Ep5 Dev update 09-01-2024 (Patreon)

Happy Sunday!

Did you enjoy last Friday's NSFW post? Have you played Episode Ayane yet? We're just getting started with the content.

Poll results:

With a total of 119 votes, Mika takes a majority win! Expect a nice piece of Mika side story in September! I will aim to release on the Sunday after next.

It will be interesting to see what mischief Mika gets up to when the player is not around.

Game update:

On the development front, I've been busy with a lot of coding, polishing, and streamlining!

Now every time you earn a new stat, you will get an animated visual confirmation of which stat and how many points you earned.

(video demo https://youtu.be/jsKDluvEtPM?si=UVqG_2a48tkmrUME )

Skill items have also been relegated to their own separate menu choice, and you will need to click the item you wish to use.

Using a skill item will pass the time - but to make them worth spending time on, I made it so that they give from 2-3 points in the relevant skill - and sometimes they yield other bonuses.

For example, using the coding book will earn you proficiency, but also unlock the ability to do coding freelance work on any eveningslot time.

So players who want to earn loads of money can work at the cafe during the day and freelance in the evening.

Gifting likewise has a large degree of polish.

(gifting video demo https://youtu.be/TigW1oLsWjU )

Every item has a unique reaction for each girl (And there is something like 60 different gifts now, so that's a lot of reactions :O!).

Each girl has certain gift preferences and it won't be too tough for you to figure out which gifts do better than others. Girls often like more than one category of items, giving you +1 affection. But only one category of gifts will give you +2 affection .

So what is the point of earning affection?

Earning affection will help you progress with a girl faster - bypassing some of the non story dates. Furthermore there will be unique scenes if you get enough affection with the girl.

With this massive overhaul of the skills and affection and gifts. Expect the next release of the game to require time for balancing. At the current time I have no idea if the player is getting enough stats/money/time or if I am being too generous with it.

Other Dev news:

I've made progress on art and story but not something I want to show at the moment due to spoilers so let's just leave it at that. But here are the updated designs of Saya's teammates and them wearing their school uniform rather than basket ball outfit.

 That's it for this weeks update! I hope to show off more story content soon and possibly get a release date for a beta prepared. I hope to get it by the end of sept but that depends on how smooth things progress.

I will be honest, we are doing the last iteration of the art model, properly integrating all the girls and it takes a lot of trial and error. It's been a costly investment but thanks to all of you supporting the project, this model is nearing it's final stage!

Now, if you use AI and want to get your hands on something similar, here is the statement from Fino -

MODEL UPDATE

If you are dabbling in AI or use sdxl you will find this interesting. If not you can skip this part of the blog.

Here is Fino's formal introduction and statement on the model.

Dear all,

I would like to take this opportunity to briefly introduce myself and my project.

My name is Fino, and I have been working in the IT sector for almost 10 years, specializing in DevOps in a technical teamleader position. I have also been an enthusiastic Stronggirls fan for the past 10 months.

Recently, I have been collaborating with Jon to improve the quality of Stronggirls and provide him with various "secret sauces." You may have already seen the results in Jon's recent DevLogs. I believe we are on a good path together, and we are both enjoying the process of elevating the Stronggirls universe to the next level.

During the development work, I realized that I could think beyond my usual scope. I have been considering how to offer creators and communities good technical possibilities for creating high-quality AI FMG content.

I quickly arrived at a solution: A custom SDXL model and matching LoRAs for a solid foundation. Since the development of such models requires a significant investment of time and money, I would like to invite you to join a community project aimed at accelerating targeted development.

The goal is to create technical models that can be freely used privately by the community and used for commercial purposes through an individual donation (not a fortune ;) ).

The focus of these developments should be on the needs of the community to provide targeted technologies. As this all sounds quite visionary at the moment, I have prepared an FAQ section to make these ideas more tangible.

I look forward to your feedback on these ideas for eventual further steps and would like to thank Jon for allowing me to use his platform and engage with this wonderful community.

P.S. I am, of course, available for further questions via PM, Discord server, or a personal conversation.

Thank you very much!

-@fino

Anybody interested in this project and wish to support Fino in developing more FMG AI models, you can hop on the discord and chat with @fino. I created two new channels for this topic specifically.

Quick FAQ from Fino:
Why do you want to train a model? Civitai is full of good models!

Civitai offers a great variety of good models. Unfortunately, these models are mostly not trained for FMG, so the prompting is always gambling and you need to use a stack of LoRAs to compensate. I want a model that works out of the box and uses a free-to-use dataset to avoid problems like legal issues.

What if we train only a LoRA?

A LoRA can be a good way to address the problem, but it is also technically limited and the results can vary strongly from model to model. Ask Jon!

What happens if you don't maintain the model anymore?

For that reason, I will try to find a new maintainer to fork the project. All sources for the model training will be provided to the new maintainer(s).

Why should I pay you money? All models on CivitAI are free!

Please consider that great models like Pony or Realistic Vision get huge monetary support to be trained. The training time is much longer and you mostly need cloud resources. Also, you need to make several training attempts. Unfortunately, there doesn't exist a recipe for all use cases.

What are my benefits as your patron?

With the support level, you will get access to the newest beta models, loras and weekly devlogs. I will try to handle the communication like Jon (never change a running system ;) ). With the super support level, you have, on top, access to polls that influence design decisions for the model. In the end, the model should primarily fit the needs of the community!

Why is the model not free to use for all purposes?

Private persons can use the model without restrictions (except for legal issues or harming other persons). Creators who use the model for commercial use need to pay an individual donation. This ensures that we get resources to develop the model further! In the end, it will be a win-win situation.

What will be the base model?

The model will be Pony or a Pony-based model. Pony provides flexibility, good license terms, and a big ecosystem.

Why not SD3?

Yes, the license has changed for the better and the quality is improved. Please consider that at the moment, the 3.1 version is not released, which is essential, and we have no training experience or an ecosystem around that. Maybe the model will be attractive enough next year.

At the end, the one important political question...?

Naomi/Kasumi simp forever. No discussion :D