MiltonPowers

Twins Basil! Twins!
Donor
Jr. Uploader
Jul 26, 2023
19,921
55,998
973


00020-3025175423.png


An update on the state of play. Naturally with the time of year as it is, there has been a fair bit of disruption to my normal routine. However, progress is still pretty good. The first 2 debriefs are almost entirely complete. The last one is fully scripted and most of the image assets are created. I need to create a few animations and then get them all into the scripts. Once that is done I will do the sound effects, but that wont take long as there isn't much by way of action in these debrief scenes. All three come in at about 10000 words, so it's a shorter update so far.

Then I should have time for a visit to Isabella, where I will get to try out some new tools... Over the weekend and also this morning, I downloaded and setup WAN2GP, one of the easier to use varieties of animation generation tools but it allows me to use WAN 2.1 and WAN 2.2 and the layout is quite similar to framepack and stable diffusion, so the learning curve hasn't been too steep. I have both models downloaded and setup. As with SDXL for image generation, you can download addons for these tools which allow them to generate certain things. These are called LORAs. They are like little booster packs. In SDXL I use them sometimes for specific lighting effects, controlling hair length or styles for some characters, changing how clothes interact with skin and body shape, sometimes for controlling breast shape etc. For animation generators they do things like control camera motion, how characters move, special effects and so on. They are essential for NSFW animation generation. So most of the time has actually been spent finding and downloading these LORAs. It isn't just a case of downloading them, they all have different prompts that need to be used to activate their content and some need different generation settings. Helpfully most of the authors don't seem to understand the impact of their file naming conventions on usability... many of the filenames come down as "fp4842903827fdasf.safetensors". So coming back to the file a week later, it will mean nothing to you. Some actually have the thought to name the file after the name of the LORA. But either way, to make sure they can be useful I keep a spreadsheet of all the file names, their activation criteria, the actual name of the LORA and then any other settings or usage guidance there is for the file. Building that spreadsheet took a fair bit of time as I've downloaded hundreds of these files now.

I also discovered that they were available for Framepack. Framepack isn't it's own thing, it's a wrapper for an open source video generator called Huanyan. I didn't realise that, so when I first got framepack and looked for LORAs and discovered none, i figured there were none, or only very few. Turns out there are loads. So I've also been doing the same for framepack. Which is good, because not all the same functionality can be found between the 3 models - framepack and WAN 2.1 and 2.2. Image quality and prompt adherence is definitely better on WAN, but it's slower to create animations (about 50% slower in fact). I now understand how and where a lot of the other, much better, NSFW animations were coming from on other games. So... I can do that now. So I'll have a go at that with the Isabella content I'll be working on next.

I go away mid January for a week, so I'm aiming to get the early access release out for around the 8th January. There are a couple more example images from 0.45 included here (not yet processed, her face needs to be re-processed to enhance the detail and fix the eyes). At some point soon I'll put up some example animations from WAN that show the difference in what it can do as well.

With that, I will wish you a Happy New Year!

openart-image_B7cm5i0W_1767195128643_raw.png openart-image_bIe2w3C4_1767195858749_raw.png openart-image_MmnUhTPc_1767193768674_raw.png
 

DarthSpitz717

Active Member
Mar 28, 2023
769
1,433
317


View attachment 5584789


An update on the state of play. Naturally with the time of year as it is, there has been a fair bit of disruption to my normal routine. However, progress is still pretty good. The first 2 debriefs are almost entirely complete. The last one is fully scripted and most of the image assets are created. I need to create a few animations and then get them all into the scripts. Once that is done I will do the sound effects, but that wont take long as there isn't much by way of action in these debrief scenes. All three come in at about 10000 words, so it's a shorter update so far.

