Xenomorph LoRA for Krea2 - V1
Overview
A purpose-built LoRA that delivers screen-accurate xenomorph renders on Krea2. While the base model recognizes the xenomorph as a concept, it consistently drifts from the source design — inventing features, losing proportions, and defaulting to generic "scary alien" tropes. This LoRA corrects all of that, locking in the iconic H.R. Giger biomechanical aesthetic across any scene, angle, or lighting condition.
Trigger word: xenomorph
Why This LoRA Exists
The Krea2 base model knows what a xenomorph is, but it doesn't understand it. After extensive side-by-side testing (same seed, prompt, sampler, scheduler — everything identical), the base model consistently exhibited the following problems:
Proportions & Silhouette
The base model renders a stocky, muscular creature — more "video game monster" than the tall, skeletal, insectoid predator from the films. The LoRA corrects this with an elongated frame, lean digitigrade legs, spindly clawed limbs, and the signature towering bipedal silhouette that makes the xenomorph instantly recognizable.
Surface & Material
Base renders often produce a rough, mottled, brownish-black surface. The xenomorph should be glossy, wet obsidian black — a sleek chitinous exoskeleton. The LoRA maintains this signature finish across all lighting environments, from neon-lit rooftops to submerged reactor rooms.
Invented Features
This is the big one. Under complex prompts, the base model starts improvising: glowing red eyes (xenomorphs are famously eyeless), random spiky protrusions, fleshy tendrils, extra organic growths — none of which belong to the design. The LoRA eliminates these hallucinations and keeps the creature faithful to the source.
Anatomical Accuracy
The LoRA preserves the details that matter: the smooth elongated dome, properly arranged metallic double-row teeth, biomechanical tubing and piping along the skull and torso, visible dorsal tubes, correct ribcage segmentation, and the bladed tail tip. In close-ups especially, the difference is night and day.
Scene Complexity Resilience
When prompts demand complex environments — action scenes, elaborate hive setups, dynamic weather — the base model's attention drifts away from creature accuracy and toward the scene. The LoRA keeps the xenomorph locked in even when the surrounding scene is highly detailed, maintaining design fidelity regardless of prompt complexity.
Recommended Settings
Trigger word:
xenomorphLoRA weight: 0.7–1.0 (start at 0.8, adjust to taste)
Works best with: detailed environmental and cinematic prompts — the LoRA handles the creature, so focus your prompt tokens on the scene, lighting, atmosphere, and action
Tip: You don't need to over-describe the xenomorph's features in your prompt — they're baked into the LoRA. Spend your prompt budget on the scene instead.
Sample Prompts
Here are some prompts to get you started:
Action scene: xenomorph bursting through a metal floor grate beneath a squad of soldiers, sparks flying, muzzle flashes illuminating the scene, spent bullet casings mid-air, industrial cargo bay with stacked crates and chain-link fencing, overhead fluorescent lights flickering, smoke and debris, wide angle action shot, cinematic motion blur
Horror close-up: xenomorph inches away from a terrified astronaut pressed against a foggy cryo-pod window, the creature's inner jaw partially extended, saliva strands connecting jaw to glass, shallow depth of field, rows of cryo-pods with blue status lights in background, cold breath visible, extreme close-up
Atmospheric wide shot: xenomorph rising slowly from black floodwater in a submerged reactor room, only head and shoulders visible above surface, water cascading off the dome skull, submerged warning lights casting distorted green and red glow beneath waterline, floating debris and oil slicks, exposed pipes and catwalks above, photorealistic
Stylized / crossover environment: xenomorph crouched on a rain-slicked rooftop at night, torrential downpour with visible rain streaks, neon signs reflecting in puddles on cracked asphalt, distant city skyline with smog, lightning flash in background, tail coiled around a rusted antenna, cyberpunk noir mood
Hive scene: xenomorph guarding a massive resin-coated hive wall covered in eggs, unconscious humans cocooned to walls with translucent secretion, dim amber emergency lighting from a half-buried control panel, organic biomechanical architecture blending with corroded ship infrastructure, condensation dripping, ground-level camera angle
Version History
VersionNotesV1Initial release. Core xenomorph design accuracy across full-body, close-up, action, and environmental scenes.
This is an ongoing project. Future versions may include refinements based on community feedback, expanded variant support (Warrior, Drone, Queen, Praetorian), and improved performance at extreme angles or unusual art styles.
Disclaimer
This is a fan-made character LoRA and is not an official Alien, 20th Century Studios, or Walt Disney Company product. The Xenomorph and the Alien franchise are properties of their respective rights holders. This LoRA is intended as a creative character-generation model for AI image generation. No commercial affiliation or endorsement is implied.
If you found this LoRA useful, a rating or review helps others find it. Feedback and sample images are always welcome — they help improve future versions.
Description
FAQ
Comments (1)
Hello and thanks for your work. Can you tell me where i can find the used diffusion model called "Solstice-V2-int8-ConvRot.safetensors". I can't find it wheater in Civitai nor in huggingface. Thanks for your answer.