DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI

Prompt book for

data lovers II ❤
An open source exploration on text-to-image and data
visualization

Version 1.0
Brought with love by Domestic Data Streamers 22nd of October of 2022
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 2

Preface
This book has been created as a free resource for
scientists, designers, journalists and storytellers, to
inspire them to look through the lenses of data language
when using text-to-image software. If you havenʼt yet
check out our materials Prompt Book here!

This document was created with lots of love by the AI

Division of Domestic Data Streamers, Veronika
Gryshchuk, Thinh Truong and Pau Aleikum.

For more resources follow our IG account and subscribe

to our newsletter.

All the images have been created using Dalle2, but with
similar results on Stable Diffusion. Thanks to Krea.ai
and Boldtron for being part of this explorative journey.
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 3

Table of contents

I. INTRODUCTION 4

II. SEMANTIC FIELDS 5

III. STANDARD SEMANTIC FAMILIES 6

IV. SUBJECTIVE SEMANTIC FAMILIES 21

V. BUILD YOUR OWN SEMANTIC FIELDS 39

VI. CONCLUSION 57
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 4

Introduction

This is just one way to build data visualization through

AI, through the following months hopefully we will
release more documents with more methods and ideas.

In this document we will specifically explore data

visualization through comparison. You will be able to
explore over 50 semantic fields, more than 400
quantitative words to describe these fields and some
examples of how they can be applied to simple
visualizations.
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 5

Semantic Fields Subjective families

A semantic field is a group of words with related Temporality Hair
meanings. The words in a semantic field are Value Mood
usually part of a category that helps define a Order Food
territory of knowledge. This is the collection of the Toughness Living stage
semantic fields that we have explored in order to Closeness Weather
generate degrees and comparisons. Turbulence Pollution
Resolution Wetness
Focus Moon phase
Image Sun position
Standard Families Body weight Technology
Skin tone Stage of development
Skin Sound frequency
Color Measurement Space Material

Brightness Quantity Area Erosion

Saturation Size Position Roughness
Contrast Height Distance Glossiness
Transparency Width Dispersion
Luminance Length Density
Volume Movement
Shape
Depth
Weight
Angle

Disclaimer: All the displayed images have been created using DALL-E.
If Stable Diffusion or Midjourney understands the quantitative scale of any family, it will show ✅ next to the family name. If they donʼt, it will show ❌
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 6

Standard semantic
families
When we talk about standard families, we refer to those
families that somehow define the basis of any
representational method, shape, color, position and so
on… These semantic families have been extracted from
the different variables of the semiology studies from the
famous and widely cited cartographer Jacques Bertin
and extended through the physical variables of
described on the researcher Yvonne Jansen on data
physicalization.

Disclaimer: We will be using as a common starting object a blue cube so we can

compare the different effects of the adjectives. That being said, in some of the
linguistic families we will need to change the blue cube for another base prompt to
be able to illustrate the visualization idea behind, as not everything can be
exemplified by a blue cube. (For example when talking about mood or facial
expressions)
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 7

Luminance ✅Midjourney
❌Stable Diffusion

Very gloomy Gloomy Slightly Slightly Glowing Very glowing Extremely

glowing
gloomy glowing

Prompt: very gloomy blue Prompt: gloomy blue cube Prompt: slightly gloomy Prompt: slightly glowing Prompt: glowing blue cube Prompt: very glowing blue Prompt: extremely glowing
cube in a black room blue cube in a black room blue cube cube blue cube

Quantity ❌Midjourney
❌Stable Diffusion

One Two Three Five Ten Fifty Hundred Thousand

Prompt: Top view of one blue Prompt: Top view of two Prompt: Top view of three Prompt: Top view of five Prompt: Top view of 10 blue Prompt: Top view of fifty Prompt: Top view of a Prompt: Top view of a
cube on white table blue cubes on white table blue cubes on white table blue cubes on white table cubes on white table blue cubes on white table hundred blue cubes on thousand blue cubes on
Note: ʻtenʼ didnʼt work but white table white table
ʻ10ʼ worked instead.
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 8

Quantity (cont.) ❌Midjourney

❌Stable Diffusion

One A few Any Some Many A lot A great

amount of

Prompt: There is no blue cube Prompt: There are a few Prompt: There are any blue Prompt: There are some Prompt: There are many Prompt: There are a lot of Prompt: There are a great
on white table blue cubes on white table cubes on white table blue cubes on white table blue cubes on white table blue cubes on white table amount of blue cubes on
Note: in this case of ʻnoʼ, it white table
always generates something

Quantity (cont.) ❌Midjourney

❌Stable Diffusion

Hundred One One One One An infinite

thousand billion trillion zillion googolplex number of

Prompt: Top view of a Prompt: Top view of a Prompt: Top view of a Prompt: Top view of a Prompt: Top view of a Prompt: Top view of an
hundred thousand blue cube billion blue cube particles on trillion blue cube particles zillion blue cube particles on googol blue cube particles on infinite number of blue cube
particles on white table white table on white table white table white table particles
Note: ʻon white tableʼ is
Note: in these large number cases, the object context changed to ʻparticlesʼ showed better result removed to not have the
ʻfiniteʼ context
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 9

Density ✅Midjourney
✅Stable Diffusion

Extremely Very Sparse Slightly Slightly Low dense Very Extremely

sparse sparse sparse dense dense dense

Prompt: aerial view of Prompt: aerial view of very Prompt: aerial view of Prompt: aerial view of a Prompt: aerial view of a Prompt: aerial view of low Prompt: aerial view of a Prompt: aerial view of an
extremely sparse group of sparse group of people on sparse group of people on slightly parse group of slightly dense group of dense group of people on very dense group of people extremely dense group of
people on the street the street the street people on the street people on the street the street on the street people on the street