Then I should have time for a visit to Isabella, where I will get to try out some new tools... Over the weekend and also this morning, I downloaded and setup WAN2GP, one of the easier to use varieties of animation generation tools but it allows me to use WAN 2.1 and WAN 2.2 and the layout is quite similar to framepack and stable diffusion, so the learning curve hasn't been too steep. I have both models downloaded and setup. As with SDXL for image generation, you can download addons for these tools which allow them to generate certain things. These are called LORAs. They are like little booster packs. In SDXL I use them sometimes for specific lighting effects, controlling hair length or styles for some characters, changing how clothes interact with skin and body shape, sometimes for controlling breast shape etc. For animation generators they do things like control camera motion, how characters move, special effects and so on. They are essential for NSFW animation generation. So most of the time has actually been spent finding and downloading these LORAs. It isn't just a case of downloading them, they all have different prompts that need to be used to activate their content and some need different generation settings. Helpfully most of the authors don't seem to understand the impact of their file naming conventions on usability... many of the filenames come down as "fp4842903827fdasf.safetensors". So coming back to the file a week later, it will mean nothing to you. Some actually have the thought to name the file after the name of the LORA. But either way, to make sure they can be useful I keep a spreadsheet of all the file names, their activation criteria, the actual name of the LORA and then any other settings or usage guidance there is for the file. Building that spreadsheet took a fair bit of time as I've downloaded hundreds of these files now.

I also discovered that they were available for Framepack. Framepack isn't it's own thing, it's a wrapper for an open source video generator called Huanyan. I didn't realise that, so when I first got framepack and looked for LORAs and discovered none, i figured there were none, or only very few. Turns out there are loads. So I've also been doing the same for framepack. Which is good, because not all the same functionality can be found between the 3 models - framepack and WAN 2.1 and 2.2. Image quality and prompt adherence is definitely better on WAN, but it's slower to create animations (about 50% slower in fact). I now understand how and where a lot of the other, much better, NSFW animations were coming from on other games. So... I can do that now. So I'll have a go at that with the Isabella content I'll be working on next.

I go away mid January for a week, so I'm aiming to get the early access release out for around the 8th January. There are a couple more example images from 0.45 included here (not yet processed, her face needs to be re-processed to enhance the detail and fix the eyes). At some point soon I'll put up some example animations from WAN that show the difference in what it can do as well.

With that, I will wish you a Happy New Year!

View attachment 5584786 View attachment 5584787 View attachment 5584788
___________________________________________+++++++++++++++_______________________________

Great artwork even though it's not finished

The whole story still is very interesting as well as the world build an interesting combination of various historical era/tech to make it
 
  • Like
Reactions: Jubs!

BigAlzBub

Active Member
Jul 20, 2020
547
514
309
Don't know if this is of any use to you Jubs, but I've had a little play with this ComfyUI workflow and the results were fairly impressive.

I only have 12 Gb of ram on my video card, and I think after your upgrade you have 16? which should mean you can use it without having to resort to the quantised versions.
 
  • Like
Reactions: Jubs!

Jubs!

Member
Game Developer
Mar 29, 2018
174
500
311
Don't know if this is of any use to you Jubs, but I've had a little play with this ComfyUI workflow and the results were fairly impressive.

I only have 12 Gb of ram on my video card, and I think after your upgrade you have 16? which should mean you can use it without having to resort to the quantised versions.
Thanks I will have a look. Although my bandwith is all tied up downloading WAN 2.2 loras today.. about a hundred of them to download and catalogue, nearly finished...

I'm using WAN2GP to access WAN and the generation times allow for 60+ seconds. It's really just 8 second chunks stitched together. But you can give it a video and ask it to continue the video as well, so you can sort of get much longer than that. Framepack studio has similar functionality as well. I'm quite looking forward to playing around with wan 2.2 and its ability to take in a scene + 2 or 3 people and have then placed in the scene, which is possible via WAN2GP (and comfy), that could be really useful .
 

MiltonPowers

Twins Basil! Twins!
Donor
Jr. Uploader
Jul 26, 2023
19,921
55,998
973


045earelease4.jpg

The early access version of 0.45 is currently being uploaded. I'll put the link up shortly for an early early access. I go away on Friday for a week, so the the public release won't be until the evening of Thursday 15th.

