itch.io

1. i'm not seeing the bug you are.
2. yes, it is needed. perhaps it does not currently work in all cases, but it does not need removal, just fixing. people kept messaging me saying they can't start a game and keep getting an error only to realize they didn't have a model loaded or weren't using the default endpoint, etc. 

Not implemented yet, but i just added the ability for the user to see and edit the start game prompt themselves which is the first step

would love if you joined the discord. these posts will unfortunately get lost/forgotten here. up to you though, thank you for taking the time to test out stuff <3

you'd need to give me more info on what you're opening and how. not hearing anyone else with that issue so it most likely isn't a bug

the browser version here on itch

I pushed a serialization fix for dictionaries/placeholders which may solve it in the future.

okay

nice, thank you for taking the time for this. sounds like there's a bug to fix but it's not from the default settings. this should reveal it and prevent others from running into it. 

You'd have to give me something reproducible. I asked in the discord but no one else seems to be having that issue. a pretty bad issue at that.

I added some info buttons to the world editor which should help to explain things, but yeah I unfortunately don't have time atm to write an actual guide. Gotta fix bugs/add polish rn.

Did you remove dictionaries from your prompt on Desktop?

By downloading it to your computer

use the browser version

if you can get solid repro steps i can fix it

download the model from itch fresh, don't modify it. does it work?

check your settings in Settings -> Endpoint

Scrolling is fixed in the latest update. As for image gen, you'll have to be more specific about your set up. Are you using A1111 or Comfy? You'll also need to expose those to your LAN.

Fixed in the latest update

by default the model already has sliders, so re-exporting isn't needed unless you want new/different ones. if it worked beforehand but not after you changed it then there's something you're missing

try it out

you need to load a model in the endpoint settings. i will smooth this out in the future with a first time user flow. 

sweet, let me know if you run into any other issues. 

you need the model i linked as well

best way is to export each thing you want carried over from v1.2 and then import them all into v2.4.0. I added multi-importing so that part should be fast. just the exporting will be time consuming.

on what platform?

yes, it now supports all versions of VRM models not just v1.0 and you can tie stats to any body sliders  on the model, not just the old defaults. the models you're using probably don't have the any sliders. you need to get a model that already has them or create the sliders in blender. the NSFW Character Model has the old default body sliders and is fully compatible if you want to simply use that.

make sure you loaded a model in the endpoint settings

You can also manually download any model you want and put it into the models folder. The recommended section is just something to get you started if you've never used local LLMs before.

That is still managed by FieryLion, I have no control over that. 

Something to keep in mind though. Renting GPUs to do anything is super expensive right now, especially so if the task is AI related. The model change was to try and get something that won't be unreasonably expensive to pay for. Best option if the current quality isn't good enough is to either create an account on an LLM hosting site or try out local stuff. Small LLMs able to run on <4GB of VRAM are surprisingly good for the hardware/effort required. The desktop app runs models natively and has a few suggestions of small models to try out which can be downloaded directly in the app itself, no technical skill required. 

Nah, it runs on any GPU. The VRAM viewer only runs on NVIDIA. Two seperate things.

ah, interesting. glad that you were able to get it working. that setting is supposed to be on by default but I guess something messed it up. I will add a short how to setup button similar to the one for image-gen which will hopefully help others in the future.

Sorry, the update wasn't live. Try again

yes, it's definitely possible. you can already play on it in the browser though, so give it a shot and let me know if you run into issues. the android app would run the exact same way, so this is a good test run.

Should be fine. Try these and see if it works:

• Settings → Generation: Thinking = Native, Native Reasoning = Default/None.
• Reload the model after the errors and try one turn — narration likely works despite the log noise.
• Update LM Studio to latest.

should probably delete that image then, that's very much still visible

Need a screenshot or something of your settings. probably an easy fix. Nothing is different in this version as far as endpoints go

just for creating images in the world editor at the moment, in game generation is still being worked on

It's up to you now, we just had a vote :)

you can try checking in '..\AppData\Roaming\formamorph' and copying everything in there over to the 'userdata' folder in the new version. Either way though, all you need to do is export your worlds and then import them into the new version. it's all backwards compatible, including saves. Just takes a bit of time.

Thank you, I'm glad you like Freya! Working on the next update atm featuring Sethi. She's getting a big overhaul to improve her story, so stay tuned~

As for running on desktop, the Formamorph windows/linux/mac app can run models natively, so you don't even need a second application running in the background for it. And on mobile, if you have LM Studio, you can actually enable the Serve on Local Network setting to allow you to connect to your PC and play on mobile too! All local, any time you want, offline. Def let me know if you have trouble setting it up by reaching out on the Discord.