Please read through the entire description (might need to be expanded) and the version change notes as they cover a lot of information about basic use cases and limitations. The description is always focused on the latest version. Thank you!This is an SD 1.5 LoRA for the character Cynthia / Shirona from Pokemon. It works for models based on SD 1.x only, it will not work for models based on SD 2.x or SDXL!Example images were picked from pure 512x512 txt2img results and then re-created at 1024x1024 using txt2img with moderate Hi-res Fix settings (upscaler Lanczos, denoising strength 0.4). This improves faces and other details that are almost impossible to get correctly and consistently at 512x512 (limitation of the technology) while still giving a realistic impression of what it looks like. Pure 512x512 results will have more distorted faces and less detail.Please see the version change notes for the training and example image generation models as well as the used weights as they might change between versions. Remember that you might need to adjust weights to best suit your use case!Remember to add the tag cynthia \(pokemon\) (with backslashes intact) to your positive prompt.The training set contained multiple variations of Cynthia's signature look so you might need to put combinations of the following tags in your positive or negative prompts to get the desired results. Some should probably be used together (separated by plus in this list):<ul><li>fur collar</li><li>fur-trimmed coat</li><li>hair ornament</li><li>hair over one eye + one eye covered</li><li>high heels</li><li>pants</li><li>shirt + v-neck</li></ul>As the training set contained a few images with floating hair, you might need to add floating hair to the negative prompts if not desired.Depending on what you want to generate, you may also want to experiment with adding cleavage to the negative prompt or specifying a breast size as they vary quite a bit.Known Limitations / Problems:<ul><li>The shoes were not present in a lot of images and not a focus for the training so they sadly will not be consistent at all.</li><li>The gems on the coat also sadly seem to be quite inconsistent despite being in a lot of images.</li><li>This model seems to be much more fixated on the simple backgrounds used during training than others I've done (despite some variation in the training set). You may want to increase the weight of background tags and/or add simple background or similar to your negative prompts.</li><li>It is also sometimes a bit fixated on the 2-3 images in the training set where a covered navel was visible under the shirt so try covered navel in your negative prompts.</li><li>Despite a lot of attempts, the separation between what is part of the coat vs. what is part of the shirt below also does not seem to have stuck properly yet.</li><li>The coat + shirt seem to get combined to a (suit) jacket fairly often so try jacket or suit in your negative prompts if that happens.</li><li>From behind, the pants tend to come out as jeans, so try jeans in your negative prompts there.</li><li>From behind if not prompting for the coat, the shirt sleeves tend to have markings where the fur trims would usually be. I have no idea how to fix this yet. Maybe try lapels in the negative prompt?</li><li>The hair can be wildly inconsistent, especially from behind (but that's also the case for every illustration of Cynthia I know of).</li><li>The hair color might sometimes bleed into the sleeve color. I'd just regenerate in that case.</li></ul>

[SD 1.5] Pokemon - Cynthia / Shirona

There are detailed changes below the next paragraph, you might need to expand this version changes box!I learned a few new things about creating LoRAs and wanted to see how far I can push my existing LoRAs without changes to their datasets (well... not quite for this one, see below). So there's still no support for alternate outfits, colors etc. for now, sorry! There are other LoRAs if you want those things. Not sure if I'll improve the LoRAs more than that in the future or make new ones but who knows.LoRA Changes:<ul><li>Now uses the standard "cynthia \(pokemon\)" instead of a custom trigger word to make things easier for everyone (please note though that the LoRA itself is still named "character_pokemon_cynthia_vX" as before!)</li><li>Switch to 32 dims (now trained directly instead of resizing from 128)</li><li>Switch to target weight 1 (increased from 0.7)</li><li>Changed a lot of training settings (too many to list, check LoRA metadata if you are interested) and updated to much newer scripts version (~3 months worth of changes)</li><li>I had to remove four images from the dataset as they confused the new training too much. As a result, some details for full-body shots as well as side shots might sadly become a bit worse in general and especially for close-ups. Hopefully not much else gets worse.</li></ul>Example image changes:<ul><li>Use CLIP skip 2 instead of 1 as I accidentally did before</li><li>Change settings for hir-res fix: Switch upscaler from None to Lanczos and increase denoise from 0.1 to 0.4 to get better results while still not changing TOO much</li><li>Use a negative embedding for the majority of negative prompting</li></ul>Recommended weight: 1Training model: <a target="_blank" rel="ugc" href="https://civitai.com/models/66/anything-v3">Anything V3</a>Example image generation model: <a target="_blank" rel="ugc" href="https://civitai.com/models/4451/abyssorangemix2-hardcore">AbyssOrangeMix2 - Hardcore</a>

Files with SHA256: dcdefb8da2dcd2e8dcd14332648574204b29b49358482a49c4a8ef116b61dc13