0.45 patch notes:

Added 3 x Zara mission debrief events.
2 x codex entries added for the neighbouring countries of Cantabria and Uchos.
Adjusted the text to have thicker borders so that it is easier to read.
A few spelling and grammar fixes.
Tidied up a few jarring image transitions in older content.
Added in 5 second fade-to-blacks at the end of each event, as a clearer indicator that the event is ending.
Updated the hub screen so that if there is no content available, the contracts or character events buttons will be greyed out, rather than just not being there at all.


In terms of stats, the 0.45 update adds:
About 13,500 words of story & 1500 words of lore added to codex.
About 1300 lines of scripting.
49 images of Zara.
23 x animations of Zara.
2 x images of Alice
2 x images of Sarah
2 x animations of Sarah.
6 x backgrounds.

This is a smaller update than usual, not in terms of the length of content, but these are simpler scenes without costume or location changes and no need for varied sound effects or more complicated image processing. Which is why I chose to get these debrief scenes done over the Christmas period. I had hoped to get a romance scene in with Isabella, but having spent most of the day on Sunday working on it, I realised there was no hope of getting that into this update without a big delay. So that is all written (6000 words, but it will need editing down), the SFW assets and backgrounds are all done and I have the skeleton images for the NSFW scene all ready to start processing for the next update. Which won't be a full contract, it will be another set of 'smaller' character events. This one with Isabella, one with Alice & Lilly and one with either Heidi or Lyra are what I am aiming for.

I've spent a lot of time mucking around with WAN2.1 and WAN2.2 ... it's quite a bit more complicated and varied to use than Framepack and results are very inconsistent, which is frustrating when combined with the longer generation times. It's definitely going to be good for improving the NSFW animations and also some of the magical stuff. I have gone back over some of the older animations to see if I can improve them, with mixed success. There is definitely a learning curve here.

Another snag is the new 5080 graphics card has developed a rattle (thanks Zotac and your shitty cheap fan supplier). Now I'm awaiting to hear on an RMA return for it. That has the potential to really screw with image and animation generation. I cant just plug my old 3080 back in and start generating images again, all the software is different and produces different results. I'm hoping I can get the card sent back whilst I'm away so that the replacement is here when I get back.

The screens here are a mix of stuff from 0.45 and some of the early, unprocessed stuff that I made for the Isabella scene that is going in the update after that.

The links for early access should go up in a couple of hours - it takes me about 1.5 hours to upload each version and there are 3 versions. I'll put the link up when the Windows & Linux version is up, then Mac version will follow about 1.5 hours later and then hopefully the android version about 1.5 hours after that.

 

BigWilli

Newbie
Nov 3, 2022
69
80
160
Jubs! Ever since I found this it's been one of my absolute favorites! Definitely threw some money towards it. Love the fantasy plot, the beautiful women (where with each one I'm going "I want to see her next, and her, and her...") that are also different, and the well-done graphics. I appreciate the effort put into it - I've tried many other similar-type endeavors, and with respect to them, this one feels so far more like a sexy regular video game with the depth, like you could take a controller and look around. A truly solid AVN that has been very enjoyable to play.
 
Last edited:

Jubs!

Member
Game Developer
Mar 29, 2018
174
500
311
Jubs! Ever since I found this it's been one of my absolute favorites! Definitely threw some money towards it. Love the fantasy plot, the beautiful women (where with each one I'm going "I want to see her next, and her, and her...") that are also different, and the well-done graphics. I appreciate the effort put into it - I've tried many other similar-type endeavors, and with respect to them, this one feels so far more like a sexy regular video game with the depth, like you could take a controller and look around. A truly solid AVN that has been very enjoyable to play.
Thank you, that's nice to hear!
 

