Development Update 19/10/2026

1.02R (well, 1.021R now) has been released to early access. The public version will go live on 28th October. The patch notes for the update are at the bottom of this post.

The last week saw me putting the finishing touches on the first cartographer's guild mission and building the Zara debrief. That went pretty well, SeeDream, while far from perfect, is certainly getting much better at not screwing around with the background image, so I've had less instances of random items being replaced with other things. Which has meant I was able to build the debrief in a day.

I lost a day when I got notification that the online AI editing service that I use has formally set the date to which it is retiring all the stuff that actually gives it a USP over what Google and every other big name AI image editing tool offers. Mid November. That's frustrating. A lot of these tools use chat to edit, and I despise it. I've talked about it before on here, so I won't go into it again. Suffice to say, i find chat to edit AI editors to be either incompetent or maliciously compliant and they are a stressful and unpleasant tool to use. The tools they offered that I liked were built around precise inpainting and built on the Stable Diffusion platform. Primarily - hand fixing, image extension, object removal, background removal and occasionally inpainting items. They weren't perfect, but they were cheap to use, only edited the specific bit of the image I told it to and had no content filtering. They do have replacement options but they have content filtering and cost 4x as much for the same thing. That's progress for you.

So, I've cancelled my subscription with them and gone with another provider. I lost a day researching and testing out alternatives. I found another provider that allows NSFW content, provided you pay the higher subscription tier. But that's ok, because I will use the benefits of that tier anyway. It also includes SeeDance, WAN3, Hailuo and their own propriety image and animation generators with NSFW added. Added bonus, they have the slightly lower image quality but cheaper versions of SeeDance, so that helps spread the cost a little bit. The hand fixing they offer is crap. This is the hardest bit, the service I used had a really good hand fixing tool. I'm still not sure how they did it. It's built in stable diffusion, and I can add hand fixers into my image gen stuff via ADetailer, but it's nowhere near as good (although free to use and can brute force solutions but churning out lots of attempts). That's also using Stable Diffusion architecture. I initially tested it on SDXL, but it seems Illustrious works better. Anyway, I have a solution. Image extension can be done via this new service - this is used when a source image has anatomy cut off and I need to extend the image to complete the full image of the person. Background removal I've setup a piece of software called IOPaint, via the Pinokio app manager. This is a nice little tool which you can access a lot of different background removing models with. I eventually found what is probably the best general purpose one. It removes the background about 4x as fast as the online subscription service did, so that's nice. It also comes with an object removal tool that is much faster and seems much better than the subscription services one, so that was two birds with one stone. So, I lost a day, but in return I got access to more and probably better overall tools.

I spent Sunday sorting out some variable stuff and also went through and adjusted every image in the game. I added to each image a reduced brightness and slightly increased contrast. AI image generators tend to produce very clear, very well lit images. Which I do want for sprite creation because if you have all sorts of fancy dramatic lighting, the sprite can look weird when put into a different scene. However, by default the end up looking a little washed out as a result. So I've adjusted this to deepen the colour and darken the image. I think it looks much better. It's more 'cinematic', but I've not gone all Christopher Nolan and made everything too dark to see and all the sound too hard to hear. It was one of those changes that I sort of stumbled on, thought it would be a good idea and then spent hours doing it. I had to adjust each image every time it's called by renpy, as I didn't want a blanket rule in place in case it wasn't right for some of them (which it isn't). I did setup blanket rules for the side images and the galleries though.

For the next update I'm aiming to do Isabella's slave raid follow-up event and the check-in with Emelia and Lilly. I say this every time, but I'm hoping these won't take too long as they are quite controlled and specific events. I already have all of Isabella's & Emelia's backgrounds done, it's taking less time to make NSFW scenes and the quality is improved with the new tools and there are aren't big variable choices that can send you on multiple versions of the same event. So... I'll see but I think maybe 3 weeks for this one.

Some screens from Zara's added event added this week.

Full 1.01R->1.021R patch notes:

Not including replacement sprites and animations, this update adds:

*He also appears in a fair few of Emelia's images and animations.