Movement ❌Midjourney
❌Stable Diffusion

Extremely slow Ponderous Slow Still Speedy Fast Extremely fast Dangerously
fast

Prompt: a big blue cube with Prompt: a big blue cube with Prompt: a big blue cube with Prompt: a big blue cube with Prompt: a big blue cube with Prompt: a big blue cube with Prompt: a big blue cube with Prompt: a big blue cube with
wheels going extremely slow wheels being ponderous compare wheels going slow compare to a wheels standing still compare wheels going speedy compare wheels going fast compare to wheels going extremely fast wheels going dangerously fast
compare to a walking person to a walking person on the walking person on the pavement to a walking person on the to a walking person on the a walking person on the compare to a walking person compare to a walking person
on the pavement front view pavement front view front view, long shot pavement front view pavement front view pavement front view on the pavement front view on the pavement front view
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 10

Brightness (tint+tone) ❌Midjourney

❌Stable Diffusion

Extremely Very bright Bright Shady Dark Very dark Extremely

bright dark

Prompt: Extremely bright Prompt: Very bright blue Prompt: Bright blue cube Prompt: Shady blue cube Prompt: Dark blue cube Prompt: Very dark blue Prompt: Extremely dark blue
blue cube near human hand cube near the human hand near the human hand near the human hand near the human hand cube near the human hand cube near the human hand

Saturation ❌Midjourney
❌Stable Diffusion

Pure color … Colorful Shady Grey

Prompt: Pure blue color (rgb Prompt: Colorful blue cube Prompt: Cube of a shady Prompt: Grey cube on the
0,0,255) in a cube on the grass on the grass blue color on the grass grass
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 11

Contrast ❌Midjourney
❌Stable Diffusion

High Medium low

Prompt: High contrast of a Prompt: Medium contrast of a Prompt: Low contrast of a blue
blue color cube compering with blue color cube compering with color cube compering with a
a high contrast green grass a high contrast green grass high contrast green grass

Transparency ✅Midjourney
✅Stable Diffusion

Maximum Slightly Slightly Maximum Extremely

opaque opaque transparent transparent transparent

Prompt: Maximum opaque Prompt: Slightly opaque Prompt: Slightly Prompt: Maximum Prompt: Extremely
blue color cube on the green blue color cube on the green transparent blue color cube transparent blue color cube transparent blue color cube
grass grass on the green grass on the green grass on the green grass
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 12

Size ❌Midjourney
❌Stable Diffusion

Micro Tiny Small Average Big Huge Gigantic Humongous

Prompt: A micro blue cube Prompt: A tiny blue cube in Prompt: A small blue cube Prompt: An average blue Prompt: A big blue cube in Prompt: A huge blue cube Prompt: A gigantic blue Prompt: A hand touching a
in a hand a hand in a hand cube in a hand a hand on a hand cube compared to a hand humongous blue cube

Size (cont.) ❌Midjourney

❌Stable Diffusion

One-tenth of Half the Double the Triple the Ten times A hundred
size of size of size of bigger times bigger

Prompt: A blue cube with Prompt: A blue cube with Prompt: A blue cube with Prompt: A hand touching a Prompt: A hand touching a Prompt: A hand touching a
the size one-tenth of a hand half the size of a hand double the size of a hand blue cube triple the size of blue cube ten times bigger blue cube a hundred times
that hand than that hand bigger than that hand
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 13

Height ❌Midjourney
❌Stable Diffusion

Extremely Very short Short Slightly Slightly Tall Very tall Extremely
short short tall tall

Prompt: An extremely short Prompt: A very short height Prompt: A person standing Prompt: A slightly short Prompt: A blue cube Prompt: A blue cube taller Prompt: A blue cube much Prompt: A person standing
height blue cube next to blue cube next to a person next to a shorter blue cube height blue cube next to slightly taller than a person than a person much taller than a person next to an extremely tall
a person a person blue cube

Width ❌Midjourney
❌Stable Diffusion

Extremely Very thin Thin Slightly thick Thick Very thick Extremely
thin thick

Prompt: side view of an Prompt: side view of a very Prompt: side view of a thin Prompt: side view of a slightly Prompt: side view of a thick Prompt: side view of a very Prompt: side view of an
extremely thin blue cube, thin blue cube, which is 5 blue cube, which is 2 thick blue cube, 10 centimeters blue cube, 20 centimeters thick, thick blue cube, 40 centimeters extremely thick blue cube, 60
which is 0.2 millimeters thin, millimeters thin, as thin as the centimeters thin, as thin as a thick, as thick as an as thick as a stack of books, next thick, as thick as an old tv, next centimeters thick, as thick as a
as thin as a needle world's thinnest smartphone, book, next to a book encyclopedia, next to that to that stack of books to that old tv tree, next to that tree
next to that phone encyclopedia
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 14

Length ❌Midjourney
❌Stable Diffusion

Extremely Very short Short Slightly Slightly Long Very long Extremely
short short long long

Prompt: top view of an Prompt: top view of a very Prompt: top view of a short Prompt: top view of a Prompt: top view of a Prompt: top view of a long Prompt: top view of a very Prompt: top view of an
extremely short blue cube, short blue cube, next to blue cube, next to a ruler slightly short blue cube, next slightly long blue cube, 10 blue cube, 20 centimeters long blue cube, 30 extremely long blue cube, 60
next to a ruler a ruler to a ruler centimeters long, next to long, next to a ruler centimeters long, next to centimeters long, next to
a ruler a ruler a ruler

