Training materials are sourced from "Peerless Tangmen", and the training materials are relatively simple without trigger words.
1. Recommendations for Image Generation:
For large models, recommended: SouthBridge_SD1.5_V1.0 (Model on the homepage)
Lora Weight Recommendation 0.6~0.8 for best results.
2. Recommendations for Relevant Parameters:
VAE: vae-ft-mse-840000-ema-pruned.safetensors
Lora Weight Suggestion: 0.6~0.8
Skip Clip: 2
Recommendation for HD Repair Redrawing: 0.2~0.4
Recommendation for Large Models: SouthBridge_SD1.5_V1.0
Recommended Sampling Method: DPM++2M Karras
Sampling Steps: 20-30
Description
Welcome to use lora【Seeking for feedback images! Seeking for feedback images! Seeking for feedback images!】
Sampling methods (Sampler:):
Euler a: 20~30 steps.
DPM++ 2S a Karras: 25~30 steps.
DPM++ SDE Karras: 20~30 steps.
DPM++ 2M SDE Karras: 20~40 steps.
DDIM: 30 steps and above.
Lora weight 0.6-0.8
[Mirra]
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.