太长的视频无法总结完整，建议加入自动分段总结 #16

Activity

1. josStorer

   added

   enhancementNew feature or requestNew feature or request

   discussion

   on Mar 17, 2023
2. josStorer commented on Mar 17, 2023

   

   josStorer

   on Mar 17, 2023

   Member

   More actions

   目前长视频的总结模式是, 保留开头结尾, 对中间部分按比例裁切, 保证在一定token范围内, 以期望保留视频整体大意
   核心逻辑见: https://github.com/josStorer/chatGPTBox/blob/master/src/utils/crop-text.mjs

   分段总结是可行的, 但会导致某些长视频消耗大量token, 并且每个片段将使用独立上下文, 导致难以与其他片段关联, 目前这个功能优先级不高, 并且不一定会支持, 如果有好的想法可以在这个issue下讨论
3. DIYer22 commented on Mar 21, 2023

   

   DIYer22

   on Mar 21, 2023

   More actions

   怪不得, 没法基于视频问答. 我觉得可以设置一个选项, 节约模式和性能模式, 或者(概览/完全), 放在视频问答窗口的按钮, 让用户自己选择. 或者在设置界面新增单个视频最长 token 长度的设置, 根据设置长度来截取字幕

   对于有 API key 用户, 额度多得消耗不完, 想要性能模式.
4. ssyzyg commented on Mar 21, 2023

   

   ssyzyg

   on Mar 21, 2023

   More actions

   对于长视频，一种简单的做法可以是拆分视频/文本长度，比如每5分钟做摘要，然后对各个摘要汇总做二次摘要，实测下来也可以达到要求，而且也可输出分段摘要。
5. PeterDaveHello commented on Jul 31, 2026

   

   PeterDaveHello

   on Jul 31, 2026 – with ChatGPT Codex Connector

   Member

   More actions

   Replied and closed by gpt-5.6-terra medium:

   Thank you for raising this limitation early. The more complete English report and subsequent design discussion about splitting and summarizing long video content are tracked in #123, so further work should be consolidated there.
6. PeterDaveHello

   closed this as duplicateon Jul 31, 2026