Volume (and Optimism) ✅Midjourney

❌Stable Diffusion

Empty Slightly Half-full Slightly Full Overfull

empty Full

Prompt: front view of an Prompt: front view of a Prompt: front view of a Prompt: front view of a Prompt: front view of a full Prompt: front view of a
empty glass cube, white slightly empty glass cube half-full glass cube slightly full glass cube glass cube containing blue overfull glass cube containing
background containing blue water, white containing blue water, white containing blue water, white water, white background blue water, white background
background background background
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 15

Depth ❌Midjourney
❌Stable Diffusion

Extremely Very deep Slightly deep Slightly Very shallow Extremely

deep shallow shallow

Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue
cube extremely deep in a very cube very deep in a very tall cube submerged slightly cube submerged slightly cube very very very shallowly cube submerged extremely
tall glass bottle glass bottle deeply in a very tall shallowly in a very tall in a very tall glass bottle shallowly in a very tall glass
glass bottle glass bottle bottle, extremely shallow

Depth (cont.) ❌Midjourney

❌Stable Diffusion

Elevated High Extremely Skyscraping

high

Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue
cube floating elevated over a cube floating high over a cube floating extremely high cube floating skyscraping
glass bottle glass bottle over a glass bottle over a glass bottle
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 16

Weight ❌Midjourney
❌Stable Diffusion

Ultralight Very light Light Medium Medium Heavy Ultraheavy

heavy

Prompt: an ultralight blue Prompt: a very light blue Prompt: a blue cube made of Prompt: a medium blue cube Prompt: a medium heavy Prompt: a heavy blue cube Prompt: an ultraheavy blue
cube on a scale cube on a scale light material on a scale on a scale blue cube on a scale on a scale cube on a scale

Angle ❌Midjourney
❌Stable Diffusion

5 degree 30 degree 50 degree 90 degree 120 degree 170 degree

Prompt: front view of 2 blue Prompt: front view of 2 blue Prompt: front view of 2 blue Prompt: front view of 2 blue Prompt: front view of 2 blue Prompt: front view of 2 blue
sticks forming a 5 degree sticks forming a 30 degree sticks forming a 50 degree sticks forming a 90 degree sticks forming a 120 degree sticks forming a 170 degree
angle angle angle angle angle angle
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 17

Area ✅Midjourney
❌Stable Diffusion

Extremely Very Narrow Slightly Very Extremely

narrow narrow wide wide wide

Prompt: an extremely Prompt: an very narrow Prompt: a narrow blue area Prompt: a slightly wide Prompt: a very wide Prompt: an extremely wide
narrow blue area blue area blue alley blue alley blue alley, extremely wide

Position ❌Midjourney
✅Stable Diffusion

Left Slightly left Middle Slightly right Right

Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue Prompt: front view of a blue
cube on the left, white cube on the slightly left, cube on the middle of the cube on the slightly right, cube on the right, white
background white background frame, white background white background background
Note: when generates position, DALL-E tends to generate the opposite direction compared to us looking at the picture. Maybe it uses the subject as the position reference so the blue cube will
stand on our left on the actual scene and then it takes the photo and it will appear on the right side of the photo.
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 18

Distance ❌Midjourney
❌Stable Diffusion

Extremely Very close Close Slightly close Far Very far Extremely far
close

Prompt: aerial view of a Prompt: aerial view of a Prompt: aerial view of a Prompt: aerial view of a Prompt: aerial view of a Prompt: aerial view of a Prompt: a blue cube
blue cube standing extremely blue cube standing very close blue cube standing close to a blue cube standing slightly blue cube standing far from blue cube standing very far standing extremely far to a
close to a person on to a person on white sand person on white sand close to a person on white a person on white sand from a person on white sand person on white sand
white sand sand

Dispersion ❌Midjourney
❌Stable Diffusion

Extremely Very Concentrated Slightly Slightly Diffused Very Extremely

concentrated concentrated concentrated Diffused diffused diffused

Prompt: Extremely Prompt: Very concentrated Prompt: Concentrated blue Prompt: Slightly Prompt: Slightly diffused Prompt: Diffused blue cube Prompt: Very diffused blue Prompt: Extremely diffused
extremely concentrated blue blue cube particles cube particles concentrated blue cube blue cube particles particles cube particles blue cube particles
cube particles particles
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 19

Shape ❌Midjourney
✅Stable Diffusion Shape (cont.) ❌Midjourney
✅Stable Diffusion

Irregular Regular Well rounded Rounded Sub-rounded Sub-angular Angular

Prompt: irregular blue Prompt: regular blue

shapes with equal angle Prompt: a cutout blue Prompt: a cutout blue Prompt: a cutout blue Prompt: a cutout blue Prompt: a cutout blue
shapes on the table front
corners on the table paper with a well rounded paper with a rounded shape paper with a subrounded paper with a subangular paper with an angular shape
view
shape shape shape

Erosion ❌Midjourney
✅Stable Diffusion

Extremely Very rusty Slightly Slightly Shiny Extremely

rusty rusty shiny shiny

Prompt: an extremely rusty Prompt: a very rusty Prompt: a slightly rusty Prompt: a slightly shiny Prompt: a shiny Prompt: an extremely shiny
blue cube blue cube blue cube blue cube blue cube blue cube, extremely shiny
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 20

Roughness ✅Midjourney
❌Stable Diffusion

Extremely Very smooth Smooth Rough Very rough Extremely

smooth rough

Prompt: an extremely Prompt: a very smooth Prompt: a smooth blue cube Prompt: a rough blue cube Prompt: a very rough Prompt: an extremely rough
smooth blue cube, blue cube blue cube blue cube
extremely smooth

