Hey Constant Readers,
We've just rolled out another huge update, focused on giving players more control, wider hardware support, and a better story-tracking experience.
Whether you're deep in Story Mode or just starting out, this update brings long-awaited features and key fixes -- and sets the stage for what's next.
📖 New: Story Objective Viewer
We’ve added a new in-game tool to help you stay oriented during structured story mode:
✅ Chapter Tracking: You’ll now see how many objectives you've completed (e.g. "3/5 complete") at the bottom of the screen.
✅ Hover for Details: Hover over the chapter name to see a full breakdown of what's been completed and what's still in progress. These are hidden and only available on hover so that we don't spoil the story for you. If you've ever wondered why the AI keeps trying to send you back to the same place -- this will make that logic crystal clear.
✅ Flexible Progression: You don’t have to check off every objective to move forward -- completing at least 70% of them is enough to unlock the next chapter. You can follow your story, bend it, or complete it however you'd like.
Once you finish the final chapter, you’ll automatically transition into Free Exploration Mode, where you're free to explore, create, and continue the story indefinitely (or as far as your hardware will allow).
💾 Save System (Finally!)
You can now save your progress and pick up where you left off -- even including your last generated image!
The current system supports one story at a time and includes your full story state — including character, item, story, and location memories -- chapter & beat progress, as well as your last generated image.
A full save manager (with multiple slots) is already in development and will be rolled out in a future update!
⚙️ New Hardware Support: CPU Inference Mode & GPU Tuning
We've made big changes under the hood to help more players run the game -- even on lower-end hardware that is below our recommended requirements:
🧠 CPU Inference Mode
We’ve added options in Settings to switch both image and text generation to CPU-based inference by setting the model to use 0 GPU layers.
While CPU mode is significantly slower, it allowed us to successfully play the game on Intel integrated graphics (13900K) at a reasonable speed -- and we’re thrilled that this gives more players a shot at enjoying the game.
⚠️ Note: System monitors may still show GPU activity in CPU inference mode. This is normal - the underlying llama.cpp engine initializes GPU contexts and keeps memory 'warm' for offloading, but all actual text generation happens on your CPU.
📊 Smart GPU Detection + Manual Override
We detect your GPU by name and now match it to a built-in VRAM database to automatically decide how many LLM layers to offload.
And for tinkerers:
✅ New Override Field in Settings: You can now manually enter the number of layers to offload if you want finer control.
📉 If your system is tight on VRAM (especially if it’s shared with background apps), lowering the offload count can greatly reduce instability.
These tweaks mainly affect a very small number of systems — but if you're playing on tight VRAM, low-end hardware, or complex Windows environments, you may see better stability here.
🖥️ Better Stability on Edge Cases
We’ve also improved compatibility with systems that were just at or barely under our recommended spec or have uncommon configurations -- especially systems affected by aggressive VRAM limits and with different protections or antivirus software that may prevent writing to folders where we save the game and settings data.
Fixes include:
Display mode fallback logic (if your driver doesn’t support switching resolutions)
Safer file system handling (for environments with locked folders or aggressive antivirus)
More graceful failures in weird setups (instead of crashing or hanging)
⚠️ Removed: Flux Model Support
We've removed support for Flux image generation models and are no longer bundling them with the game. If you were using Flux previously, the game will now automatically fall back to the standard Stable Diffusion model -- no action needed on your part.
We will be adding them as a DLC package (free, of course) later this week, that way customers that don't intend on using them don't have to waste disk space on them anymore as they more than doubled the size of the game!
We are still huge believers in flux as the quality of generation really is fantastic, we just want to ensure we deliver it in the best possible way for our customers that aren't running high powered battle-stations.
More to come on this in a midweek update!
🧪 Sneak Peek: New Settings Menu In Development
We’re currently testing a redesigned settings UI that’s a massive step up in usability and clarity.
📸 Here’s a preview of what’s coming:
We’ll share more about this in the next update -- but we wanted to keep you in the loop on what's coming next!
With all our love and neverending imagination,
🪄 The Neverending AI Storybook Team