换新的时长或尺寸后,第一条视频更早开始,模型的三个内核对所有时长、尺寸和分头方式只编译一次。挂 LoRA 的请求更快,每个适配器的输入降维乘法每层只算一次,不再每个注意力头组各算一次。应用被意外关闭或结束后自动开启的放慢,完成 1 次生成就解除,不再需要 3 次;显卡报告错误引起的放慢仍需 3 次。",
          "生成:提示词为空时点生成,按钮旁会显示“提示词为空。”,并可点“继续生成”。剪辑视图中,剪切或粘贴在描述里留下的残缺字符会在发送前去掉。有多张显卡的电脑上,排队的视频会同时生成,每张显卡一条。",
          "工作流:另存的工作流中,FreeVideo 节点上有不再接受的取值时,打开即可运行;该取值恢复为默认值,工作流其余内容不变。",
          "作品库:可以用心形按钮收藏视频,顶部的“全部”和“收藏”可切换显示全部视频或只显示收藏;下次打开作品库时保留所选。",
          "AI 全自动:对话可以在菜单中“重命名”。名称在重启后保留,清空名称后重新以第一句话作为标题。",
          "安装:Linux 上从启动器安装时,缺少 Git 和 C++ 编译器会改为私有下载,不再需要 sudo;只有 curl 等基础工具仍需系统提供,启动器会给出安装命令。即使 nvidia-smi 无法响应安装检查的查询,低于 580 的显卡驱动也会显示带检测到显卡的提示。系统早于 2022 年或缺少窗口所需的图形库时,Linux 桌面版会说明原因和处理办法,不再无提示地退出。",
          "更新:所有平台的更新下载都可走国内镜像,下载源过慢时自动换源续传。",
          "诊断:Mac 上因可用内存不足而在开始前被拒绝的生成,在错误面板中单独一行说明,报告会记录当时的内存。作品库中视频打不开时,反馈会附上播放器错误码和文件响应。诊断报告会记录这次运行使用的显卡。"
        ]
      }
    },
    {
      "product_version": "0.4.24",
      "en": {
        "summary": "A person in a source video can be replaced with the character in a person image; reference videos can be used for depth only; the built-in AI sets up a white-model scene from one sentence; an earlier video's prompt and settings can be reused; decoding on cards with less video memory is quicker.",
        "changes": [
          "Studio: \"Replace person\" in the Media section replaces the person in a source video with the character in a person image, and can keep the source video's sound; if that sound cannot be added, the finished video is still kept. A reference video can be set to \"Depth only\"; its depth is computed once and reused by later requests.",
          "Creation panel: after an earlier video is selected in Recent videos, \"Reuse settings\" fills its prompt, picture size and duration back into the panel; \"Undo\" restores the previous values within 10 seconds.",
          "White models: when no external Agent is connected, the white-model button in the chat box works with the built-in AI. Describe the shot in one sentence; if a ready-made scene fits, a white model is generated for confirmation, otherwise the reason is shown and the sentence is kept.",
          "Long videos: when a description is split across parts, the story is no longer repeated in every part; the shared text keeps only what stays the same, such as people, place, lighting and style. Without the prompt model, the description is split across the parts by sentence instead of every part using the whole description, and the creation panel offers to download the prompt model; a download in progress is paused when generation starts.",
          "Editor: a continued part can be as short as 5 seconds, and on Mac and on computers that cannot make 10 seconds at once a new part starts at 5 seconds; a video with a continued part shorter than 10 seconds has at most 4 parts. A description over the length limit is pointed out before sending, including on a retry.",
          "Generation: after other ComfyUI workflows have run, the video memory they hold is released before a picture or a prompt rewrite starts, without restarting ComfyUI. A missing text encoder model is detected before encoding starts, with repair guidance; if a model repair fails while downloading, the original file is restored. When generation cannot start because the installation is in use, the failure details name the programs holding it open. When the GPU checks before switching to int8 fail, the failure card lists the reason for each check.",
          "Memory and speed: on graphics cards that cannot hold the whole decoder, decoding no longer prepares the same layers again for every video segment or every new request while memory allows, with identical output. On most Windows graphics cards with 8 to 12 GB of video memory, the first pass of two-pass videos between 10 and 12 seconds without LoRA computes 8 attention heads at a time instead of 4, as shorter videos already did. On Windows, when other programs free video memory during a generation, the next pass and decoding use the recovered memory.",
          "Install: setup no longer stops with a dependency mismatch after the installation folder was moved, copied or restored, and a dependency folder too deep for Windows is reported as a long path. Text encoder and tokenizer files deleted after installation are restored from the local copy before encoding and during repair, without downloading.",
          "Phone access: the phone page works when local network access is switched on in the launcher, and no longer shows \"Connection lost. Reconnecting...\" with no content.",
          "Agents: the install-skill command installs the FreeVideo skill for an Agent, also from deeply nested Windows folders. The tool list an Agent receives is shorter: common tools are listed in full and the others are opened through a help tool.",
          "Diagnostic reports: when ComfyUI restarts during a generation, the report of that run is marked as interrupted instead of running, and a long video keeps the report of the part that was interrupted."