Glossiness ✅Midjourney
✅Stable Diffusion

Matte Slightly Slightly Shiny

matte shiny

Prompt: a matte blue cube Prompt: a slightly matte blue Prompt: a slightly shiny blue Prompt: a shiny blue cube,
with matte material, close up cube, close up cube, close up close up
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS
QUANTITATIVE
FOR AISEMANTIC FIELDS 21

Subjective semantic
families
We call subjective families the ones that their qualitative
attributes have different interpretations depending on
the context you show it. So for example a thumbs up
(the gesture) is widely recognised sign of approval or
agreement BUT is actually used as an insult in
Bangladesh. This contextual approach is basic to
understand that the families that you will find now are
words that mean different things for different people,
cultures and contextual realities and they could be
misunderstood both by the AI and the audience.

Disclaimer: We will be using as a common starting object a blue cube so we can

compare the different effects of the adjectives. Said that in some of the linguistic
families we will need to change the blue cube for another base prompt to be able to
illustrate the visualization idea behind, as not everything can be exemplified by a
blue cube. (For example when talking about complex emotions of facial
expressions)
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 22

Temporality ❌Midjourney
✅Stable Diffusion

Ancient Antique Vintage Old Relatively New Extremely

new new

Prompt: an ancient blue Prompt: an antique blue Prompt: a vintage blue cube Prompt: an old blue cube Prompt: a relatively new Prompt: a new blue cube Prompt: an extremely new
cube in museum cube in museum in museum in museum blue cube in museum in museum blue cube in museum

Value ❌Midjourney
❌Stable Diffusion

Worthless Very cheap Cheap Austere Low-cost Expensive Very Luxurious Precious
expensive

Prompt: worthless blue Prompt: very cheap blue Prompt: cheap blue watches Prompt: austere blue Prompt: low-cost blue Prompt: expensive blue Prompt: very expensive Prompt: luxurious Prompt: precious
watches watches watches watches watches blue watches blue watches blue watches
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 23

Order ❌Midjourney
❌Stable Diffusion

Very chaotic Chaotic Organized Very organized

Prompt: very very chaotic Prompt: chaotic blue cubes Prompt: organized blue Prompt: very organized
blue cubes on the table on the table cubes on the table blue cubes on the table

Toughness ✅Midjourney
✅Stable Diffusion

Extremely Very soft Soft Slightly Slightly Rigid Very rigid Extremely
soft soft rigid rigid

Prompt: blue cube made of Prompt: blue cube made of Prompt: blue cube made of Prompt: blue cube made of Prompt: blue cube made of Prompt: blue cube made of Prompt: blue cube made of Prompt: blue cube made of
extremely soft material on the very soft material on the soft material on the table, slightly soft material on the slightly rigid material on the rigid material on the table, very rigid material on the extremely rigid material on
table, close up table, close up close up table, close up table, close up close up table, close up the table, close up
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 24

Openness ❌Midjourney
❌Stable Diffusion

Sealed Locked Closed Half closed Slightly Slightly Half opened Opened
closed opened

Prompt: sealed blue letter, Prompt: locked small blue Prompt: closed small blue Prompt: half closed small Prompt: slightly closed Prompt: slightly opened Prompt: half opened small Prompt: opened small blue
realistic photo box box blue box small blue box small blue box blue box box
Note: didnʼt generate it

Turbulence ✅Midjourney
❌Stable Diffusion

Relaxed Balanced Moderate Unstable Severe Extreme

Prompt: fluid in relaxed Prompt: balanced Prompt: moderate Prompt: unstable Prompt: severe turbulence Prompt: extreme turbulence
state, top view turbulence fluid, top view turbulence fluid turbulence fluid fluid fluid
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 25

Resolution ❌Midjourney
❌Stable Diffusion

Extremely Pixelated Low High 4k 8k 10k

pixelated

Prompt: an extremely Prompt: image with a Prompt: a low resolution, Prompt: image with a high Prompt: image with a 4K Prompt: image with a 8K Prompt: image with a 10K
pixelated resolution blue pixelated, bad quality bad quality image of a blue resolution of a blue cube on resolution of a blue cube on resolution of a blue cube on resolution of a blue cube on
cube on the sand resolution of a blue cube on the cube on the sand the sand the sand the sand the sand
sand
Note: itʼs hard for DALL-E to
give bad quality images

Resolution (cont.) ❌Midjourney

✅Stable Diffusion

Poorly drawn Basic drawing Detailed Extremely Realistic Hyper realistic

drawing detailed
drawing drawing drawing

Prompt: poorly drawn blue Prompt: basic drawing of a Prompt: detailed drawing of Prompt: an extremely Prompt: realistic drawing of Prompt: hyper realistic
cube and sand blue cube and sand a blue cube and sand detailed drawing of a blue a blue cube and sand drawing of a blue cube and
cube and sand sand
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 26

Focus ✅Midjourney
❌Stable Diffusion

Completely out Very blurry Medium Slightly In focus Sharp focus Extremely
of focus blurry blurry sharp focus

Prompt: completely out of Prompt: very blurry photo Prompt: medium blurry Prompt: slightly blurry Prompt: in focus photo of a Prompt: sharp focus photo Prompt: extremely sharp
focus photo of a blue cube of a blue cube photo of a blue cube photo of a blue cube blue cube of a blue cube focus photo of a blue cube

Image ❌Midjourney
❌Stable Diffusion

Severely Highly distorted Medium Slightly Undistorted

distorted distorted distorted

Prompt: severely distorted Prompt: highly distorted Prompt: medium distorted Prompt: slightly distorted Prompt: undistorted image of
image of a blue cube image of a blue cube image of a blue cube image of a blue cube a blue cube, realistic photo
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 27

