info.md

Usage

Enter a prompt and click Generate. Civitai has an excellent guide on prompting.

Compel

Positive and negative prompts are embedded by Compel, enabling weighting and blending. See syntax features.

Arrays

Arrays allow you to generate different images from a single prompt. For example, a cute [[cat,corgi,koala]] will expand into 3 prompts. Note that it only works for the positive prompt. You also have to increase Images to generate more than 1 image at a time. Inspired by Fooocus.

Autoincrement

If Autoincrement is checked, the seed will be incremented for each image. When using arrays, you might want to uncheck this so the same seed is used for each prompt variation.

Models

All use float16 (or bfloat16 if supported). Recommended settings are shown below:

• fluently/fluently-v4
  • scheduler: DPM++ 2M, guidance: 5-7, steps: 20-30
• linaqruf/anything-v3-1
  • scheduler: DPM++ 2M, guidance: 12, steps: 50, vae: default
• lykon/dreamshaper-8
  • scheduler: DEIS 2M
• prompthero/openjourney-v4
  • scheduler: PNDM
• runwayml/stable-diffusion-v1-5
  • scheduler: PNDM
• sg161222/realistic_vision_v5.1
  • scheduler: DPM++ 2M, guidance: 4-7

Schedulers

All are based on k_diffusion except DEIS and DPM++. Optionally, the Karras noise schedule can be used.

• DEIS 2M
• DPM++ 2M
• DPM2 a
• Euler a
• Heun
• LMS
• PNDM

VAE

All models use madebyollin/taesd for speed.

TODO

• Support LoRA
• Support embeddings
• Add VAE radio
• Add styles
• Clip skip
• DeepCache with T-GATE
• Hires fix
• Latent preview