---
title: 'Stable Diffusion Prompt Word Guide | Basic tutorial'
source: 'https://youtube.com/watch?v=NilOK7dExtc'
video_id: 'NilOK7dExtc'
date: 2026-07-28
duration_sec: 328
---

# Stable Diffusion Prompt Word Guide | Basic tutorial

> Source: [Stable Diffusion Prompt Word Guide | Basic tutorial](https://youtube.com/watch?v=NilOK7dExtc)

## Summary

This video explains the meaning of symbols in Stable Diffusion prompts and how to write effective prompts. It covers basic principles, weight adjustments, advanced syntax, and recommended prompt structure.

### Key Points

- **Basic Prompt Principles** [0:18] — Prompts are separated by commas; words at the front get higher weight; keep prompt words within 75 for best control.
- **Parentheses, Brackets, and Braces** [0:48] — Parentheses increase weight by factor 1.1 per layer; curly braces increase by 0.05; square brackets decrease weight by factor 0.9 per layer.
- **Direct Weight Adjustment** [1:36] — Use colon after word in parentheses to set exact weight (recommended 0.3-1.5). Example: (cherry blossoms:1.2).
- **Calling LoRA with Angle Brackets** [2:00] — Format: <lora:trigger_word:weight>. Tensor offers one-click LoRA trigger words.
- **Underscores as Links** [2:18] — Underscores connect words so Stable Diffusion treats them as a single concept (e.g., milk_cake).
- **Advanced Syntax: Timing Prompts** [2:35] — Use [prompt:number] to start prompt at that step; [prompt::number] to end at that step; [prompt1:prompt2:number] to switch prompts; vertical bar for alternate sampling.
- **Recommended Prompt Structure** [3:28] — Line 1: quality/style words; Line 2: main subject description; Line 3: environment/lighting; Line 4: LoRA trigger words.
- **Negative Prompts and Tools** [4:30] — Use general negative prompts; Flux model doesn't allow custom negative prompts. Tensor's workbench supports language input and translation.

### Conclusion

Mastering prompt syntax and structure helps generate more accurate images. Tensor provides free online models and daily credits for experimentation.

## Transcript

what do all those symbols in stable
diffusion really mean how can we write
prompt words so that the pictures we get
are more what we want when you see those
complex stable diffusion prompt words do
you feel confused too don't worry this
video will help you with all kinds of
problems when you fill in prompts first
off let's check out the basic principle
of prompt words prompt words are
separated by commas and you can put
prompt words on different lines but you
still need a comma at the end of each
line as for weight each prompt word has
a default weight of one but the words at
the front get a higher weight so put
important words at the front if you can
finally keep the number of prompt words
within 75 if there are too many words
they won't have much control over the
picture next up let's talk about what
each symbol means first off there are
parentheses square brackets and curly
braces these are mainly used to tweak
the weights of keywords when you put a
hint word in parentheses it's weight
becomes 1.1 you can use up to three
parentheses at the same time when n
parentheses are used the weight of the
hint is 1.1 to the power of n curly
braces are also for increasing the
weight of a hint curly braces are used
to increase weight too each layer of
curly braces increases the weight by
0.05 the adjustment with curly braces is
much smaller than that with parentheses
square brackets are used to decrease
weight one square bracket makes the
weight of a hint word 0 point 9 times
less three square brackets make the
weight 0.7 to 9 the easiest way to
adjust weights is to add a colon after
the hint word in parenthesis and then
fill in the desired weight value
directly after the colon It's
recommended to set this value between
0.3 and 1.5 for example if you enter
cherry blossoms kittens green leaves and
set a different weight value for cherry
blossoms you can control the proportion
of cherry blossoms in the whole picture
the pointed bracket are mainly used to
call Laura the format is Laura trigger
word weight value after calling Laura
images with specific features can be
generated it should be noted that in
tensor Laura tritter words can be
directly used with one click which is
very convenient underscores act as a
link to make the keywords more closely
connected for example a milk cake might
be understood by stable diffusion as a
glass of milk and a cake but if milk and
Cake are connected by Under scores then
stable diffusion will understand milk
and cake as a whole next let's check out
some Advanced syntax now how do we
control when a prompt kicks in first off
you can use a combo of square brackets
and colons there's a number after
cologne it means the prompt starts
rendering from that point a double
cologne that means the prompt in
parentheses renders till that point when
there's one prompt it's like this but
there can be two prompts in parentheses
like this the two prompts are separated
by a cologne so for the first 70% of the
rendering stone is on and for the next
30% flower is on if the two prompts in
parentheses are separated by a vertical
line like this it's alternate sampling
so red hair and blue hair prompts
alternately and you'll end up with red
and blue hair after you get the hang of
the basic syntax let's check out the
recommended way to describe the overall
prompt words picture quality and
painting style words have a big impact
on the overall look of the picture so
it's a good idea to write them on the
top I've got some tips on picture
quality prompts mostly they work for
everything but for different styles of
pictures there will be some specific
quality words too like pixel style Pixar
style ink painting style and so on add a
comma at the end of this and you can
start a new line if you want then
there's the description of the main part
of the picture my character age
hairstyle hair color what they're doing
and so on the more detailed you are the
more accurate the generated picture will
be add a comma after writing and start a
new line next up is the description of
the lighting in the environment like a
Snowy Evening or a Sunny Meadow add a
description of the lighting used at the
end of this the last line is for adding
the trigger word of the Laura you want
to use after positive prompts fill in
the negative prompts
generally just filling in some general
negative prompt is fine if you're using
the flux model it won't let you fill in
negative prompts on your own when you're
using tensor's classic workbench you can
just fill in the prompts in the language
you know after that click the
translation button right below when
you're out of ideas for the prompts you
can choose to generate them randomly or
by recognizing pictures oh and you can
also check out the posts recommended on
the homepage to get some
inspiration there are a bunch of tensor
models for you to pick from all of
tensor's models are online no need to
download they're one click and don't ask
too much of your graphics card best of
all they're free you can get free
credits every day by doing some stuff
click the website link under the video
to give it a try any questions or things
you want to hear about next time just
subscribe me in the comment section see
you