Sciguy77

Newbie
Oct 23, 2025
22
18
54
Probabiy NOT the place to post this but forum searches don't help a whole lot. I haven't done any modeling since Sexy Beach with another user's software out of the Netherlands. What's the easiest way to get started with AI CG? If I'm gonna learn this, I wanna make sure I don't invest in something that'll be obsolete in 6 months, but worth the investment as a rank beginner. Using a GTX 1660
 

-CookieMonster666-

Message Maven
Nov 20, 2018
14,149
21,382
1,057
Probabiy NOT the place to post this but forum searches don't help a whole lot. I haven't done any modeling since Sexy Beach with another user's software out of the Netherlands. What's the easiest way to get started with AI CG? If I'm gonna learn this, I wanna make sure I don't invest in something that'll be obsolete in 6 months, but worth the investment as a rank beginner. Using a GTX 1660
You should post a new thread over on the Tutorials board. Then folks there who know about this type of thing (and not Jubs only) can see your thread and answer if they have suggestions. Just click the appropriate button in the top right of that board:

1767911057705.png
 

Jubs!

Member
Game Developer
Mar 29, 2018
174
500
311
Probabiy NOT the place to post this but forum searches don't help a whole lot. I haven't done any modeling since Sexy Beach with another user's software out of the Netherlands. What's the easiest way to get started with AI CG? If I'm gonna learn this, I wanna make sure I don't invest in something that'll be obsolete in 6 months, but worth the investment as a rank beginner. Using a GTX 1660
I don't mind giving advice, but as I'm going away tomorrow I don't have much time. My development updates on here and the patreon page (either open access or free membership access) detail my experiments and learnings with AI generation.

But my knowledge isn't the best, it's just been done through trial and error. I'm pretty sure that for this kind of thing, if you don't want to pay for every image you generate (which would be horrendously expensive the way that I do things), then you need the open source free generators and that is pretty much Stable Diffusion. You can run framepack on a 6gb VRAM card as well, but it would probably be a bit slow. There are different varieties of SD - 1.5, SDXL and Pony are the most populat. 1.5 is the older one that a lot of people stuck with because it had loads of LORAs setup for it - I never got on with it, found the mutations and hallucinations really bad. I'm sure that's partly down to how I was using it, but ... it wasnt for me.

SDXL is the more recent version and then there is Pony which is a version of SDXL that is more focused on cartoon stuff and so has more NSFW stuff in it. It can do realism, but for realism you want SDXL. I use the WebUI version (Automatic1111 I think it's sometimes called), which is relatively easy to setup and use, but not as versatile as the comfyUI version. If you have questions, I'd recommend asking Gemini , as it will provide you with advice and recommendations based on the scenario you give it.

Also, I have a subscription to OpenArt, which is where I do all the hand fixing and background stripping but it also give sme access to pretty much all the image and video generation (and sound, but i havent played with that) tools via integrated APIs. I think it is by far the most cost effective way to get access to all the different 'paid for' tools.
 

Sciguy77

Newbie
Oct 23, 2025
22
18
54
I don't mind giving advice, but as I'm going away tomorrow I don't have much time. My development updates on here and the patreon page (either open access or free membership access) detail my experiments and learnings with AI generation.

But my knowledge isn't the best, it's just been done through trial and error. I'm pretty sure that for this kind of thing, if you don't want to pay for every image you generate (which would be horrendously expensive the way that I do things), then you need the open source free generators and that is pretty much Stable Diffusion. You can run framepack on a 6gb VRAM card as well, but it would probably be a bit slow. There are different varieties of SD - 1.5, SDXL and Pony are the most populat. 1.5 is the older one that a lot of people stuck with because it had loads of LORAs setup for it - I never got on with it, found the mutations and hallucinations really bad. I'm sure that's partly down to how I was using it, but ... it wasnt for me.

