modèles/DMD2 Speed LoRA [SDXL, Pony, Illustrious] - dmd2_sdxl_4step_lora

DMD2 Speed LoRA [SDXL, Pony, Illustrious] - dmd2_sdxl_4step_lora

7/27/2025

11:19:04 AM

Mots-clés et tags associés

6tz,distribution matching distillation,dmd2,dmd2_sdxl_4step_lora,fast image synthesis,hyper,lcm,lora,sdxl 1.0,speed,stable diffusion,outil,turbo

A cyberpunk female cyborg crouching on rubble with glowing eyes, illuminated by bright lightning in a rainy night cityscape.

A happy orange tabby cat with raised arms and closed eyes, smiling with blush on cheeks, standing under a floating orange camera and yellow stars on a white background.

Silhouette of aku bouzu with dripping outline holding a rose in a thorny bush, set against a vibrant key lime to deep crimson gradient backdrop with dripping watercolor effects.

Flat vector illustration featuring a girl with blonde hair sitting barefoot on a large concrete docking station by the ocean, with a science fiction spacecraft flying in the blue sky.

Girl lying on bed in a blue onesie reaching for a smartphone, lit by sunset through window with potted plants on sill in a dimly lit room

A white-haired girl wearing a long black sleeveless dress stands against a wall on a neon-lit sci-fi street, eating a donut with a long sword behind her back.

Watercolor illustration of a blue cat monster with one long leg and one short leg, cat ears, standing on one hand, interacting with two smaller cats in grass with flowers.

A glowing india-green jackalope character with long ears and antlers in a dynamic pose, standing in water surrounded by blue and coral-colored plant life under a dark night sky.

A striking android woman with white hair and robotic limbs wearing sleek mecha armor and a flowing yellow cape stands against a bright blue background, accompanied by a large mechanical figure.

Vibrant graffiti mural with a large, detailed eye surrounded by swirling pink shapes on a blue background in a surreal urban street art style.

A vibrant graffiti-style image of a female cyclops with a long patterned neck and a single large eye, multicolored hair flowing with vivid colors on a cityscape background.

Woman with short striped hair and ivory skin in an orange jumpsuit and embroidered heels posed dynamically near a neon-lit water tower at sunset.

Prompts recommandés

A closeup of a SFW

Prompts négatifs recommandés

NSFW, naked, nude, watermark, text, simple background

Paramètres recommandés

samplers

LCM, lcm

steps

cfg

1.4 - 1.8

other models

Checkpoints\Z\ILXL\WIP\Perfection_ILXL_Realistic_3.2.safetensors (83a0ec21ca), Checkpoints\Checkpoints\SDXL\Photorealistic\juggernautXL_v9Rundiffusionphoto2.safetensors (c9e3e68f89)

Conseils

Recommended LoRA weight: 0.7, but feel free to adjust.

Use the LoRA with any SDXL or similar checkpoint using the recommended settings.

You can merge this LoRA into a checkpoint to include its effects directly.

If results appear washed out or bland, try increasing the CFG scale.

Points forts de la version

dmd2_sdxl_4step_lora

dmd2_sdxl_4step_lora_fp32

Sponsors du créateur

This is the official DMD2 implementation LoRA from the Distribution Matching Distillation for Fast Image Synthesis paper.

Original model at Huggingface tianweiy/DMD2.

Full credit goes to Tianwei Yin.

This is the official DMD2 implementation LoRA from the Improved Distribution Matching Distillation for Fast Image Synthesis paper.

dmd2_sdxl_4step_lora

Original model at Huggingface tianweiy/DMD2.

Recommended Settings

Steps: 8

LoRA weight: 0.7

Sampler: LCM

Scheduler: Normal, Simple

CFG: 1.4 - 1.8

I use similar settings on both a low resolution base image, and the upscaled refinement pass (hires fix).

Usage

Load the LoRA like any other. I recommend a weight of 0.7, but feel free to play around.

You should be able to use it with any SDXL or similar checkpoint. Use the settings above.

You can also merge this into a checkpoint and save that, to include the result of the LoRA into the checkpoint directly. This means you no longer need to load this as a LoRA, and you can still use it with LCM and low steps for good result.

The LoRA itself doesn't add any speed improvement, it just allows you to generate at fewer steps, with better result.

If the results are washed out and bland, try increasing the CFG.

Licence

Creative Commons Attribution Non Commercial 4.0

I am not the creator of the model, just uploading the model here to make it easier to download and reference in creations. Full credit goes to Tianwei Yin.