Prompt:
cinematic film still (Ethereal Wonderland:1.4), (Mystical Atmosphere:1.3),(Fantasy Realism:1.2),
(space_style:0.6), cosmic outer space, Unsplash contest winner, 16k resolution, 300 dpi, 600 dpi,
4k, Contest winner, High definition, detailed, realistic, 8k uhd, high quality,, daylight,
(masterpiece,best quality:1.5), ultra realistic,32k,RAW photo,(high detailed skin:1.2), 8k uhd, dslr,
soft lighting, high quality, film grain, shallow depth of field, vignette, highly detailed, high budget,
bokeh, cinemascope, moody, epic, gorgeous, film grain, grainy, ((ChalkDustStyleSDXL)),
((ahanhatbinh)), ((locon_flmngprsn_v1_from_v1_64_32))
Negative Prompt:
anime, cartoon, graphic, text, painting, crayon, graphite, abstract, glitch, deformed,
mutated, ugly, disfigured, dark, blue skin, cartoon, semi-realistic, blurry, logo, watermark,
signature, cropped, out of frame, worst quality, low quality, jpeg artifacts, poorly lit, overexposed,
underexposed, glitch, error, out of focus, (semi-realistic, cgi, 3d, render, sketch, cartoon, drawing,
digital art, anime, manga:1.3), amateur, (poorly drawn hands, poorly drawn face:1.2), deformed
iris, deformed pupils, morbid, duplicate, mutilated, extra fingers, mutated hands, poorly drawn
eyes, mutation, deformed, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face,
disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs,
fused fingers, too many fingers, long neck, incoherent, naked, nsfw, (worst quality, low
quality,nsfw,nipple, pussy:1.5)
Model:
dreamshaper_8
Prompt
A good prompt should include: Subject (mandatory), medium, style,
artist, website, resolution, additional details, color, lighting.
You don't need to include keywords from all categories, use this list as a
reminder to know what content to use.
I will use the basic DreamShaper_8 model, all images generated with
DPM++ SDE Karras sample in 20 steps, with image size 512*768. To focus
on keyword effects, I won't use negative keywords for now. But don't
worry, we'll also discuss negative keywords later.
Subject
Subject is the content you want to see in the picture. A common
mistake often made is not writing enough about the subject. For
example, if we want to generate an image of a female witch using her
magic, a beginner might write:: A sorceress
For character creation, a common trick often used is to mention the
names of celebrities. This has a strong effect and is a very good way to
control the appearance of the subject. However, it should be noted that
these names may not only change the face but also the pose and other
aspects.
As a demonstration, I will depict the female witch as Emma Watson,
one of the most frequently used keywords in "Stable Diffusion." Imagine
her as strong and mysterious, casting lightning magic. We want her outfit
to be very detailed to look appealing.:
Emma Watson as a powerful mysterious sorceress, casting lightning
magic, detailed clothing
Media
Media refers to the materials used to create artworks. For example,
illustration, oil painting, 3D rendering, and photography. Media has a
very strong effect, as with just one keyword it can significantly change
the style. Let's add the keyword.: digital painting.
Emma Watson as a powerful mysterious sorceress, casting lightning
magic, detailed clothing, digital painting
Style
Style refers to the artistic style of the image. Some examples include
impressionism, surrealism, pop art, etc. I've added it to the keywords:
hyperrealistic, fantasy, Surrealist, full body.
Artist
Artist names are powerful modifications. They allow you to
customize Chocobo's style by referring to a specific artist. Using multiple
artist names to combine their styles is also common. Now, let's add
superhero comic artist Stanley Artgerm Lau and 19th century portrait
painter Alphonse Mucha.
Emma Watson as a powerful mysterious sorceress, casting lightning
magic, detailed clothing, digital painting, hyperrealistic, fantasy,
Surrealist, full body, by Stanley Artgerm Lau and Alphonse Mucha.
Dapat dilihat bahwa gaya dari kedua artis tersebut telah berhasil
digabungkan dan menghasilkan efek yang konsisten.
Website
Graphic websites like Artstation and Devian Art have tons of images
from various genres. Using them in the instructions is a reliable way to
direct images to these styles. Let me add Artstation to the instructions:
Artstation.
Perubahan ini mungkin tidak begitu mencolok, tetapi gambar
memang tampak memiliki gaya yang mirip dengan situs web Artstation.
Resolution
Resolution represents the clarity and detail of the image. Let's add
the keywords : highly detailed, sharp focus.
Efek mungkin tidak begitu besar, karena gambar sebelumnya sudah
sangat jelas dan detail. Namun, menambahkannya tidak ada salahnya.
Detail
Other details are additions to modify the image. I will add elements
of science fiction, beauty, and anti-utopia to give some atmosphere to
the image : sci-fi, stunningly beautiful, dystopian.
Color
You can control the overall color of the image by adding color
keywords. The color you specify will probably appear as a base color or
in objects. I will use the keyword 'iridescent gold' to add a bit of gold to
the image: iridescent gold
Munculnya baju perang emas, tampak bagus.
