SoftSketch is designed to deliver a refined mix of semi-realistic and stylized illustration. It captures the elegance of soft shading, clean lines, and expressive details, making it perfect for character portraits and artistic compositions.
For the best results, it is recommended to add the trigger word "flat color" to enhance the signature's smooth and balanced aesthetic.
If you'd like to use my checkpoint for online generation, please visit here:
👉 SeaArt AI | Tensor.Art | ShakkerAI(No longer updated)
Thanks for trying this model! 🤗 I hope it meets your creative needs and brings you joy. And don't forget to give a 👍 if you like it! ;>
Recommended Setting Examples:
Clip skip: 2
Recommended resolution: 1024x1024, 864x1152, 896x1152, 1152x864, 1152x896
Sampler: Euler a
Steps: 20~40
CFG: 4~7
Recommend using Hires Fix.
Upscaler: Remacri, R-ESRGAN_4x+, R-ESRGAN_4x+ Anime6B
Use Adetailer for face details
Feel free to share results!
Please DO NOT re-upload my model to other platforms.
Description
SoftSketch v3.0 Update:
Enhanced Contrast
Improved Lighting
Refined Skin Textures
Increased Detail Density
Vibrant Color Palette
FAQ
Comments (2)
When looking at your showcased pics and prompts for v3.0 compared to previous versions, it seems that v3.0 doesn't use the "flat color" tag and doesn't render pics in a "flat color" way anymore. Yet you still have using "flat color" in prompts as a recommendation. Is it an error?
Also how does your checkpoint handles LoRAs? Asking because when looking at your pics it seems to only be able to produce a single face type. Meaning that you ll need LoRAs to get really different faces.
v3 is bonkers beautiful
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.