Image (cont.) ❌Midjourney

❌Stable Diffusion

Severely Highly distorted Medium Slightly Undistorted

distorted distorted distorted

Prompt: severely distorted Prompt: highly distorted Prompt: medium distorted Prompt: slightly distorted Prompt: undistorted image of
image of people image of people image of people image of people people

Body weight ✅Midjourney

✅Stable Diffusion

Anorexic Skiny Well built Muscular Chubby Fat Overweight Overly obese

Banned word in DALL-E

Prompt: front view of a Prompt: front view of a Prompt: front view of a Prompt: front view of a Prompt: front view of a fat Prompt: front view of an Prompt: front view of an
skinny body, white well-built body, white muscular body, white chubby body, white body, white background, overweight body, white overly obese body, white
background, studio lighting background, studio lighting background, studio lighting background, studio lighting studio lighting background, studio lighting background, studio lighting
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 28

Skin tone ✅Midjourney

✅Stable Diffusion

Pale white Fair Medium Olive Brown Black

Prompt: front view of pale Prompt: front view of fair Prompt: front view of Prompt: front view of olive Prompt: front view of Prompt: front view of black
white skin tone skin tone medium skin tone skin tone brown skin tone skin tone

Skin ✅Midjourney
✅Stable Diffusion

Velvety Silky Smooth Soft Wrinkled Very Crumpled

wrinkled

Prompt: front view of Prompt: front view of silky Prompt: front view of Prompt: front view of soft Prompt: front view of Prompt: front view of very Prompt: front view of a face
velvety skin skin smooth skin skin wrinkled skin wrinkled skin with crumpled skin
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 29

Hair ✅Midjourney
✅Stable Diffusion

Extremely Very hairy Hairy Slightly Bald

hairy hairy

Prompt: extremely hairy head Prompt: very hairy head Prompt: hairy head Prompt: slightly hairy head Prompt: bald head
Note: long shot is banned
with hair prompts and only
men came out

Hair (cont.) ✅Midjourney

✅Stable Diffusion

Extremely Very hairy Hairy Slightly Bald

hairy hairy

Prompt: extremely hairy Prompt: very hairy blue cube Prompt: hairy blue cube Prompt: slightly hairy Prompt: bald blue cube
blue cube blue cube
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 30

Mood ❌Midjourney
✅Stable Diffusion

Agonized Very sad Sad Slightly Slightly Happy Very Delighted

sad happy happy

Prompt: Portrait of an Prompt: Portrait of a very Prompt: Portrait of a sad Prompt: Portrait of a Prompt: Portrait of a Prompt: Portrait of a happy Prompt: Portrait of a very Prompt: Portrait of a
agonized face, color sad face, color photography face, color photography slightly sad face, color slightly happy face, color face, color photography happy face, color delighted face, color
photography photography photography photography photography

Mood (cont.) ❌Midjourney

✅Stable Diffusion

Extremely Very Disappointed Slightly Slightly Satisfied Very Extremely

disappointed disappointed disappointed satisfied satisfied satisfied

Prompt: Portrait of an Prompt: Portrait of a very Prompt: Portrait of a Prompt: Portrait of a Prompt: Portrait of a Prompt: Portrait of a Prompt: Portrait of a very Prompt: Portrait of an
extremely disappointed face, disappointed face, color disappointed face, color slightly disappointed face, slightly satisfied face, color satisfied face, color satisfied face, color extremely satisfied face,
color photography photography photography color photography photography photography photography color photography
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 31

Mood (cont.) ❌Midjourney

✅Stable Diffusion

Furious Very angry Angry Slightly angry Slightly calm Calm Tranquill

Prompt: Portrait of a Prompt: Portrait of a very Prompt: Portrait of an Prompt: Portrait of a Prompt: Portrait of a Prompt: Portrait of a calm Prompt: Portrait of a
furious face, color angry face, color angry face, color slightly angry face, color slightly calm face, color face, color photography tranquill face, color
photography photography photography photography photography photography

Mood (cont.) ❌Midjourney

✅Stable Diffusion

Mortiferous Aggressive Hostile Neutral Friendly Cute Adorable

Prompt: Portrait of a Prompt: Portrait of an Prompt: Portrait of a hostile Prompt: Portrait of a Prompt: Portrait of a Prompt: Portrait of a cute Prompt: Portrait of an
mortiferous face, color aggressive face, color face, color photography neutral face, color friendly face, color face, color photography adorable face, color
photography photography photography photography photography

Note: DALL-E and Stable Diffusion both shows gender bias when generating human expression showing male when it scales towards aggression and female on the opposite spectrum
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 32

Food ✅Midjourney
✅Stable Diffusion

Very Unhealthy Healthy Very healthy

unhealthy

Prompt: very unhealthy Prompt: unhealthy food Prompt: healthy food Prompt: very healthy food
food

Food (cont.) ✅Midjourney

❌Stable Diffusion

Uncooked Rare Medium rare Well done Overcooked

Prompt: uncooked food Prompt: rare food Prompt: medium rare food Prompt: well done food Prompt: overcooked food
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 33

Living stage ✅Midjourney

❌Stable Diffusion

Moldy Very rotten Rotten Half rotten Slightly rotten Fresh

Prompt: a moldy apple, Prompt: a very rotten apple, Prompt: a rotten apple, Prompt: a half rotten apple, Prompt: a slightly rotten Prompt: a fresh apple,
highly detailed, white highly detailed, white highly detailed, white highly detailed, white apple, highly detailed, white highly detailed, white
background background background background background background

