by Lykon
Versatile SD 1.5 fine-tune handling diverse styles from photorealism to anime and fantasy art. One of the most popular community checkpoints, runs on 4GB+ VRAM.
VRAM requirements, GPU fit, and setup notes for DreamShaper 8, including 8GB/12GB fit guidance where relevant. Recommended runtimes: ComfyUI, Automatic1111, and Diffusers support. Best download size: ~2.0 GB at FP16.
Your hardware
Detecting...
Measured quality metrics for DreamShaper 8 outputs.
How often humans prefer this model's output (0-100%)
Visual quality and composition rating (5-9 scale)
Compare which GPUs can run DreamShaper 8 at different precisions. FP8 uses less memory than FP16 when available, and the grade shows how comfortably each GPU handles the workload.
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 4.0 GB | S● | S● | S● | S● |
| 768×768 | 4.8 GB | S● | S● | S● | S● |
| 1024×1024 | 5.9 GB | S● | S● | S● | S● |
Turbo / LCM distillation
Use distilled scheduler at 4-8 steps for faster iteration
ControlNets available
Add guided generation with 3 adapters (+0.7 GB VRAM each)
Rich LoRA ecosystem
Customize style, characters, and quality with community LoRAs
from diffusers import StableDiffusionPipeline
import torch
pipe = StableDiffusionPipeline.from_pretrained(
"Lykon/dreamshaper-8",
torch_dtype=torch.float16
)
pipe.to("cuda")
image = pipe(
prompt="your prompt here",
num_inference_steps=25,
guidance_scale=7.5,
height=768,
width=768,
).images[0]
image.save("output.png")Get started
Setup instructions for running DreamShaper 8 locally
1. Download the model
Get the checkpoint from HuggingFace
2. Place in:
ComfyUI/models/checkpoints/3. Launch ComfyUI
python main.pyComfyUI Workflow
Basic txt2img workflow for DreamShaper 8
Drag & drop into ComfyUI or use File → Import
VRAM allocation at 1024×1024 on RTX 4090 24GB (24 GB)
Time per image at 1024×1024, 28 steps, FP16.
Download DreamShaper 8 in the precision that matches your GPU. Lower precision usually means less VRAM pressure, while higher precision keeps more quality.
| Format | Precision | Size | Provider | |
|---|---|---|---|---|
| safetensorsRecommended | FP16 | 2.0 GB | Lykon | Download |
3 ControlNets available for DreamShaper 8. ControlNets add guided image generation (edges, depth, pose) at the cost of extra VRAM.
Inherits SD 1.5 base model ControlNet compatibility. Edge-based structural guidance.
Inherits SD 1.5 base model ControlNet compatibility. Depth-based spatial control.
Inherits SD 1.5 base model ControlNet compatibility. Human pose control.
Inherits full SD 1.5 LoRA ecosystem — 50,000+ LoRAs on CivitAI.
Approximately 50,000 LoRAs available on CivitAI. Each LoRA adds ~0.1 GB VRAM.
Browse all LoRAs on CivitAIFrequently asked questions
DreamShaper 8 (0.86B parameters) requires approximately 5.9 GB of VRAM at FP16 precision for standard 1024×1024 image generation. If you want a lighter setup, lower precisions like FP8 can reduce memory use when available.
DreamShaper 8 usually needs more than 8GB for comfortable local use. Check the VRAM table above for the exact resolution and precision trade-off.
DreamShaper 8 is marked for ComfyUI, Automatic1111, and Diffusers support in our catalog, so those are the runtimes we recommend first for local setup. If your workflow uses another front end, check the model's available formats and workflow notes above before downloading.
Yes, the RTX 4090 (24 GB VRAM) can run DreamShaper 8 comfortably at FP16. Expected generation time is around ~1s per image at 1024×1024.
Yes, DreamShaper 8 has 3 ControlNet adapters available: Canny Edge (SD 1.5), Depth Map (SD 1.5), OpenPose (SD 1.5). Each ControlNet adds roughly 0.7 GB of extra VRAM.
Inherits full SD 1.5 LoRA ecosystem — 50,000+ LoRAs on CivitAI. The LoRA ecosystem for DreamShaper 8 is rated as "massive". There are approximately 50,000 LoRAs available on Civitai. Each LoRA adds roughly 0.1 GB of extra VRAM.
On a reference GPU (RTX 4090 24GB), DreamShaper 8 generates a 1024×1024 image in approximately ~1s at FP16 with 28 inference steps. Faster GPUs with higher memory bandwidth will produce images more quickly.
See also