Downloads · 30 days
155
20% of all-time downloads
pimp-my-pony-gguf
huchukato/pimp-my-pony-gguf
pimp-my-pony-gguf is a text-to-image model from huchukato. Use it when you need an image from a text prompt. The card lists the license as apache-2.0.
GGUF-quantized versions of huchukato/pimp-my-pony — an SDXL/Pony-based merge — for use with ComfyUI-GGUF.
All-time downloads
758
Public
Repo size
33.4 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.gguf33.4 GB · 100%
At a glance
- Task
- Text-to-Image
- License
- apache-2.0
- Access
- Public
- Created
- Feb 21, 2026
- Updated
- Sep 24, 2026
- SHA
- 4c25c655
Base models
Model card
From the Hugging Face model README
Pimp My Pony — GGUF
GGUF-quantized versions of huchukato/pimp-my-pony — an SDXL/Pony-based merge — for use with ComfyUI-GGUF.
Each model variant lives in its own folder, with multiple quantization levels.
Contents
| Variant | Quants | Notes |
|---|---|---|
SemiRealAnime/ (pimpmypony-SRA) | Q8_0, Q6_K, Q5_K_M, Q4_K_M | semi-realistic anime style |
More variants will be added over time.
Usage (ComfyUI)
These files contain only the UNet — CLIP and VAE are not included. You need three separate loaders:
| Component | Node | What to load |
|---|---|---|
| UNet | Unet Loader (GGUF) | the .gguf file from this repo |
| CLIP | DualCLIPLoader | SDXL text encoders (clip_l + clip_g), e.g. from the base repo or any SDXL CLIP pack |
| VAE | Load VAE | any SDXL VAE (a special VAE is included in the base model — recommended) |
Requires the ComfyUI-GGUF custom node
(pip install --upgrade gguf).
Which quant should I pick?
SDXL is conv2d-heavy, so gains are smaller than on transformer models (Flux/Wan) — but still meaningful on limited VRAM / unified memory.
| Quant | Quality | Use case |
|---|---|---|
Q8_0 | ~lossless | plenty of VRAM, max quality |
Q6_K | very good | balanced |
Q5_K_M | good | balanced, slightly smaller |
Q4_K_M | decent, some loss | tight VRAM (≤8GB) |
Recommended generation settings are the same as the base model: clip skip 2, included VAE, avoid excessive negative prompts.
Conversion
Converted with the official
ComfyUI-GGUF tools:
UNet extracted to safetensors → convert.py (BF16 GGUF) → llama-quantize
(llama.cpp b3962 + lcpp.patch).
Limitations
- Conv2d layers cannot be quantized; low quants may degrade fine details
- Quantized CLIP is not provided here — text encoders stay FP16/BF16
- GGUF files are not usable by llama.cpp itself (image-model format for ComfyUI-GGUF only)
License
Apache-2.0, same as the base model. Free for personal use and personal merging.
- Task
- Text-to-Image
- License
- apache-2.0
- Created
- Feb 21, 2026
- Updated
- Sep 24, 2026