Jeshky
Newbie
- Oct 29, 2022
- 23
- 20
- 79
Ren'Py Story of Stephanie [Week 1] [LustyWhisper]
- Thread starter LustyWhisper
- Start date
StrongEpsilon
Newbie
- Apr 29, 2025
- 95
- 295
- 63
Is she a shapeshifter? I'm just curious.
You don't have permission to view the spoiler content.
Log in or register now.
Yes, she has at least 40hrs on her Steam in game Among Us in last two weeks.
Torchwood2000
Newbie
- Sep 24, 2020
- 21
- 25
- 94
Looks good to me. I'm curious how the story develops.
Maybe a way to enlarge the text would be good
Maybe a way to enlarge the text would be good
#SHORTNEWS FOR WEEK 2 UPDATE:
The story is finished.
All that remains is to finish the illustrations.
Game will stay vertical for now 735:1080.
Text will be enlarged.
The story is finished.
All that remains is to finish the illustrations.
Game will stay vertical for now 735:1080.
Text will be enlarged.
Hi, thanks for the feedback. I plan to release regular updates. But I'll see how it goes; first, I need to verify my account with Subscriberstar. Then I plan to start regular publication. As for the Mac version, it will definitely be included in a future version. However, the game is currently new with only one update, so I don't want to rush anything. I'd rather collect feedback at this point.
Kryptozoid
Well-Known Member
- Sep 3, 2016
- 1,239
- 2,650
- 503
I won't comment on the game as I have not played it, I just looked at the pictures and I'm not a noob on AI image making so I will explain a few things :
It looks like you're making your pictures locally with Stable Diffusion Auto1111 or Forge or ComfyUI and you're using some SD 1.5 model or maybe SDXL. If I'm wrong and you're using some online service, stop it and do it locally (you can still use online GPU if yours sucks).
So, main problem is the face, eyes are fucked up most of the time and face look slightly different from pic to pic.
1\ solutions for crossed eyes syndrome :
a/Use ADetailer (plugin for Auto1111 and probably others). It will not fix face looking slightly different from picture to picture, but at least it will make it sharper and more detailed.
or
b/Use Topaz Photo AI's face improvement AI (150-200$ or free on w@r3z l33t haxxors forums) Super easy to use, detect faces automatically, you just have a slider to move to decide how much of the effect is applied. It will also not make the face consistant from pic to pic. And it does a lot of other things like AI upscaling, color grading, etc... all simple to use.
2\Solution for face not looking like the same person every time :
Use a control net called ip-adapter face something V2, there's 2 Loras going with it, one for SD 1.5 and one for SDXL, that you can include in your prompt for more weight control, but it also works just changing the weight value in the plugin options.
How it works : you give 1 (or more) pictures of the character you want to the ip-adapter, then set the weight either in the prompt itself, or the control net, or both, until it satisfies you. (usually 0.7 is fine) And now your character is consistent. If you use a real person as base, make sure you use pictures where she's roughly the same age, doesn't wear super weird makeup etc... otherwise it will still look slightly different each time. The best solution is to find a photoshoot of someone, where angles and facial expressions or clothes are different, but makeup and age are consistant.
Also I'd recommend the model TheSuperRealistic V3 for SD 1.5 (maybe there's V4 or 5 these days, but 2-3 months ago v3 was still the best).
Of course look for tutorials on youtube or Civitai if something I said isn't detailed enough.
And don't hesitate to combine several Loras (I don't mean character loras but more light lighting style, places, effect, clothes etc...) because for now your images look very generic. This advice is only applicable to SD 1.5, don't even try with SDXL, it will always look more generic because it doesn't like long prompts and it doesn't have any finessing with multiple Lora weights, it's always 0% or 100% effect. It's a subtle as a rock. So stay on the latest good photoreaistic Sd 1.5 models.
Last thing, I don't know what's the story, but at some point it looks like the main girl is at school, and she looks about 10 years older than she should be. Maybe she's a teacher I don't know. Anyways, unless it's a "29yo milf story" I'd suggest to make her look younger.
And second last thing : go in the settings of Auto1111 or whatever you use and tell it to render in .jpg, not .png which takes unnecessary space.
It looks like you're making your pictures locally with Stable Diffusion Auto1111 or Forge or ComfyUI and you're using some SD 1.5 model or maybe SDXL. If I'm wrong and you're using some online service, stop it and do it locally (you can still use online GPU if yours sucks).
So, main problem is the face, eyes are fucked up most of the time and face look slightly different from pic to pic.
1\ solutions for crossed eyes syndrome :
a/Use ADetailer (plugin for Auto1111 and probably others). It will not fix face looking slightly different from picture to picture, but at least it will make it sharper and more detailed.
or
b/Use Topaz Photo AI's face improvement AI (150-200$ or free on w@r3z l33t haxxors forums) Super easy to use, detect faces automatically, you just have a slider to move to decide how much of the effect is applied. It will also not make the face consistant from pic to pic. And it does a lot of other things like AI upscaling, color grading, etc... all simple to use.
2\Solution for face not looking like the same person every time :
Use a control net called ip-adapter face something V2, there's 2 Loras going with it, one for SD 1.5 and one for SDXL, that you can include in your prompt for more weight control, but it also works just changing the weight value in the plugin options.
How it works : you give 1 (or more) pictures of the character you want to the ip-adapter, then set the weight either in the prompt itself, or the control net, or both, until it satisfies you. (usually 0.7 is fine) And now your character is consistent. If you use a real person as base, make sure you use pictures where she's roughly the same age, doesn't wear super weird makeup etc... otherwise it will still look slightly different each time. The best solution is to find a photoshoot of someone, where angles and facial expressions or clothes are different, but makeup and age are consistant.
Also I'd recommend the model TheSuperRealistic V3 for SD 1.5 (maybe there's V4 or 5 these days, but 2-3 months ago v3 was still the best).
Of course look for tutorials on youtube or Civitai if something I said isn't detailed enough.
And don't hesitate to combine several Loras (I don't mean character loras but more light lighting style, places, effect, clothes etc...) because for now your images look very generic. This advice is only applicable to SD 1.5, don't even try with SDXL, it will always look more generic because it doesn't like long prompts and it doesn't have any finessing with multiple Lora weights, it's always 0% or 100% effect. It's a subtle as a rock. So stay on the latest good photoreaistic Sd 1.5 models.
Last thing, I don't know what's the story, but at some point it looks like the main girl is at school, and she looks about 10 years older than she should be. Maybe she's a teacher I don't know. Anyways, unless it's a "29yo milf story" I'd suggest to make her look younger.
And second last thing : go in the settings of Auto1111 or whatever you use and tell it to render in .jpg, not .png which takes unnecessary space.
Last edited:
Thanks for the detailed advice. This is my first game. I'm more interested in the story. However, I hope the visuals will improve over time. I'll save it somewhere and hopefully work my way through it over time. Thanks again!
Kryptozoid
Well-Known Member
- Sep 3, 2016
- 1,239
- 2,650
- 503
Nothing wrong about caring for the story, but it's visuals that will make people click on it first. So you do you, but for a game made for men, the deal is nail the visuals first, then make sure the story doesn't suck so people keep playing the game. And yes you will improve over time for sure, unless visuals are forever secondary to you.