My prompts tend to get away from me, extensive in determination and detail, and their structure tended to vary wildly from one mini project to another. So I got organized. This was possible because I installed and enabled the Dynamic Prompts extension, which allows you to mark lines as comments, excluded from the actual prompt itself, only holding your notes.
For different aims I have different templates. For images that feature humans, I use the following, with examples for each section. The sections guide me to enter what's important, and I ignore or leave out line items for what's not important.
# POSITIVE CATEGORY : HUMANS
# SETUP: SHOOT, FOCUS, DETAIL, LIGHT
model shoot, crystal clear, intricate, sunlight,
# MODEL/PERSON
a gorgeous forty year old model,
# LOCATION / BACKGROUND
on the shore of the Isle of Man,
# GENRE
vacation,
# WEARING/CLOTHING: OUTFIT/DRESS, SHIRT, PANT/SHORTS/SKIRT, SHOES, HAT, OR NUDE
winter clothes, fur hat,
# VIEWING / LOOKING AT
looking at viewer,
# BODY ATTRIBUTES: SKIN TYPE N COLOR, BODY PARTS (NOSE...)
creamy skin,
# EYES : COLOR, LIDDED, EYE ENHANCEMENT
brown eyes,
# HAIR COLOR
{__hair_colors__},
# HAIR STYLE
ponytail,
# POSITION / ORIENTATION
sitting on a park bench,
# EXPRESSION
smirking,
# ACTIVITY
sticking her tongue out,
# EXTRAS
# ENHANCERS (OVERALL)
<lora:wowifierV3:500.1>, <lyco:add_a_box_of_crazy_details>
Highly organized, but if you also turn on the Dynamic Prompts setting for turning all whitespace into a single space each, it doesn't eat up your tokens, as this is what that template actually spills out:
model shoot, crystal clear, intricate, sunlight, a gorgeous forty year old model, on the shore of the Isle of Man, vacation, winter clothes, fur hat, looking at viewer, creamy skin, brown eyes, blonde, ponytail, sitting on a park bench, smirking, sticking her tongue out, <lora:wowifierV3:1.4>, <lyco:add_a_box_of_crazy_details>
...and this is also the cover image for the article. The template I use for artworks is a little stranger, with different sections. :)
I've also created an organization for my negative prompts. Most are incorporated into Automatic1111 styles now, but the exercise to organize was informative and will pay off immediately when I need to amend a negative prompt. Here's the template I use, with a few examples. With Dynamic Prompts enabled, I keep the # comment lines in.
# NEGATIVE CATEGORY : ART PAINTINGS / PGM
# TASK/PROJECT SPECIFIC
(leather helmet),
# QUALITY
(blurry:1.3),
# COMPOSITION and COLORS
monochrome, desaturated, face out of frame,
# DEFORMITY
extra heads,
# WATERMARKS
logo, signature,
# GENERAL EMBEDDINGS
bad-ears-mark-229,
I hope you find this technique as useful as I do.