Living stage (cont.) ✅Midjourney

✅Stable Diffusion

Infant Baby Young Adult Senior Ancient Fossil

Prompt: an infant human Prompt: a baby human Prompt: a young human Prompt: an adult human Prompt: a senior human Prompt: an ancient human Prompt: a fossil human
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 34

Weather ✅Midjourney
✅Stable Diffusion

Stormy Thunder Pouring Rainy Foggy Cloudy Sunny

Prompt: stormy weather Prompt: thunder weather Prompt: pouring weather Prompt: rainy weather Prompt: foggy weather Prompt: cloudy weather Prompt: sunny weather
outside outside outside outside
Note: DALL-E shows
raindrops on the glass mainly

Weather (cont.) ✅Midjourney

✅Stable Diffusion

Freezing Cold Chilly Cool Warm Hot Melting hot

Prompt: freezing weather Prompt: cold weather Prompt: chilly weather Prompt: cool weather Prompt: warm weather Prompt: hot weather Prompt: melting hot
outside outside outside outside outside outside weather outside
Note: DALL-E shows it on the Note: DALL-E took it
glass literally
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 35

Pollution ✅Midjourney
✅Stable Diffusion

Extremely Very polluted Medium Slightly Clean Extremely clean Pure Pristine
polluted polluted polluted

Prompt: extremely polluted Prompt: very polluted water Prompt: medium polluted Prompt: slightly polluted Prompt: clean river Prompt: extremely clean Prompt: pure river Prompt: pristine water
water water water river

Wetness ✅Midjourney
❌Stable Diffusion

Dehydrated Dry Unmoistened Moist Humid Wet

Prompt: dehydrated soil Prompt: dry soil Prompt: unmoistened soil Prompt: moist soil Prompt: humid soil Prompt: wet soil
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 36

Moon phase ✅Midjourney

✅Stable Diffusion

New Waxing First quarter Waxing Full moon Waning Last quarter Eclipse
crescent gibbous gibbous

Prompt: nasa picture of Prompt: nasa picture of Prompt: nasa picture of first Prompt: nasa picture of Prompt: nasa picture of Prompt: nasa picture of Prompt: nasa picture of Prompt: nasa picture of
new moon waxing crescent moon quarter moon waxing gibbous moon full moon waning gibbous moon Last quarter moon eclipse

Sun position ✅Midjourney

✅Stable Diffusion

Sunrise Morning Noon Afternoon Sunset

Prompt: sunset sun position Prompt: morning sun Prompt: noon sun position Prompt: afternoon sun Prompt: sunset sun position
on the sea position on the sea on the sea position on the sea on the sea
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 37

Technology ✅Midjourney
✅Stable Diffusion

Prehistoric Ancient Outdated Old Modern Contemporary Advanced Futuristic

Prompt: a prehistoric tool Prompt: an ancient tool Prompt: an outdated tool Prompt: an old tool Prompt: a modern tool Prompt: a contemporary tool Prompt: an advanced tool Prompt: a futuristic tool

Technology (cont.) ✅Midjourney

❌Stable Diffusion

Prehistoric Ancient Outdated Old Modern Contemporary Advanced Futuristic

Prompt: a mobile phone Prompt: a mobile phone Prompt: a mobile phone Prompt: a mobile phone Prompt: a mobile phone Prompt: a mobile phone Prompt: a mobile phone Prompt: a mobile phone
designed with prehistoric designed with ancient designed with outdated designed with old designed with modern designed with contemporary designed with advanced designed with futuristic
technology technology technology technology technology, studio technology, studio technology, studio technology
photography, front view photography, front view photography, front view
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 38

Stage of development ✅Midjourney

✅Stable Diffusion

Under- Early Developing Half- Late Fully Highly Over

developed developed developed developed developed developed developed

Prompt: underdeveloped area Prompt: early developed area Prompt: developing area Prompt: half-developing area Prompt: late developed area Prompt: fully developed area Prompt: highly developed Prompt: over developed area
area

Sound frequency ✅Midjourney

✅Stable Diffusion

Muted Very low Low pitched Pleasant Slightly high High pitched Very high Noisy
pitched pitched pitched

Prompt: muted waveform Prompt: very low pitched Prompt: low pitched Prompt: pleasant waveform Prompt: slightly high pitched Prompt: high pitched Prompt: very high pitched Prompt: noisy waveform
waveform waveform waveform waveform waveform
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS
QUANTITATIVE
FOR AISEMANTIC FIELDS 39

Build your own

semantic fields
This section is to give you a step by step guide
to building your own spectrum of semantic
fields for your own projects.
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 40

Step 1: Find your word

Find an adjective that you think can add interesting qualitative ideas
behind the data you want to express.

For example I have some information about pollution on a certain area that I
want to talk about, in this case the adjective to use will be:

“polluted”
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 41

Step 2: Antagonize
Then you need to first find the opposite word to the one you have first
defined. For doing so we recommend using free webs like Thesaurus.

polluted clean
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 42

Step 3: Expand
Afterwards you have to explore word past your antonyms:

Pure /
noxious polluted clean
Pristine
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 43

Step 4: Define the spectrum

Then you can use some of these quantitative modifiers to build a
richer spectrum:

Quantitative modifiers
Insignificantly Lightly Slightly Very Highly Extremely Ultra

You can use these modifiers in both directions, in the one of the
original adjective and the one of the opposite:

Pollution
Noxious Extremely Very polluted Slightly Clean Extremely clean Pure Pristine
polluted polluted
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 44

Step 4.2: Diversify

And of course you can think about more quantitative modifiers that can
add richness and comparative value to your prompt.

