Base model: REAL PHOTOGRAPHY (realism/photorealistic) by QUARTERJACK

JJJ · Chimera E7 is a cross-medium visual LoRA built around style elasticity, strong material rendering, adult character variety, and high prompt responsiveness.

It is not intended to impose one fixed house style. Chimera is designed to move with the prompt instead.

It can shift between photorealistic portraiture, cinematic realism, semi-realistic illustration, painterly rendering, watercolor, ink wash, gouache-like work, graphic illustration, impasto, fashion-sketch aesthetics, and other mixed-media directions while retaining a recognizable underlying visual language.

The LoRA tends to actively interpret age, condition, styling, lighting, mood, and surface cues rather than simply preserving the base model output. This can produce strong character presence, but vague prompts may also be pushed toward CHIMERA's own cinematic/editorial preferences.

Strengths

• Strong handling of adult women without collapsing everything into youthified white-young-thin defaults

• Good age retention across mature, tired, older, weathered, and non-idealized faces

• Strong prompt responsiveness across realistic, cinematic, fantasy, athletic, formal, streetwear, and stylized character prompts

• High style elasticity across watercolor, ink wash, painterly blocks, monochrome illustration, graphic rendering, impasto, fashion sketch, and other media

• Good character presence and mood readability

• Strong material separation across skin, fabric, leather, metal, glass, hair, fur, smoke, paper, paint, and other surfaces

• Handles accessories, layered clothing, structured outfits, props, and environmental objects well

• Good multi-subject coherence

• Surprisingly stable non-human and anthropomorphic character handling

• Can move across different rendering media without forcing every prompt back into one fixed visual style

Weaknesses

• Persistent glossy-skin / high-specular tendency, especially on the REAL PHOTOGRAPHY base

• Skin, hands, fabric, and other materials can occasionally become overly polished or lacquered

• Tendency toward editorial polish and glamourization

• Some prompts may be beautified or sexualized beyond what is strictly requested

• Attractive or formal subjects may become more camera-aware, polished, or fashion-editorial than intended

• Extreme style shifts usually preserve mood and overall visual identity better than strict facial identity fidelity

• Very natural, dry, understated realism may require more explicit prompting

Best Use Cases

• Mature women

• Androgynous / butch / elegant adult characters

• Moody portraits

• Character introduction images

• Cinematic realism

• Formalwear and tuxedo portraits

• Fantasy or historical-inspired character work

• Athletic or physically worn characters

• Painterly character studies

• Fashion/editorial leaning concept work

• Stylized realism

• Cross-medium experimentation

• Bot PFPs and character-card artwork where variation is preferred over one fixed house style

Prompting Tips

If you want more natural realism and less shine, explicitly request:

matte skin, subdued highlights, dry skin texture, natural surface response, restrained specular highlights, non-editorial lighting

If you want a more understated character rather than a glamour portrait, try:

not glamorous, not seductive, understated, practical styling, non-fashion pose, candid presence, ordinary body language

If using older subjects, avoid overstacking age-related wording. Chimera already reads mature age well, so combining too many terms such as:

older, elderly, weathered, deeply lined, tired, aged

may push the result much older than intended.

For painterly or traditional-media results, medium-specific language generally works better than photography terminology. Describe the actual rendering behavior you want, for example:

transparent watercolor washes, cold-pressed paper, pigment bleeding

or:

heavy impasto, palette-knife marks, thick paint buildup

rather than relying only on broad words such as painting or artistic.

Chimera responds well to explicit descriptions of:

• age

• facial structure

• fatigue or physical condition

• body type

• clothing structure

• material texture

• lighting

• composition

• rendering medium

• overall scene atmosphere

Recommended LoRA Weight

0.55–1.0

• 0.55–0.7 — lighter influence, more freedom from the base model

• 0.75–0.9 — balanced range

• 1.0 — strongest CHIMERA character, material, and style influence

These are only starting points. Feel free to experiment. :3

Cross-Base Use

Chimera was trained on REAL PHOTOGRAPHY base model, so neutral prompts naturally retain some photographic influence.

It can also work on other KREA2-derived bases, but the base model still matters. Surface rendering, realism, gloss level, and overall style may change noticeably depending on the checkpoint underneath it.

In other words: the LoRA transfers well, but the base still has a personality.

Reference Image Settings

All reference images were generated under:

• LoRA weight: 1.0

• Sampler: Euler

• Sampling steps: 8

• CFG scale: 1

• Resolution: 768×1152

• Trigger word: not used

No trigger word is required for normal use.