Generates images directly from text prompts using diffusion models. This foundational technique is essential for testing LoRA outputs, fine-tuning character anatomy, and exploring style or pose variations purely through descriptive input.

Transforms an existing image into a refined or altered version based on a new prompt. Excellent for enhancing LoRA results by maintaining structure while improving details like fingers, facial alignment, or artistic consistency.

Allows selective editing of specific regions in an image (e.g., redrawing hands or fixing eyes) while preserving the rest. Extremely powerful for correcting flawed anatomy in LoRA outputs without starting from scratch.

Refers to a structured pipeline combining multiple steps—txt2img, ControlNet, LoRA merging, upscaling—into a single process. Critical for advanced LoRA generation, ensuring consistency, precision, and high-quality anatomy in every output.

Transforms an existing video into a new animated sequence guided by prompts or reference styles. Useful for applying LoRA styles to motion footage, preserving body proportions and consistency over time.

Creates videos directly from text prompts. Combines the creativity of txt2img with temporal coherence. Ideal for exploring how LoRA-trained characters perform across animated timelines, including hand gestures and body movement.

Converts static images into short animated clips, often using motion interpolation or AI animation tools. Excellent for testing if LoRA-generated character designs retain anatomical clarity in motion.

Provides fine-grained control over composition and anatomy using reference maps (e.g., pose, depth, line art). A game-changer for LoRA-based generation, especially when aiming for precise hands, limbs, or symmetrical proportions.