---V1.1 update---
Fast update, tested with multiple noise, until find a better result's.
It add details and variance.
The prompt is simple:
(image in frist line)
woman Día de los Muertos face paint
(image in second line)
woman Día de los Muertos face paint. In bottom left conrner a text in drip graffiti lectering showing "Kkw"
Thi is copy/paste workflow.
Replace sampler in your workflow with this, just open workflow, copy all nodes, and paste in your workflow, connect the needed nodes.
I saw many way to get variance with ZIT, this is my way, with no custom nodes.
The image show how it work with simple prompt and detailed prompt.
---Simple prompt----
tiger fighting against cobra snake, in a jungle
-----Advanced prompt----
[Scene] (A dense, humid jungle clearing, dappled with sunlight filtering through broad green leaves. The air hangs heavy and still, hinting at an impending storm.)
[Subject] (A magnificent Bengal tiger, male, fully grown, with thick, muscular limbs and a glossy, orange and black striped coat. His eyes are a piercing, predatory gold, fixed intently on an unseen target. His whiskers are long and sensitive, twitching slightly. Sweat glistens on his brow.)
[Action] (The tiger is stalking forward with deliberate, silent steps, muscles coiled and tense. He is moving with a focused intensity, extending his right paw forward as if preparing to strike a cobra snake)
[Location] (The clearing is surrounded by towering, ancient jungle trees, their roots thick and gnarled. Thick vines hang down from the branches, and a small stream meanders through the undergrowth. A decaying log lies partially submerged in the stream.)
[Lighting] (Shattered sunlight, a cool, green-tinged hue dominates the scene. The intensity is high, creating strong highlights and deep shadows, enhancing the tiger's menacing presence. The light source is slightly above and to the right, giving the tiger a three-dimensional quality.)
[Mood] (Predatory, powerful, dangerous.)
[Atmosphere] (Tense, expectant, primal.)
[View Angle] (Low angle shot, as if the viewer is on the ground, observing the tiger from its perspective. The focus is sharp on the tiger’s eyes and face.)
[Composition] (Dynamic, creating a sense of movement and threat. The diagonal lines of the tiger's body draw the viewer's eye towards the unseen target.)
---------------------
As you can see, with advanced prompt variance is minimal, seem affecting only non detailed descripted snake.
But with simple prompt, or old prompt style like sd1.5, seem working good.
I also have tested with settings included in workflow, may using other sampler or scheduler work better.
In the zip file is included json and png workflow, and the 2 sample images.