Search Write

StableDiffusion

This member-only story is on us. Upgrade to access all of Medium.

Member-only story Featured

STABLE DIFFUSION

Mastering SDXL Prompts ( 1 )

Advanced Guide for Lens
Perspective
Combining different lens distances, angles and scenes forms
compositions

Edmond Yip Follow 6 min read · Apr 30, 2024

186 2

Mastering SDXL Prompts ( 1 ) Advanced Guide for Lens Perspective (Image Generate by Edmond Yip)

Read full story here if you not a paid member

In the realm of Stable Diffusion, mastering the art of lens perspective with
prompts alone is now achievable, eliminating the need for ControlNet or
other extensions. This advanced guide delves into controlling lens
perspective solely through prompts, introducing new scene composition
prompts for reference.

Exploring the capabilities of the SDXL model unveils its superior

comprehension of natural language, empowering users to describe desired
visuals using plain language alongside prompts.

Basic Setting
For this experiment, Animageine XL serves as the checkpoint model.

Animagine XL V3.1 - v3.1 | Stable Diffusion Checkpoint | Civitai

Animagine XL 3.1 is an update in the Animagine XL V3 series, enhancing the previous
version, Animagine XL 3.0. This…
civitai.com

The placeholder {prompts} will be substituted with various focal length

prompts, each assigned a weight of 1.5 .

Basic Setting

prompts： {prompts} of a girl, brown hair, streetwear, in new york

broadway, intricate detailed, looking at viewer

Negative prompts : low quality, worst quality, out focus, blur

Lens Distances
The exploration will categorize focal lengths, starting from close distances,
moving to mid-range, and ending with far distances.

1 ) Extreme Close-Up shot / Close-Up shot

The extreme close up shot resembles proximity without the eye
enlargement effect seen in SD 1.5.

To achieve eye enlargement, append of eye after extreme close up shot .

This technique can be applied to different body parts, rendering them

finely detailed and slightly enlarged.

Extreme Close Up shot / Close-Up shot (Image Generate by Edmond Yip)

extreme close up shot will enlarge body parts (Image Generate by Edmond Yip)

2 ) Mid-Range / Upper Body

upper body , portrait depict the head and shoulders.

medium shot , chest_shot , half body will show body above waist.

cowbody_shot , knee_shot will show body above thigh.

For instance, resolving the issue with cowboy_shot by inserting an

underscore between cowboy and shot .

chest_shot , cowboy_shot , knee_shot , and waist_shot may alter clothing.

Mid-Range (Image Generate by Edmond Yip)

Upper Body (Image Generate by Edmond Yip)

3 ) Full Body Shot

full shot or full body shot reveals the entire figure from head to toe.

establishing shot position characters slightly lower than long shots for
broader background views.

long shot or extreme long shot capture upper body to mid-thigh

regions, occasionally appearing smaller.

Full Body Shot (Image Generate by Edmond Yip)

Lens Angles
A variety of angles are covered, streamlined for easy reference.

1 ) Front View
dutch angle shot maintains a 45-degree tilt while keeping the subject
frontal.

selfies add a touch of cuteness with hand-held self-portraits and

inherent beautification effects.

Front View (Image Generate by Edmond Yip)

Side View
side view , lateral view and flank view offer random left/right side
angles.

profile and side view with the character appearing more elegant in
profile shots.

Side View (Image Generate by Edmond Yip)

Back View
back view , from behind or back depict the majority of the back.

Turn around signify a shift in perspective.

Back View (Image Generate by Edmond Yip)

High Angle View

bird_eye_view introduces barrel distortion in the background.

high_angle like bird_eye_view or top down view offer elevated

perspectives.

overhead shots and from above minimize background distortion.

high_angle or slightly above perspectives are slightly lower.

High Angle View (Image Generate by Edmond Yip)

Low Angle View

low_angle , low_view slightly lower the perspective.

worm_eye_view may inadvertently introduce bug-like patterns, distorting

eyes into goggles or compound eyes.

from bottom views occasionally include the buttocks.

Low Angle View (Image Generate by Edmond Yip)

Wide Angle View

ultra wide angle view and fisheye_shot warp backgrounds into circular
forms.

fisheye_shot may impact image quality.

wide angle view exhibit minimal background distortion.

upside_down_shots reverse the entire scene, sometimes creating circular

backgrounds.

Wide Angle View (Image Generate by Edmond Yip)

Dynamic Shots
We change prompts to a car driving in New York City, with a {prompts} of

the background complement scenes, as the effectiveness of using a girl as the

subject may vary.

pan shot , tracking_shot and motion_blur introduce varying degrees of

motion blur into the background.

zoom burst incorporates comic-style diffusion lines across the entire

scene, including the subject and background, creating a dynamic and
immersive effect.

Dynamic Shots (Image Generate by Edmond Yip)

Architectural Perspectives
Shifting prompts to {prompt} of a clock tower in a city highlights angles
suitable for architectural scenes.

vanishing_point converge the scene towards a focal point, creating

linear depth.

aerial shot emulate drone footage, offering a high-angle and somewhat

wide-angle effect.

tilt_shift shots simulate miniature photography, featuring exaggerated

depth of field.

isometric views showcase buildings from a 45-degree angle, minimizing

distortion.

Architectural Perspectives (Image Generate by Edmond Yip)

Symmetrical Compositions
Using prompts a girl, a dress with {prompt} magic patterns of Shipibo-

Konibo, looking at the viewer to generates geometric shapes while

reducing weight to avoid odd distortions.

bilaterally symmetrical shots exhibit overall left-right symmetry, albeit

with color abnormalities.

vertical_symmetry or horizontal_symmetry showcases symmetry

variations, although characters may vanish in certain scenes.

bilateral_symmetry_shot and horizontal_symmetry_shot aim for a roughly

symmetrical composition, although color aberrations may occur at
times.

vertical_symmetry_shot may result in the disappearance of the girl due to

vertical symmetry.

