Note: Preview images are generated without the use of LORAs. Some inpainting might be present.
I rarely use 2D models since I prefer realistic and 2.5D artstyle, but I still need a 2D model just in case I wanna use it. And thus, this merge was born. However, I'm not really that into anime artstyles and western style animations are a little too flat and simplistic. So for this model, I want to find that middle ground between anime and western animation artstyle. This model also have that hand drawn digital art feel to it, where the line art is rough and bold, and the shading is a mix of soft and cell shading, which I really like. It might not be to everyone's taste, but for those who are interested feel free to try it out..
These are just some examples of the settings I usually use and it's not really a hard requirement so you're free to experiment and see which setting works best for you:
Sampling Method: DPM++ 2M Karras or DPM++ 3M SDE Karras
Sampling Steps: 25 - 50
Hires. fix: On (not optional)
Upscaler: R-ESRGAN 4x+ Anime6B
Hires Steps: I usually put it at 0 so it'll be the same as Sampling Steps
Denoising Strength: 0.25 - 0.35
Upscale: 1.5 - 2
CFG Scale: 5.5 - 7
Try to keep the prompts short and concise. No need to fill up the prompts with unnecessary wording that does the same thing like high detail, ultra detail, absurdres, hires, photorealistic, hyperrealistic, etc. Especially realism prompts since its 2D so you don't need it. Keep it simple.
Also, avoid putting multiple ????? tags like 'erection, ?????, ????? ?????, large ?????, etc' because you might get exaggerated ????? proportions or multiple ???????. Sometimes, just a simple '?????', 'erection' or 'flaccid' is enough.
ADetailer is a must if you want to get good looking faces. Without it, you will get distorted or blurry looking faces. I also highly suggest using this ????? ADetailer if you're working with ????? stuff, but only if its necessary. Sometimes it'll do more harm than good.
Hires fix is mandatory. I don't know if its a common thing for 2D models or is it just mine, but images tend to get distorted a lot, especially far away faces. So hires fix is absolutely necessary.
If you want finer details, then after generating with Hires Fix then send it through img2img with the same prompts and settings with denoising strength set around 0.3-0.4.
Use negative embeddings instead if writing paragraphs of words into the negative prompts. Here are some examples of the ones I usually use:
v4.0
What's new:
I tried to replicate an artstyle that's somewhere between v2.0 and v3.0-alt
Easier to create more 'bara' characters with exaggerated anatomy
This includes ????? as well, so if you want regular sized ????? you might need to use lora.
1. I diritti dei modelli ripubblicati appartengono ai creatori originali.
2. I creatori originali che desiderano reclamare il proprio modello devono contattare il personale di SeaArt AI tramite canali ufficiali. Clicca per rivendicare
