Mary Dahl Mature - Baby Doll (Batman the Animated Series - BTAS) by TheLightningCrab

Created by TheLightningCrab

https://civitai.com/models/870643/mary-dahl-mature-baby-doll-batman-the-animated-series-btas

Mary Louise Dahl (Baby Doll) from her 2nd appearance in Batman the Animated Series (BTAS). A fully mature woman stuck forever in a young body. This model is only trained using images of her mature form, not herself dressed in her younger clothes. I might make a second LoRA with that as an option sometime later. But you can easily ask the LoRA to give her twintails.

Remix Tip: Remove the term Baby from your prompts.

This LoRA is made with 1080p video sources from 72 separate images. I manually edited approximately 73% of all the images, ensuring their quality, and also avoiding over-saturating with the same limited outfits. (This took MUCH more time than expected.)

(I'm still new to LoRAs, so let me know if something is wrong.)

Trigger words:

Mary_Dahl_Mature, 1girl

Use the word 'chibi' or some sub-level (chibi:0.#) value to achieve desired petite-ness level. For example: (chibi:0.5), or (chibi). You can see this in my example posted images:

chibi

(chibi:0.5)

single hair bun, hair bun

twintails

Image Resolutions Make your resolution images at (or larger than) 720x720, or use else results will be slightly blurry. For example, suggested image resolution is 832x1216. HiRes Fix not required but can improve image quality, usually with a 0.3 to 0.4 denoise.

(optional) SDXL Recommended Resolutions/setting:

640 x 1536(5:12)

768 x 1344 (4:7)

832 x 1216 (13:19)

896 x 1152 (7:9)

1024 x 1024 (1:1)

1152 x 896 (9:7)

1216 x 832 (19:13)

1344 x 768 (7:4)

1536 x 640 (12:5)

Checkpoints Best results tested so far used checkpoints like:

Mistoon Copper

Toonify

Autism SDXL

Prefect Pony

NOT RECOMMENDED: The current default Pony V6 checkpoint seems the LEAST attuned to produce the best semi-cartoonish results. (Which is funny, because that is what I used Pony V6 to train this LoRA.)

Tested with other LoRAs I tested this LoRA model with several other LoRAs, including the following (many all at once):

ExpressiveH (Hentai LoRa Style) エロアニメ: This works, but try lower values since it might distort facial expressions.

Detail Tweaker XL

nagatiti_Lora: This works but she's so small, the results are limited.

Bad artist Negative

CyberRealistic Negative

Negative Hand

NeoLight

d3f0rm3d