PixelWave 10
A general-purpose fine-tuned model, great for art and photo styles.
Model Description
PixelWave 10 has been fine-tuned to provide excellent results for both artistic and photographic styles. The focus for version 10 was to train the CLIP models, which improves reliability, ensures a wide variety of styles, and enhances prompt adherence.
Usage Instructions
Recommended Settings
- Standard Use:
- 20 steps, DPM++ SDE, CFG 4
- 40 steps, 2M SDE Karras, CFG 4
Tips for Best Results
Photographic Styles: Avoid terms like 'vibrant', 'intense', 'bright', 'high contrast', 'neon', 'dramatic' for a more natural look.
Art Styles: Specify the art style, near the beginning. I usually put it after a short description of the image. Mention artist names to get their specific style, 'by artist Van Gogh'
PAG Recommended: Use 1.5 Scale with CFG 3.
Example Prompts:
Elegant Lady with Pearls, color digital photograph, woman, blonde hair, magenta sweater, pearl necklaces, earrings, thoughtful expression, indoor, plants, warm lighting, shallow depth of field, natural light, portrait, focused gaze, long hair, gold jewelry, greenery, selective focus
Mountain Scene at Sunrise by artist Albert Bierstadt, oil painting, mountain scene, sunrise, landscape, adventure, journey, rugged terrain, rocky peaks, river flowing, golden light, blue sky, atmospheric perspective, dramatic composition, textured brushwork, detailed foliage, ethereal mood, natural beauty, serene expression, dynamic movement, depth of field.
Features
- Improved CLIP Training: Enhanced reliability and versatility in styles.
- Wide Style Variety: Capable of producing a vast range of artistic and photographic styles.
- Prompt Efficiency: Better adherence to prompts without needing 'quality' terms.
No Need for the Refiner Model
PixelWave 10 achieves high-quality results without the need for a refiner model.
Additional Tools
I also created Mikey Nodes, which contains a lot of useful nodes. You can install it through comfy manager.
License
creativeml-openrail-m
- Downloads last month
- 38