        ]
      },
      "zh": {
        "summary": "可以用人物图中的角色替换源视频里的人物;参考视频可以仅参考深度;内置 AI 可以用一句话搭出白模场景;可以复用之前视频的提示词和设置;显存较小的显卡解码更快。",
        "changes": [
          "工作室:“素材”一栏新增“替换人物”,用人物图中的角色替换源视频里的人物,可选保留源视频的声音;声音未能加入时,生成的视频仍会保留。参考视频可以勾选“仅参考深度”,深度只计算一次,之后的请求直接复用。",
          "创作面板:在“最近作品”中选中之前的视频后,点“复用设置”可以把它的提示词、画面尺寸和时长填回面板;10 秒内点“撤销”可以恢复原来的内容。",
          "白模:未连接外部 Agent 时,聊天框的白模按钮也可以使用。用一句话描述镜头,有合适的现成场景就先生成白模供确认,没有则说明原因并保留输入。",
          "长视频:描述分到各段时,不再在每一段重复整个故事,各段共用的部分只保留人物、场景、光线和风格等不变的内容。未安装提示词模型时,描述按句子顺序分到各段,不再每段都用整段描述,创作面板会提示下载提示词模型;开始生成时,正在进行的下载会暂停。",
          "剪辑视图:接着生成的每段最短可以是 5 秒,在 Mac 和无法一次生成 10 秒的电脑上,新的一段默认 5 秒;含短于 10 秒的接续段时,一个视频最多 4 段。描述超过长度上限时,发送前(包括重试时)就会提示。",
          "生成:在 ComfyUI 里跑过其他工作流之后,生成图片和改写提示词前会先释放这些工作流占用的显存,无需重启 ComfyUI。文本编码器模型缺失时,会在编码开始前检测并给出修复指引;模型修复在下载中失败时,会还原原来的文件。安装正被占用而无法开始生成时,失败详情会写出占用它的程序。切换到 int8 前的 GPU 检查失败时,错误卡片会列出每项检查的失败原因。",
          "内存与速度:显存放不下整个解码器的显卡,在内存允许时解码不再为每个视频片段或每次新请求重复准备同样的层,输出不变。显存 8–12 GB 的大多数 Windows 显卡上,10–12 秒、不带 LoRA 的两遍视频,第一遍每次计算的注意力头由 4 个增加到 8 个,与更短的视频相同。Windows 上,生成过程中其他程序释放显存后,下一遍和解码会用上恢复的显存。",
          "安装:安装目录被移动、复制或还原后,安装不再因依赖不一致而停下;依赖目录层级过深时会按长路径问题提示。安装后被删掉的文本编码器和分词器文件,会在编码前和修复时从本地副本还原,不需要下载。",
          "手机访问:在启动器里打开局域网访问后,手机页可以正常使用,不再一直显示“连接断了,正在重新连接…”且没有内容。",
          "Agent:install-skill 命令可以为 Agent 安装 FreeVideo 技能,Windows 上层级很深的目录也能装。Agent 拿到的工具列表更短:常用工具完整列出,其余工具通过帮助工具打开。",
          "诊断报告:生成过程中 ComfyUI 重启后,这次运行的报告会标为已中断,不再显示为运行中;长视频会保留被中断那一段的报告。"
        ]
      }
    },
    {
      "product_version": "0.4.23",
      "en": {
        "summary": "Each part planned in the editor can have its own description; low-memory generations explain why they are slower.",
        "changes": [
          "Generation: When a generation runs in low-memory mode, the progress view says why it is slower.",
          "Editor: Each part of a video planned in the editor can now have its own description, and a continued part uses exactly what you write for it. The highlight on characters mentioned with @ in the prompt box updates as soon as a rewritten prompt is filled in."
        ]
      },
      "zh": {
        "summary": "剪辑里规划的每一段可以单独写描述;低显存模式下的生成会说明较慢的原因。",
        "changes": [
          "生成:低显存模式下,进度区会说明这次生成较慢的原因。",
          "剪辑视图:剪辑里规划的每一段现在都可以单独写描述,接着生成的一段会用为它写的描述。改写后的提示词填回时,@人物 高亮立即刷新。"
        ]
      }
    }
  ]
}