Stable Diffusion Prompts: Positive + Negative Template
In Stable Diffusion the order matters as much as the negative prompt; this template sets subject, style, light, camera and quality in sequence.
You are writing Stable Diffusion prompts. Weights and order both matter. IMAGE [IDEA] MODEL: [SDXL / SD 1.5 / FLUX] STYLE: [STYLE] POSITIVE PROMPT ORDER 1) Subject and action (most important, first). 2) Setting and background. 3) Style and medium. 4) Light. 5) Camera and lens. 6) Quality and detail terms (few, not many). NEGATIVE PROMPT — start from this baseline and trim to the image lowres, blurry, jpeg artifacts, watermark, text, signature, extra fingers, extra limbs, deformed hands, bad anatomy, duplicate, cropped, out of frame, worst quality, low quality, normal quality RULES - Use weights sparingly: (term:1.2) for emphasis, (term:0.8) to soften. Never stack weights on one term. - Do not add more than 8 quality terms; they crowd out the subject. - Keep the positive prompt under 75 tokens for SD 1.5; SDXL and Flux tolerate more. - One subject per prompt. OUTPUT A) POSITIVE — ordered, comma-separated, with weights where needed. B) NEGATIVE — trimmed list for this image type. C) SETTINGS — steps, CFG, sampler, resolution suggestion with a one-line reason. D) 2 VARIATIONS — same subject, different style and different light. E) TROUBLESHOOTING — if hands deform, if the style overpowers the subject, if the image is too busy.
In Stable Diffusion, position in the prompt is a weight. Put what matters first.
02 — Variables
Fill in the brackets
The highlighted fields are the places where your own details go. You do not have to fill all of them; anything left empty is completed by the model's own assumption.
[IDEA]
[SDXL / SD 1.5 / FLUX]
[STYLE]
03 — How to use
How to use it, step by step
- 01
Token count matters: SD 1.5 loses the end of long prompts, so put the subject first.
- 02
A negative prompt is not a wish list — long negatives remove things you wanted.
- 03
Change one variable at a time when tuning; otherwise you cannot tell what worked.
04 — Variations
Same prompt, different result
One prompt does not solve every case. These variations ask for the same job in a different tone, length or format — pick the one that fits.
Photoreal
Same structure, photoreal: camera, lens, film stock, natural light, minimal quality terms.
Illustration
Same structure, illustration: line weight, palette, shading style, no photographic terms.
05 — Tips
Notes that take the result up a level
- Weights above 1.4 usually break the image. Stay between 0.8 and 1.3.
- If the style eats the subject, move the style terms after the light and camera, or lower their weight.
06 — FAQ
Frequently asked about Midjourney Prompts
Why is the negative prompt so long?
It is a baseline of common artefacts. Keep the parts that apply to your image and delete the rest — an over-long negative removes detail you wanted.
Do quality terms help?
Marginally, and they cost tokens. Three or four is enough; ten makes every image look the same.
Next
You may also find these useful
Midjourney Prompt Writing Guide: Order, Structure, Parameters
The only way to get controllable results in Midjourney is the right order: subject, setting, light, style, technical detail, parameters.
Intermediate MidjourneyMidjourney Portrait Prompts: 12 Ready Templates
Twelve ready templates that produce portraits with natural skin, correct light direction and no plastic finish.
Intermediate MidjourneyMidjourney Logo Prompts: Brand Mark Generation
Produces many logo directions, then narrows them to simple marks that can be redrawn as vectors.
Intermediate MidjourneyMidjourney Product Photography Prompts (E-commerce)
Product images at studio quality for e-commerce listings: surface, light, shadow and framing defined so the product stays the hero.
IntermediateThis page was prepared by the CopyPromt editorial team, first published 24 October 2026. Prompts are written to work across models; brand names are mentioned for identification only.