Lets say for example that SD1.5 could perfectly generate Naruto in kishimoto/shippudenanime's style. Even if it could make naruto as perfect as that there are like hundreds or thousands of model/merges that probably can't because of mixing/training, and in those cases you can have a 144 mb (you know the pretty standard lora size) file to recreate naruto in the style (at least somewhat close).

But lets come back to reality, can you really just recreate naruto in that exact style using just SD1.5?

@merra none of which is mentioned in the description. Proper tags for available clothing, accessories and the like are essential for characters like this.

@whyusecivit 144mb is enormous for anime LoRAs, and you very, very rarely even need anything more than textual inversion for those - SD contains ample information on how to draw anime. Most of the time the whole training dataset is far smaller than 144mb...

@merra model dimensions are equivalent to the LoRA size. Being that your model is 144 MB, you probably made a 128 dimension LoRA. It's not a big deal. Take a look at the SuperMerger extension for the A1111 repo, it has a button to remake the dimensions. Not that many people are complaining about 144MB unless their connection is stuck at 2004 dial-up speed. I do agree with OP tho, that more information on your training would be helpful, such as useful prompts and so on. Anyway, thanks for being a creator and giving us free content to enjoy :)