About This Model
This model was commissioned by a user who reached out to me; I have a few different versions that I felt were mostly inconsistent. This version is what I feel is most authentic to Nowa's style, in that you can create images with a similar layout as well as style.
At the current time, I consider V3 to be the definitive version of this style LoRA. You might be wondering: "Twisted, why in God's name did you upload so many different versions of this style?".
Long story short, I have been slowly but surely doctoring every image in this dataset to completely remove text, speech bubbles, and page numbering that might pollute or disturb your generations. Now that the dataset has been cleaned in its entirety, there is not a single image within this LoRA containing a speech bubble or a wall of text.
Recommendations
Please keep in mind that this positive prompt is only necessary if you wish to replicate his NSFW acts. The syntax below is using the Dynamic Prompts extension for A1111/ForgeUI; you can also just choose whichever you want to generate or you can engineer your own prompts.
If you would like to add the "PLAP" and "Ah!" SFX/VFX just add plap to prompt. Some weighting might be required.
Positive Prompt:
plap, {before and after, full nelson|before and after, mating press}Negative Prompt:
speech bubble, text, page number, watermark
Possible Issues
Sometimes images with
multiple viewsorbefore and afterlayout struggle to generate results that align with expectations.
Description
No speech bubbles or text (aside from PLAP! sound effect) exist in dataset; if these are appearing then it might be related to the checkpoint you are using.
Page numbering issue from V2 has been resolved; depending on which checkpoint you are using some might try to make the trigger word into a watermark. Include
watermarkin negative prompt to mitigate this issue, if it presents itself.