Symmetrical Compositions (Image Generate by Edmond Yip)

Composition and Framing

Crafting Compositions with Varied Lens Distances, Angles, and Scenes.

Image Generate by Edmond Yip

(selfie:1.5), (wide angle view:1.5) of 1girl, black hair, air bubble,

floating in water, underwater ripples, sun shine

Image Generate by Edmond Yip

(from below:1.3), (profile:1.5) of a girl wearing a pink kimono, looking at

viewer, firework in the sky, intricate detailed

Image Generate by Edmond Yip

(close up shot:1.5) of a beautiful the cloud elf queen busks on the

streets, (wide angle view:1.3) of new york, playing guitar and singing,

dynamic pose, highly detailed, smooth, sharp focus, intricate detailed

Image Generate by Edmond Yip

(medium shot:1.5), (dutch angle shot:1.5) of a girl sitting with tuck knees

in a Capsule hotel, cyberpunk, neolight, intricate detailed

Image Generate by Edmond Yip

(side view:1.5), (medium shot:1.5) of a girl hold on to the tree vine, tree

from which the foliage is flying around, motion_blur background, intricate

detailed, looking at viewer

Image Generate by Edmond Yip

(medium shot:1.5), (back_view:1.5) of a girl, wearing Wedding dress, (back

light:1.5), in a church

Image Generate by Edmond Yip

(low view:1.5), (chest_shot:1.3) of a girl drawing a glowing light magic

circle in front of two hand, wind, intricate detailed

While this guide focuses on SDXL, its principles are applicable to other
models like Midjounery or Dalle 3. Experiment and learn to master
prompts, bringing your visions to life effortlessly!

For All Stable Diffusion Lovers

Thank you for being a part of the Stable Diffusion community!

Stable Diffusion Want Writer for Prompt Engineering, Technology

Guide and Model Training
Exploring the Intersection of Technology, AI Art, and Model Training
medium.com

Stable Diffusion Sdxl Guides And Tutorials Prompt Creativity

186 2

Published in StableDiffusion Follow

326 followers · Last published Jul 18, 2024

Story about Stable Diffusion, if you also want to public story here, please contact
me.

Written by Edmond Yip Follow

1.4K followers · 98 following

🤩 Latest News and Updates of Stable Diffusion. ✍Tutorial of installation,

extension and prompts for Stable Diffusion.
https://medium.com/@edmond.yip/subscribe

Responses (2)

Zeton

What are your thoughts?

Edward Re
Oct 9, 2024

hi Edmund,

very informative article.

Make thanks for sharing your tips

Reply

Daliy
Jul 25, 2024

Hello,

I'm wendy from Mimicpc，which is an online tool designed to help individuals who want to use AI tools such as
Stable Diffusion and ComfyUI but do not have high-end computers. we want to make friendly backlink
exchange by this article on your… more

Reply

More from Edmond Yip and StableDiffusion

In StableDiffusion by Edmond Yip In StableDiffusion by Edmond Yip

Understanding the Meaning of 100+ Negative Prompts everyone

score_9 in Prompts are using
How to ensure the generated image quality is In Stable Diffusion, although negative
consistently controllable prompts may not be as crucial as prompts,…

Jun 25, 2024 18 Jun 14, 2023 613 4

In StableDiffusion by Prompting Pixels In Bootcamp by Edmond Yip

Beginner’s Guide to a Text-to- Set up nginx-proxy with automatic

Image (txt2img) Workflow in… SSL certificate renewal in Docker…
Getting started with ComfyUI can seem Previously, I used jwilder/nginx-proxy and
overwhelming at first, but once you… jrcs/letsencrypt-nginx-proxy-companion to…

May 1, 2024 4 Apr 8, 2023 27 3

See all from Edmond Yip See all from StableDiffusion

Recommended from Medium

Webster In Engineering @ Draw Things by Liu Liu

Revolutionize AI Image Editing in Metal FlashAttention 2.0: Pushing

ComfyUI with FLUX.1 Tools for… Forward On-Device Inference &…
FLUX.1 Tools enhance ComfyUI with Metal FlashAttention underpins Draw Things’
advanced inpainting, outpainting, and imag… claim of fastest image generation inside the…

Nov 21, 2024 3 1 Jan 7 119 1

In 𝐀𝐈 𝐦𝐨𝐧𝐤𝐬.𝐢𝐨 by ArtfullyPrompt - Nathan Cash In Prompt Studio by Konstantinos Patronas

The Art of Consistency — Episode How to Create Food Photos in the

1 Style of Hiroshi Yoda Using…
LoRA Training and Character Identity Hiroshi Yoda’s food photography is renowned
for its minimalist aesthetics, vibrant colors,…

Mar 10 21 Jan 3

Jason Griffin In Level Up Coding by Thuwarakesh Murallie

Flux + ComfyUI on Apple Silicon I Tried 20+ Open-Source LLMs.

with Hardware Acceleration — … These 5 Perform Remarkably We…
Create AI-generated art on your Mac M1, M2, If you can’t afford a GPU but need to handle
M3, or M4 using ComfyUI with the amazing… inferences privately, this is the solution.

Mar 6 254 3 6d ago 108 6

See more recommendations

Help Status About Careers Press Blog Privacy Rules Terms Text to speech