SDXL is the more recent version and then there is Pony which is a version of SDXL that is more focused on cartoon stuff and so has more NSFW stuff in it. It can do realism, but for realism you want SDXL. I use the WebUI version (Automatic1111 I think it's sometimes called), which is relatively easy to setup and use, but not as versatile as the comfyUI version. If you have questions, I'd recommend asking Gemini , as it will provide you with advice and recommendations based on the scenario you give it.

Also, I have a subscription to OpenArt, which is where I do all the hand fixing and background stripping but it also give sme access to pretty much all the image and video generation (and sound, but i havent played with that) tools via integrated APIs. I think it is by far the most cost effective way to get access to all the different 'paid for' tools.
Copying this reply to my rapidly growing library of tips and tricks. I'm still reeling from the scary power inherent in AI CG tech, the freakiest being its ability to create scenarios straight from text. My first calculator had 4 functions, vacuum tubes for the display for chrissakes, and it was a huge step up from my Keuffel & Esser Ivory slide rule. I retired last year and REALLY need a new hobby. Making and/or tweaking games trapped me in its event horizon for almost a decade. If you don't get that yet, you will soon (attention to detail demands it). Never got my own game off the ground. Prolly never will, I just need to play with a new toy.
Thanks Jubs!
 

Hatchet80

Newbie
Mar 22, 2023
53
93
86
I'm not being argumentative here. I just wanna provide my thoughts on the situation because I grew up with violence every day and I might be able to shed some light on some contention about how the MC reacts.

I just dont agree about the MC being able to kill without it affecting him. Most people these days couldn't kill a chicken to eat it without feeling bad or sick about it, I certainly would feel that way. Soldiers undergo a minimum of 6 months training, usually more before they are deployed and they are people that volunteered for it. If there are active warzones they can reasonably expect to be involved in them and more effort will have been spent preparing them for it. They also tend to kill at a distance, by pulling a trigger, not right next to them by cutting their throats. He also witnessed other horrific sights, it isn't just the fact he killed someone, it's the whole experience of that rescue mission. I don't think a well educated western teenager dropped into that situation would shrug that off without some kind of mental trauma. He does get better at dealing with it, as time goes on, as you would expect as someone is exposed to more of it. Lilly also helps him with it.
This makes sense to me. It was repeated that MC lived a pretty good sheltered life before coming into the new world. The most violent thing he experienced is rugby, so for him to be affected like he was makes total sense. Even after months of training, taking a life in CQB would be jarring to someone like him. If he grew up like I did, then I could see the argument that the affect on his mental state wouldn't be as bad.

On a side note, soldiers don't get that much training in USA. I realize that the USA isn't the only western nation, so you might not be talking about us, but our soldiers are usually considered better trained excluding special forces. Those people, no matter where in the world, get massive amounts of training comparitively. 3 months of boot camp for USMC which has the longest boot camp in the US armed forces by double the amount of the Army which is 1.5 months long, which is the second longest boot camp in the US armed forces. Then comes ITB (Infantry Training Battalion). That can be longer depending on your MOS (Military Occupational Specialty), because you're learning your specific job. Your estimation of 6 months of training is close depending on what your MOS is, but ONLY if you explicitly talk about Marines. Any other branch, there is a high probability that they didn't get that much training. You said soldier, which in USA speak, means Army. USMC makes Marines, US Army makes soldiers, US Navy makes sailors, and US Airforce makes....well not just pilots and mechanics. IDK what they make. Military members who consider bicycle riding as PT (physical training)? OK OK OK. I can't be that harsh to our flyboys. They make the best damned pilots on the planet....you flyboys happy now? Real talk, air support is essential to ground forces. I appreciate every pilot in the US military.

Hopefully, if you weren't talking about US armed forces, you view this little tangent as a learning opportunity (which is what it is intended as) rather than seeing it as American ego. Maybe this info can help you in the future.

BTW Dev, the harem tag isn't implemented yet but in the planned section, so does that mean that if I choose a girl, I'm temporarily locked out of the others?
 
Last edited:
4.10 star(s) 18 Votes