Lighting
Any photographer will tell you that lighting is the key to creating
successful images. The keyword lighting will have a huge impact on how
the image looks. Let me add in a prompt: cinematic lighting, dark.
So the prompt is:
Emma Watson as a powerful mysterious sorceress, casting lightning
magic, detailed clothing, digital painting, hyperrealistic, fantasy,
Surrealist, full body, by Stanley Artgerm Lau and Alphonse Mucha.
Artstation, highly detailed, sharp focus. sci-fi, stunningly beautiful,
dystopian. iridescent gold, cinematic lighting, dark.
Negative prompt
Common negative prompt is : ugly, tiling, poorly drawn hands,
poorly drawn feet, poorly drawn face, out of frame, extra limbs,
disfigured, deformed, body out of frame, bad anatomy, watermark,
signature, cut off, low contrast, underexposed, overexposed, bad art,
beginner, amateur, distorted face, blurry, draft, grainy.
Negative prompt help to emphasize the image and avoid local
distortions in the image.
Other common negative prompt :
ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face,
out of frame, extra limbs, disfigured, deformed, body out of frame, bad
anatomy, watermark, signature, cut off, low contrast, underexposed,
overexposed, bad art, beginner, amateur, distorted face
Weight Prompt
You can adjust the weight of keywords through the syntax (Keywords:
factor), where factor is a value with a default of 1. A value less than 1
indicates less importance, while more than 1 indicates more importance.
For example, we can adjust the weight of the keyword lightning magic:
(casting lightning magic: 1.5)。
Emma Watson as a powerful mysterious sorceress, (casting
lightning magic: 1.5), detailed clothing, digital painting, hyperrealistic,
fantasy, Surrealist, full body, by Stanley Artgerm Lau and Alphonse
Mucha. Artstation, highly detailed, sharp focus. sci-fi, stunningly
beautiful, dystopian. iridescent gold, cinematic lighting, dark.
Adding weight to lightning magic will usually result in more lightning,
while subtracting it will usually result in less. This may not always apply
to every image, but statistically, it is true.
This technique can be applied to theme keywords and all categories.
Such as style and lighting. Sintaks () dan []
An alternative method to adjust the keyword strength is to use () and [].
(Keyword) will increase the keyword strength by 1.1 times, equal to
(Keyword: 1.1). [Keyword] will decrease the strength by 0.9 times, equal
to (Keyword: 0.9). You can combine several together, similar to
multiplication in math.
(Keyword): 1.1，((Keyword)): 1.21，(((Keyword))): 1.33；
[Keyword]: 0.9，[[Keyword]]: 0.81，[[Keyword]]: 0.73。
Pencampuran Kata Kunci
You can combine two keywords. The correct term is "keyword
scheduling". Syntax: [Keyword1: Keyword2: factor]. For example, if we
use keywords:
[Emma Watson: Angelina Jolie: 0.5] ， at 20 sampling steps, steps 1-10
have Emma Watson clues, while steps 1-10 have Angelina Jolie clues.
This factor determines when the keyword changes: 20 * 0.5 = 10.
The effect of changing factors is that the two female actresses are
mixed to varying degrees. This is a perfect example of an important rule
in keyword blending: the first keyword determines the global
composition. The initial diffusion step determines the overall
composition, while the subsequent steps refine the details.
What if I swap Emma Watson's position with Angelina Jolie's:
[Angelina Jolie: Emma Watson: 0.5], what would I get?
Dari Kata Kunci ke Kata Kunci
By combining keywords, you can achieve an effect similar to the
transition from keyword to keyword, by editing a series of very similar
images. We can replace the lightning in the image with fire and give it a
different factor : [lightning: fire: 0.8], [lightning: fire: 0.2] ， with the
same seed and steps.
Emma Watson as a powerful mysterious sorceress, casting
[lightning: fire: 0.8] magic, detailed clothing, digital painting,
hyperrealistic, fantasy, Surrealist, full body, by Stanley Artgerm Lau and
Alphonse Mucha. Artstation, highly detailed, sharp focus. sci-fi,
stunningly beautiful, dystopian. iridescent gold, cinematic lighting,
dark.
This 'factor' often needs to be carefully adjusted. How does it work? The
theory behind it is that the overall composition of the image is
determined by the initial calculation step. Once the diffusion is trapped
in a small space, replacing any keywords will not have a big effect on the
overall image, only changing a small part of the image.
How Many Prompt We Can Use?
If you are using the SD V1 model, you will face the limitation of
keyword length, which is limited to 75 Tokens. Tokens are different from
words.
While in AUTOMATTIC1111, there is no Token limitation. If the
keyword contains more than 75 Tokens, it will start a storage block of 75
additional Tokens, bringing the new limitation to 150. This process can
continue endlessly or until your computer memory runs out.
In AUTOMATIC1111, you can check the number of tokens by looking
at the small box on the top right corner of the keyword input field.