Diffus
Get started
models/DMD2 Speed LoRA [SDXL, Pony, Illustrious] - dmd2_sdxl_4step_lora_fp16

DMD2 Speed LoRA [SDXL, Pony, Illustrious] - dmd2_sdxl_4step_lora_fp16

6tZ
|
9/17/2025
|
10:47:52 PM
| Discussion|
0
A golden cat with mysterious purple eyes walks gracefully amid a dreamlike, painterly scene with purple and blue watercolor lighting and floral accents.
Silhouette of a girl standing in a flower-filled meadow during sunset with a large sun, a mauve sky, and twisted shadow-like strands rising upwards.
Female figure with dark silhouette, highlighted by scarlet color theme and tenebrism lighting, sitting at a workshop table with instruments against a crimson background.

Recommended Negative Prompts

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

Recommended Parameters

samplers

LCM, lcm

steps

8

cfg

1.4 - 1.8

other models

Checkpoints\Checkpoints\SDXL\Unsorted\juggernautXL_juggXIByRundiffusion.safetensors (33e58e8668)

Tips

Try combining both the fp32 and fp16 LoRAs at 0.5 weight each for better performance.

Use the LoRA weight setting of around 0.7, though weights of 1.0 also show good results.

If images appear washed out or desaturated, increase the CFG value.

The LoRA enables generating with fewer steps without speed improvement on its own.

You can merge the LoRA into a checkpoint to use results directly without loading the LoRA.

Version Highlights

dmd2_sdxl_4step_lora_fp16

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.

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). The settings above are my personal recommendations, but I suggest you play around with the settings to find what works for you and the model you are using.

Combining Both LoRAs

It's also been reported to perform better to use both the fp32 and fp16 at the same time at 0.5 weight each.

Load the LoRA like any other. I personally recommend a weight of 0.7, but others swear it's better at 1.0 in weight.

Usage

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, desaturated or 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.

Contributor

6tZ

Previous
Photorealistic Skin ⛔️ No plastic [FLUX] - FLUX v0.1
Next
Image Enhancer Flux By Devildonia - v1.0 Extreme
Use this model

Model Details

Model type

LORA

Base model

SDXL 1.0

Model version

dmd2_sdxl_4step_lora_fp16

Model hash

b3d9173815

Creator

6tZ's avatar
6tZ

Discussion

Please log in to leave a comment.

Images by DMD2 Speed LoRA [SDXL, Pony, Illustrious] - dmd2_sdxl_4step_lora_fp16

A golden cat with mysterious purple eyes walks gracefully amid a dreamlike, painterly scene with purple and blue watercolor lighting and floral accents.
Silhouette of a girl standing in a flower-filled meadow during sunset with a large sun, a mauve sky, and twisted shadow-like strands rising upwards.
Female figure with dark silhouette, highlighted by scarlet color theme and tenebrism lighting, sitting at a workshop table with instruments against a crimson background.

lora Images

A white-haired girl with red eyes, wearing a black outfit and a golden spiked gauntlet, holding a red glowing sword, standing amid rolling fog in a sword tomb.
A fantasy male with demon wings kneeling in a blood-red pool under a glowing red moon, surrounded by swords and dramatic lighting.
Close-up of a dark fantasy female elf with white hair, black horns, pale skin, wearing intricate golden armor with skull motifs, holding a glowing red sword.
A realistic renaissance-style female elf with long black hair, green eyes, and pointed ears, wearing a long black dress with silver buttons, holding a gnarled staff in a mist-shrouded graveyard.
An AI generated image using stable diffusion showing a grand wizard in a blue robe with a mystical figure behind him.
Anime-style depiction of a female warrior with green hair, wearing a teal outfit with a brown belt, a purple cape, and orange gloves and boots. The image is generated using stable diffusion.
A vintage woman with red hair in a black dress holding two skulls and laughing joyfully. This is an AI generated image using Stable Diffusion.
Samurai standing in a dimly lit alley at night, generated by AI using Stable Diffusion.
A realistic 3D model of a man standing in a uniform created using Stable Diffusion.
Anime girl with long blonde hair wearing a cap, orange t-shirt, white pants, and red sneakers sitting in an urban environment, generated using Stable Diffusion AI.

speed Images

A young woman in a high-tech racing suit leans on a metal railing at the edge of a Shanghai Formula 1 track at night, surrounded by neon city lights and racing motion trails.
A night scene of a 1989 Mazda 787B racing car speeding on a wet Le Mans track, with light trails, rain and wet tarmac reflecting headlights.
Black and white luxury racing boat speeding on calm sea at sunset with pink tinted sky and motion blur effect, two men at the helm wearing helmets.
A crimson WRC rally car speeds through a snowy forest at night with heavy snowing and warm headlights cutting through darkness.
A vibrant illustration of Pikachu charging forward with an aggressive look, surrounded by colorful particles and motion blur effects.
Illustration of a vibrant red racing car with headlights on, speeding on a racetrack with a dynamic, abstract background and a visible driver in racing gear.
A heavily armored spearman with a flaming helmet and red spear kneeling amidst a foggy battlefield of ruined spears and scattered debris.
A green sports car, generated by AI using Stable Diffusion, parked on a dirt path in a nighttime forest setting.

tool Images

A furry humanoid character with a backpack and axe walking through a post-apocalyptic landscape. This is an AI generated image using Stable Diffusion.