She-Hulk (Krea 2)

She-Hulk

This LoRA recreates my interpretation of She-Hulk as two linked versions of the same woman: Jennifer Walters as a petite, sharp, natural-skinned lawyer, and her transformed green form as a towering, muscular, green-skinned powerhouse. Same face language and same attitude, very different body.

Use the trigger word: shehulmar

The two physical states must be called separately. Use shehulmarhuman for her petite natural-skinned form, and shehulmargreen for her large green form. Every prompt with her in it should use shehulmar plus exactly one form token. That split was the hard part of this dataset. I wanted one LoRA that kept them recognizable as the same person, while forcing the scale, muscle, skin, and proportions to stay apart. It took a lot of consistent wording to get there, but I think it works well.

The LoRA is designed to keep her recognizable across portraits, full-body shots, courtroom and office scenes, candid street moments, action sequences, destruction scenes, intimate moments, and a wide range of visual styles. It handles clean professional portraits, sitcom-style workplaces, rainy nights, city streets, gym sessions, talk-show sets, comic-book splash pages, and more explicit scenes with equal ease.

Her default appearance depends on the form you call. shehulmarhuman is petite, with natural skin and long wavy brown hair. shehulmargreen is much larger, with green skin, visible muscle definition when she is flexing or fighting, and usually long wavy black hair. Say those body facts in the prompt. Do not lean on the tokens alone. Size against other people matters a lot for the green form, so describe her towering over coworkers, pedestrians, or anyone standing next to her when that is the point.

The model strongly recognizes her signature looks:

• classic-outfit: sleeveless white high-cut leotard with purple side panels, a chest window, and often white athletic shoes with purple stripes

• lawyer-outfit: navy pinstripe tailored jacket, white silk shirt, short matching navy skirt, and black stilettos

• torn versions of either outfit, using torn as the damage word (torn jacket, torn silk shirt, torn leotard, torn chest window)

• nude when she is fully unclothed

Other clothes work if you describe them. Workout gear, gowns, coats, casual knits, and so on are not locked tokens. Write the garments out.

Basic example
Cinematic full-body shot of shehulmar, shehulmargreen, green skin and defined muscular legs, standing in a modern office wearing a lawyer-outfit consisting of a navy pinstripe tailored jacket, white silk shirt, short matching navy skirt, and black stilettos, looking at the viewer with a confident expression, left hand on her hip, towering over the coworkers around her, bright daylight through floor-to-ceiling windows, city skyline behind the glass

She works well in many different settings and styles (photorealistic, cinematic, sitcom, comic-book, graphic novel, surveillance still, etc.) while also allowing casual clothing, formal dresses, lingerie, or full nudity when desired. The LoRA responds especially well to clear descriptions of pose, body language, expression, eye gaze, camera angle, lighting, and who else is in the frame and where they stand relative to her.

For transformation and power scenes, describe the effect as visible: glowing green eyes, glowing green cracks across the face or skin, torn lawyer-outfit or torn classic-outfit splitting off the larger body, debris, bent metal, or a bus, beam, or slab of concrete actually leaving the ground.

The model is most reliable when prompts use the right tokens and then describe the rest in plain visual language. Call shehulmar, call the form, name the outfit with consisting of plus the actual garments, say green skin or petite natural skin, say where she is looking, and say how big she is next to anyone else. That combination is what this LoRA was trained to follow.

This is a fan creation made out of love for the character and is not affiliated with or endorsed by Marvel. I hope people have fun with it, please post your results on the page, I would love to see what you create! Cheers :)