Please do share your creations if you make something that you like. We can all learn from each other.For best results:Start with a resolution of 1024x768 I recommend using the DPM++ SDE or DPM++ 2S a sampler with a CFG scale between 1 and 2.5, and 5-15 steps. Clip skip: 2. And then experiment.I've seen users make beautiful pics with other settings, so it may be worthwhile to experiment to see what suits your desire.VAE is included. Generally speaking, adding steps increases detail but if you want to use any sort of LoRA or hiresfix, a higher step count may fry your image. Lower cfg tends to follow the prompt better; higher cfg tends to give more contrast.2.2 - Hopefully more realistic, though colors are less punchy. Definitely itterative.1.9 SPO - better colors, and composition, now generates at 1288x12881.8 in addition to finer dark elements, now makes coherent pictures up to 1152x1152 without the aid of controlnet.1.6 improvements to contrast and composition.1.5 improves color and detail.1.4 merged in some of my favorite LoRAs.1.2 Is more creative and tends to handle hands better.Version 1.0 is functionally the same as 0.9t, the only notable difference being more life-like and saturated colors.0.9t represents a huge leap over the previous iteration. I've merged in some of the realness of Juggernaut and the responsiveness of OpenDalleV1.1 and crammed it back down to Turbo(merge) speed.

Ultraspice

<ul><li>Improved color and detail</li></ul>

ultraspice_v15.safetensors

v1.5

v2.3

v2.2

v2.1

v1.9

v1.8

v1.6

v1.4

v1.2

v1.0

v0.9t

v0.9t (sans VAE)

v0.4

Realistic XLTURBO_v15 large model

If you have created something you like, please share your creation. We can all learn from each other.For the best results:Start from 1024x1024 or 1024x768. I recommend using DPM++ SDE Karras sampler, with CFG ratio between 1 and 2.5, step size 5-11. I have seen users create beautiful images using other samplers, so it may be worth trying to see what suits your desires.Including VAE. In general, adding steps will increase detail, but if you want to use any kind of LoRA or hiresfix, higher step numbers may distort your image. Lower cfgs often better follow hints; higher cfgs yield higher contrast.ADetailer and Highresfix may require slightly different settings from non-Turbo models.If you find a piece you like but the fingers are not good, save the seed and adjust the CFG ratio to 0.1 or 0.01. They often generate successfully within a few attempts. 1.5 enhanced color and detail.1.4 combined some of my favorite LoRA.1.2 more creative and better at handling both hands.Version 1.0 is functionally the same as 0.9t, with the only significant difference being more realistic and saturated colors.0.9t represents a significant leap from previous iterations. I integrated some authenticity from Juggernaut and responsiveness from OpenDalleV1.1 and restored it to Turbo (merged) speeds.

写实XLTURBO_v15.safetensors

Realistic XLTURBO_v15

Files with SHA256: 069737f0a7484dc5aa86e2907d16f93bf31e9aef7a93a330a60e3ad7195814f9