Radically Violently Strongly Hazardous Moderate Fair Harmless Pure Divine

 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 45

Step 5: Prompt it
With that you will be ready to write the prompts for each stage of the
spectrum.
We recommend giving a bit more context to the description of the prompt
so the different images have less variability and itʼs easier to recognize the
quantitative differences in them. For that we recommend defining the
environment, such as the background, the kind of shot, pictorial
technique and several other specs. You can check our first prompt book
for more ideas on that.

Pollution
Radically Violently Strongly Hazardous Moderate Fair Harmless Pure Divine

Prompt: radically air Prompt: violently air Prompt: strongly air Prompt: hazardous air Prompt: moderate air in a Prompt: fair air in a city Prompt: harmless air in a Prompt: pure air in a city Prompt: divine air in a city
polluted city polluted city polluted city pollution in a city city DALL-E city DALL-E DALL-E
DALL-E DALL-E DALL-E DALL-E DALL-E DALL-E
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 46

Step 6: Render
And finally the only thing you need is to render it:

Pollution
Radically Violently Strongly Hazardous Moderate Fair Harmless Pure Divine

Prompt: radically air Prompt: violently air Prompt: strongly air Prompt: hazardous air Prompt: moderate air in a Prompt: fair air in a city Prompt: harmless air in a Prompt: pure air in a city Prompt: divine air in a city
polluted city polluted city polluted city pollution in a city city DALL-E city DALL-E DALL-E
DALL-E DALL-E DALL-E DALL-E DALL-E DALL-E
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 47

Step 7: Create you data story

Air Pollution rank between countries

Finland Poland Senegal Uganda Pakistan Nigeria Nepal

PM2.5 level: 5.86 PM2.5 level: 20.88 PM2.5 level: 40.7 PM2.5 level: 50.49 PM2.5 level: 58.28 PM2.5 level: 71.8 PM2.5 level: 99.73

Prompt: rock in the shape of Prompt: rock in the shape of Prompt: rock in the shape of Prompt: rock in the shape of Prompt: rock in the shape of Prompt: rock in the shape of Prompt: rock in the shape of
Finland map covered in Poland map covered in green Senegal map covered sparsely Uganda map covered in Pakistan map covered in Nigeria map covered in Nepal map, extremely dead
extremely extremely green moss, top view white in green moss, top view slightly slightly green moss, slightly dried yellowish moss, extremely brownish dead rock, no sign of life, top view
moss, top view white background white background top view white background top view white background moss, completely dried moss, white background
background top view white background

PM2.5 levels, micrograms per cubic meter of air (µg/m³)(2017)

higher number means more polluted (scale from 0 to 100) Source: Air Pollution across the world
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 48

Higher GDP – less air pollution I

This data visualization shows level of air pollution in different countries. Fine particulate matter (PM2.5) is an air pollutant that is a concern for people's health when
levels in air are high. The Gross Domestic Product per capita, or GDP per capita, is a measure of a country's economic output that accounts for its number of people.

Mali Tunisia Thailand Trinidad & Tobago Hungary UK Finland

PM2.5 level: 38.53 PM2.5 level: 37.66 PM2.5 level: 26.26 PM2.5 level: 24.11 PM2.5 level: 15.93 PM2.5 level: 10.47 PM2.5 level: 5.86
GDP per capita: 2216.77 GDP per capita: 9727.5 GDP per capita: 17.287 GDP per capita: 23.728 GDP per capita: 31.007 GDP per capita: 41.627 GDP per capita: 47.260

Prompt: extremely rusty Prompt: very rusty coins one Prompt: rusty coins one on Prompt: slightly rusty coins Prompt: slightly shiny coins Prompt: shiny coins one on Prompt: very shiny coins
coins one on top of another on top of another with top of another with Thailand one on top of another with a one on top of another with top of another with British one on top of another with
with Mali flag long shot Tunisia flag long shot white flag long shot white flag of Trinidad and Tobago Hungary flag long shot white flag long shot white Finland flag long shot white
white background background background long shot white background background background background

Note: this graph is not the best, since columns of coins are not precise and might be confusing. The flags of countries occupy a lot space and donʼt need to be shown
when we try to visualize air pollution and GDP per capita in each country. Letʼs try again in the next example.
Source: Air Pollution across the world
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 49

Higher GDP – less air pollution II

This data visualization shows level of air pollution in different countries. Through a coin which evolves from extremely rusty to very shiny, we show the air pollution,
simultaneously, from micro coin to very very huge coin, we show countryʼs GDP per capita level.

Mali Tunisia Thailand Trinidad & Tobago Hungary UK Finland

PM2.5 level: 38.53 PM2.5 level: 37.66 PM2.5 level: 26.26 PM2.5 level: 24.11 PM2.5 level: 15.93 PM2.5 level: 10.47 PM2.5 level: 5.86
GDP per capita: 2216.77 GDP per capita: 9727.5 GDP per capita: 17.287 GDP per capita: 23.728 GDP per capita: 31.007 GDP per capita: 41.627 GDP per capita: 47.260

Prompt: extremely rusty Prompt: very rusty tiny Prompt: rusty small coin, Prompt: slightly rusty Prompt: slightly shiny big Prompt: shiny very big coin, Prompt: very shiny very
micro coin, white coin, white background, top white background, top view medium size coin, white coin, white background, top white background, top view very huge coin, white
background, top view view background, top view view background, top view

