model-compatibility

1# ComfyUI Model Compatibility Matrix
2 
3## Stable Diffusion 1.5 (SD 1.5)
4 
5### Overview
6 
7The original widely-adopted Stable Diffusion model. Huge ecosystem of fine-tunes, LoRAs, ControlNets, and embeddings. Still the most compatible and lightweight model family.
8 
9### Configuration
10 
11| Parameter | Value |
12|-----------|-------|
13| **Loader** | `CheckpointLoaderSimple` |
14| **Native Resolution** | 512x512 |
15| **Supported Resolutions** | 512x512, 512x768, 768x512, 768x768 (some fine-tunes) |
16| **VAE** | Built-in or external (`vae-ft-mse-840000-ema-pruned.safetensors`) |
17| **CLIP** | Single CLIP-L (output index 1 from checkpoint) |
18| **Text Encoder Node** | `CLIPTextEncode` |
19| **CFG Range** | 7-12 (typical: 7.5) |
20| **Negative Prompt** | Yes — very important for quality |
21| **Steps** | 20-30 (standard samplers) |
22| **Sampler** | All standard samplers: `euler`, `euler_ancestral`, `dpmpp_2m`, `dpmpp_sde`, `ddim` |
23| **Scheduler** | `normal`, `karras` |
24| **Denoise** | 1.0 (txt2img), 0.5-0.8 (img2img) |
25| **VRAM (FP16)** | ~2-3GB |
26 
27### Workflow Pattern
28 
29```
30CheckpointLoaderSimple → MODEL(0), CLIP(1), VAE(2)
31  CLIP(1) → CLIPTextEncode (positive) → CONDITIONING
32  CLIP(1) → CLIPTextEncode (negative) → CONDITIONING
33EmptyLatentImage (width=512, height=512) → LATENT
34KSampler (cfg=7.5, steps=20, sampler="euler", scheduler="normal") → LATENT
35VAEDecode → IMAGE
36SaveImage
37```
38 
39### VAE Notes
40 
41- Most SD 1.5 checkpoints have a built-in VAE, but it's often mediocre
42…