Verified: a year ago
SafeTensor
Type | |
Stats | 125 0 |
Reviews | (18) |
Published | Mar 17, 2024 |
Base Model | |
Hash | AutoV2 D2BDAA5F39 |
Introduction
VisionMix will be my second release of a checkpoint model. My last one was called EpicCeleb. The goal of this model is to take creative arts and photorealism and combine them into one.
Suggested Settings
I heavily encourage everyone who is reading this to use my suggested settings, as it will most likely boost performance.
Sampling Method: For quality, use DPM++ SDE Karras. For speed, use DPM++ 2M Karras.
Sampling Steps: 25-30
Size: 512x768 or 768x512
CFG Scale: 5 for realism, or 7.
I highly suggest that you make a TensorRT model for my model. IT WILL BOOST SPEED.