Source: Air Pollution across the world

 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 50

Mass shootings in the US between 1985 and 2018

Nobody Extremely sparse Sparse Slightly dense Low dense

Prompt: aerial view of Prompt: aerial view of Prompt: aerial view of Prompt: aerial view of Prompt: aerial view of low dense
nobody on the street extremely sparse group of sparse group of people on slightly dense group of people group of people on the street
people on the street the street on the street

1985: 0 shooting 1989: 2 shootings 2007: 4 shootings 2012: 7 shootings 2018: 12 shootings

Source: Number of mass shootings in the United States between 1982 and October 2022
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 51

Mass shootings in the US between 1985 and 2018

Zero 2 4 7 12

Prompt: top view of 2 copper

bullets lying on white table,
straight, white background

Prompt: top view of no

bullets lying on white table, 1989: 2 shootings
straight, white background Prompt: top view of 4 copper
bullets lying on white table, Prompt: top view of 7 copper
Note: didnʼt work in the straight, white background bullets lying on white table,
case of ʻno thingʼ straight, white background

1985: 0 shooting

2007: 4 shootings
2012: 7 shootings

Prompt: top view of 12 copper bullets

lying on white table, straight, white
background
2018: 12 shootings
Source: Number of mass shootings in the United States between 1982 and October 2022
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 52

Global coral coverage over the last decade

35

30
Hard Coral Cover (%)

Prompt: a tall cement block with

colorful corals growing on it, white
background, studio photography

Editing and erasing tool was used to

generate this graph with bars in
different height

25
Source: Status of Coral
2010 2015 2018 2020 Reefs of the World: 2020
 DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 53

Global coral coverage over the last decade

35

30
Hard Coral Cover (%)

DALL-E image is enhanced with

Dream Studio

Prompt: a cement block with colorful

corals growing on it, extremely detailed
corals, extremely realistic corals,
contemporary installation,
white background
25
Source: Status of Coral
2010 2015 2018 2020 Reefs of the World: 2020
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 54

Ice melting in Greenland

This data visualization shows a big cube of ice that represents Greenland and through the years it can be seen how much ice loss is happening. Ice melting is shown
with a metrical unit gigattone.

2006 loss: 1300 2010 loss: 2600 2014 loss: 3200 2018 loss: 3800 2022 loss: 4100
gigatonnes gigatonnes gigatonnes gigatonnes gigatonnes

Prompt: a vertical long cube of ice in 6

blocks in a full height on the plain
floor, white background, full shot, front
view

Editing and erasing tool was used to

generate this graph with bars in
different height
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 55

Happiness scores at work (scale 1 to 6) Source: Feeling at work

22.5%
22%
21.5%
17.5%

8%

4.5%

3%

Extremely sad, in tears Extremely sad Sad Neutral Slightly happy Happy Extremely happy
Prompt: an extremely short robot with Prompt: an extremely extremely Prompt: a sad robot with Prompt: a very very tall robot, Prompt: a very very tall Prompt: a very very tall robot Prompt: a very very tall robot
LED face showing extremely extremely sad robot with LED face showing LED face showing his with LED face showing a robot with LED face showing with LED face showing happy standing with LCD face
sad face in tears, front view, white his extreme sadness, extremely sadness, tall, white neutral expression, white slightly happy face, white face, white background, front showing extremely extremely
background, studio photography short, white background, front background, front view, background, front view, background, front view, view, studio photography happy face, white background,
view, studio photography studio photography studio photography studio photography front view, studio photography
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 56

Happiness scores at work (scale 1 to 6) Source: Feeling at work

22.5%
22%
21.5%
17.5%

8%
Note: when generating
human emotion but not in
4.5% ʻhumanʼ context, the result is
not as accurate as the prompt
describe. In this case, thereʼs
not much difference between
ʻextremely sad in tearsʼ and 3%
ʻextremely sadʼ

Note: when using editing

tool in DALL-E, the more
ʻgenerating framesʼ added
to the image, it started to
become ʻless realisticʼ.
For example when it
reached to this robot, the
robot is like a drawing
instead of realistic
material like the
robots before

Extremely sad, in tears Extremely sad Sad Neutral Slightly happy Happy Extremely happy
Prompt: an extremely short robot with Prompt: an extremely extremely Prompt: a sad robot with Prompt: a very very tall robot, Prompt: a very very tall Prompt: a very very tall robot Prompt: a very very tall robot
LED face showing extremely extremely sad robot with LED face showing LED face showing his with LED face showing a robot with LED face showing with LED face showing happy standing with LCD face
sad face in tears, front view, white his extreme sadness, extremely sadness, tall, white neutral expression, white slightly happy face, white face, white background, front showing extremely extremely
background, studio photography short, white background, front background, front view, background, front view, background, front view, view, studio photography happy face, white background,
view, studio photography studio photography studio photography studio photography front view, studio photography
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 57

Conclusions
We hope you like it, actually…. if you are still
reading this, it means that you are interested as
heck. Well, we hope that this language
exploration can be a first step to bridge the gap
between the qualitative and the quantitative
ways we can express meaning.

Maybe this contribution can help explore new

ways to further disseminate valuable
information, ideas, poetry and science.

Fingers crossed…
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS
QUANTITATIVE
FOR AISEMANTIC FIELDS 58

Want to
contribute?
If you want to be part of this book and add new
sections with your best quantitative semantic
fields, write us a short email with your idea at
ai@domesticstreamers.com and we will reach out
to involve you in the next edition of the book.
🌝👍
DOMESTIC DATA STREAMERS QUANTITATIVE SEMANTICS FOR AI 59

For more news & resources:

Follow us on Instagram, Linkedin, Twitter or
subscribe to our beloved newsletter.
Ping us at: ai@domesticstreamers.com
Lots of love and data! Bona nit!