Music: Rumors of a Dungeon by human gazpacho — Free Music Archive

August 2026 Update

Full release notes, with videos: https://eu.daimonia.app/articles/release-august26

A huge thank you to everyone supporting the development of Eudaimonia, and to everyone who took the time to report bugs, send feedback or share ideas. It really shapes what I work on. Many of the features and improvements in this update were suggested by users, in the request threads on the forum.

⚠️ Deprecation notice

Two things become deprecated with this update, and will be removed in the next one:

Stereo v1 — replaced by Stereo V2. Note that Stereo V2 has its own stim format, so unfortunately it's not possible to migrate your Stereo stims to the reworked engine.
The "Legacy devices" mode in EstimAI — the Multi-Devices mode is just better overall, and doesn't change much on your end.

Nothing breaks today, and nothing changes in this update. But if you rely on one of these two, it's a good time to start migrating.

Stereo V2 — Reworked engine for audio E-stim `NEW`

The Stereo engine has been fully rewritten to give you better and richer stims, more control, and a more robust safety model. Up to three carriers per channel, a new sequence editor, and a lot of work on the safety side.

Stream mode `NEW`

With Live streaming enabled, the scene starts playing as soon as the first sentence is written, and the lines keep coming as they are generated. You now see the first lines a few seconds after hitting Send, instead of staring at a loading spinner.

Body detection `NEW`

A small detection model now runs directly in your browser and locates body areas on the generated images. No image is ever sent anywhere for this. It unlocks the Censor Module (the AI decides to censor images during the scene, a rather cruel tease tool) and the new camera effects, that move the camera over the detected body zones.

Multi-character image generation `NEW` `SUBSCRIBERS`

The Intelligent image generation agent now supports multiple characters. Every visible character keeps its own appearance and outfit, with its own pose, expression and position in the frame.

Sound Effect Module `NEW` `SUBSCRIBERS`

EstimAI has a new agent: a Sound Director. It listens to the narration and places short sound effects on it — moans, gasps, impacts, ambiance beds, props, and so on. Still a work in progress, and a bit hit or miss for now.

Character Studio `NEW DESIGN` & Character sets `NEW`

The characters page has been completely redesigned. And with the new Character sets, you can now build galleries of your favourite characters in different outfits, straight from image generation, and use them in any session.

Scenario phases `NEW`

A dedicated phase system has been added to the scenario editor, on top of the existing trigger/event system. A new scenario that takes full advantage of it should be available soon.

Local TTS: KokoroTTS & one unified server `IMPROVED`

KokoroTTS is in, and instead of multiplying repositories and installation steps, the local TTS server is now a single unified server that hosts every engine — you only install it once: https://github.com/diataxis/local-TTS

Onboarding: making the first hour less brutal `NEW`

The first steps in Eudaimonia, especially for EstimAI, are confusing and intimidating. First steps in that direction: Quick Play, a Get Started checklist, a ComfyUI wizard, a TTS wizard, and demo videos.

Writing quality `IMPROVED`

A new system prompt has been written: EudaimoniaV3, even more centered on Eudaimonia's format and principles. Plus a Scene State agent for physical continuity, and a rewrite of the MCQ paths.

Important improvements

Dedicated audio output for stims — use TTS or the new sound effects while your stims run on Stereo V2 or the 2B.
ComfyUI Cloud support — no GPU? Point Eudaimonia at cloud.comfy.org instead of a local ComfyUI.
New screen & character effects — lots of new FX, and sound cues attached to some of them, so a whip crack actually cracks.
Image generation: new templates & styles — Anima, Krea 2, a new two-pass generation type, and a big quality pass on all the presets.

A word about moderation

A fair amount of moderation and content filtering has been added over the last weeks. It only targets one thing: content involving minors. Adult content, however extreme, is not concerned.

In practice you should not notice anything. But automatic filters do produce false positives. If you get a [SMO-CF] error on something perfectly legitimate, please send it to me. I'd rather fix a false positive than leave the filter loose.

What's next?

First: a new scenario that takes full advantage of the new features. I believe you'll like it, as it's a Eudaimonia adaptation of a classic.
Second: a new game mode, which I wasn't able to finish in time for this update. I hope to release it in the next one, at least as a beta.
Third: lots of things, most of them coming from the request threads on the forum.

That's it for August. If something behaves weirdly, please tell me, either here or by email.

I hope you'll like it, have fun!

August 2026 Update

⚠️ Deprecation notice

Stereo V2 — Reworked engine for audio E-stim NEW

Stream mode NEW

Body detection NEW

Multi-character image generation NEW SUBSCRIBERS

Sound Effect Module NEW SUBSCRIBERS

Character Studio NEW DESIGN & Character sets NEW

Scenario phases NEW

Local TTS: KokoroTTS & one unified server IMPROVED

Onboarding: making the first hour less brutal NEW

Writing quality IMPROVED