Beta 0.9.16
Added a minimal scene reaction popup delay configuration (default: 0)

添加了一个场景演出气泡最小间隔秒数（默认：0s)

謝謝作者!

@虛無_鬼王 这个可能确实是AI模型自己的问题，以后加一个强制最小delay

不好意思,頻繁出現是我改出bug了,剛剛按照默認的[示例输出]改回去後就正常了
可是時不時連續兩句(無先後間隔)同時跳出的現象還是存在(不是一個人兩句)
不確定是不是gemini忘記寫delay了,這個現象其實從開始用以來就有不少看見的記憶

可是我確認過,並不是兩次對話對撞造成的
請稍等一下,我在確認是不是自己改提示詞改出bug了

@虛無_鬼王 试试把回合冷却设置高一点

請問可以設定最低顯示速度嗎?
不管我怎麼調整提示詞,都會不斷出現兩句甚至三句同時出現的現象
我用的是彈出氣泡,因此看到的就是高速刷屏,每次都得停下來打開視窗看看他們在說什麼
這種情況在我高速走動時特別容易出現

@Tim Add auto explore keyword in the prompt filter from the panel. I don't use that mod so I have no idea what message it outputs, but as long as the filter contains the exact word it outputs, it will be filtered.

Cool mod, but this mod interferes with AutoExplore. It no longer shows you switching between harvesting and mining mode. The AI completions sometimes outputs the AutoExplore messages. Anyway to fix?

NPC をより魅力的にしてくれる神 Mod
The Good Mod That Makes NPCs More Charming

Ew, genslop

@NobeOddy Try increasing the global AI Service Cooldown or Turn Cooldown/Seconds Cooldown in the config setting. You can open the file in panel.

Adding Grok has made NPCs excessively chatty, especially party members. Anyway to make them talk less?

@Inquisition Time! Yes, using OpenAI compatible endpoints. As for sillytavern, you can use it as LLM provider. But not prompt cards.

compatible with KoboldCPP or Sillytavern?

Thanks a lot.

@Submissive + Greetable In the Emmersive panel -> Backgrounds -> Uncheck Enable Popup Chat

Sorry for the silly question, how do I disable chat popups in the pop notification and make them pop up as speech bubbles above the character?

ew, ai

好棒，能聊天了，赞美你！

@Grizzly Yes, the config panel UI is only translated to English/Chinese/Japanese, but you'll get Korean responses. The mod automatically enforces localization rules on output (most of the popular LLM models can handle it correctly)

Can I use it in Korean?

@DK Ok that worked, thanks it wasn't clear you could bypass the API field for local models.

哇！謝謝作者更新！聊天功能期待好久了！

0.9.14 Update:
Added player chat feature to initiate scene request.
Added show name in chat log config, default enabled.
Fixed some internal logic issue.

添加了玩家聊天功能，并同步AI对话。
默认显示NPC姓名至消息记录中。
修复了一些内部逻辑。

@agentimith Add OpenAI Provider - Enter Any Key - Change Endpoint

Is local AI usage disabled? I only see where to add APIs for OpenAI providers and Gemini. I don't see anywhere you can enter an IP address of a server.

@DK Nevermind it was me being stupid, i changed the params when it was http but didnt change it after i switched to https, it works now and i got succesful replies. Thanks so much for your help my dude.

@DK yeah i copied the params from the instruction page, given i might have overlooked something or am just too stupid but i think i've at least copied correctly.

@Aurelius Did you change the params according to the instruction page - DeepSeek only supports legacy json_mode output so it must be changed

also double and triple checked my api key then send a request with it on another service and got a succesfull response on that service.

@DK Yeah tried that too but then i get another error which goes along the lines of "This Response_Format type is unavailable now" and "invalid_request_error"

@Spyridon @Aurelius The steps I took: Add OpenAI Provider -> Paste Key, Enter -> Change Nickname(optional), Reload
-> Change Endpoint to the one stated in the wiki important: it seems that DeepSeek API has blocked access from http now. At the time the doc was written, it was accessible. Now it must be https - I have updated the wiki page.
-> Change params accordingly

@DK Ahh I see now, thank you!

But I am having the same problem Aurelius is having. Upon creating an OpenAI and attempting to use the DeepSeek key, it didn't seem to be working. It simply said "unauthorized".

Are you able to confirm it's working and no other steps necessary that aren't in the documentation?

@Aurelius 401 Authentication fails due to the wrong API key. Please check your API key. If you don't have one, please create an API key first. Make you you paste the key as-is into the game.

i cant seem to connect it to the official deepseek service, it just gives me the 401 "Unauthorized" error.

@Spyridon Because the instructions says "OpenAI Compatible Providers: Using different providers is as simple as swapping the endpoint, model, and correcting params."

Why is it that only Gemini and OpenAI are showing up and none of the other models used in the API_Setup instructions?

哇！謝謝作者！一直頻繁打擾真的不好意思！

0.9.13
Added recent actions filter under Prompt Setting panel
添加了最近对话过滤器至提示词设置面板

喔喔，很抱歉我沒有說清楚，我有個習慣
等待的時候會休息，戰鬥的時候會休息（幫隊友回 HP MP），時不時手滑按錯到休息

所以我的同伴就因為這個一直將話題轉到休息的事情上，不管談什麼都會被休息打斷，因為不太希望這部分影響到她們，所以一直很困擾

我嘗試使用了 FilterMessages 來隱藏訊息，但就算訊息隱藏了她們依舊會被影響
不確定是不是文字被隱藏了但記錄還在之類的原因，再之後就是找到 DisabledProviders 了，不過依照你的說法這貌似並不是這樣用的
其實我也不確定她們是對反覆出現的大量'休息'文字，還是對我的狀態做出的反應...

@虛無_鬼王 这种情况发生的时候，重新读一次档 清空最近行动日志。AI很容易陷入无限循环。DisabledProviders不是这个用法，它是用来禁用上下文提供器的

訊息還是一直處於隱藏，是說隱藏的訊息甚至收不到通知，這我第一次知道

那個不好意思可以問一下嗎？
因為隊友一直在關注我坐地上休息的事，因此想消除影響，因此我這樣做了
但是沒有產生作用，她們還是一直關注我休息的事，是我填寫方式錯誤了嗎？

DisabledProviders = The knowledge seems too complex without proper guidance,你坐下来闭上了双眼,你结束了休息

作者,你的訊息一直處於隱藏

Spent hours trying to figure this out. You say we can run with ollama, but I cant figure out how to setup local models such as gemma3 with this. The only options for setup are ai services from gemini or openai. Those require API keys (local models dont need api keys) to set up. Looking over dry_run.txt indicates that the system is only set up to prompt those two cloud models. Can anyone clarify how exactly to set up local models? I have ollama running gemma3 and i now understand that there are generate and chat endpoints available, but i cant seem to make anything work with the mod. Just keep getting 404 not found when trying to test.

哇！真的非常感謝！

0.9.12
Added dedicated popup bubble toggle
已添加独立的气泡开关，在背景设定栏可以为单独的角色启用弹出气泡（会取消掉常规气泡）