How To Become a Master Prompter With AI in 2024

Hey guys! Schieler here….
I wanted to deep dive, not delve, into how and why we get such good image prompts at Stock MJ. A lot of people ask this question and have a desire to create the same quality of images we do and can, and I don’t blame them!
So what’s it really take and how are we getting these images? Well, it’s not easy, and it takes a ton of time….here’s why!
There is nothing straightforward about image prompting. What I mean by this is effectively the image prompt changes every time depending on the specific circumstance and can take up to 50 runs before we get the image exactly the way we want!
But what exactly does that mean? Well let’s follow the breadcrumbs….
Expanded Overview of Image Prompting
Complexity of Image Prompting:
Each image prompt is a unique challenge influenced by a myriad of factors. For instance, the context of the image plays a crucial role. An indoor setting might require different lighting and shadow details compared to an outdoor setting. Additionally, the time of day, artificial lighting, and even the weather can impact the desired outcome of the image. Understanding and incorporating these elements into the prompt can significantly change the AI’s interpretation and output.
Experimentation and Refinement Process:
Achieving the perfect image often involves a process of trial and error. This experimentation isn’t just about tweaking the details but also about understanding how different elements interact within an image. For example, changing the angle of light can alter the way shadows fall and can dramatically affect the mood and realism of the image. It’s about finding that sweet spot where all elements of the image cohesively work together to create a realistic and compelling visual.
Let’s take eyes for example. Mid Journey can create a wide range of quality in eyes, but getting them to look realistic can be quite the challenge depending on the scenario of the image, how the light is hitting the eye, if the image is outside or inside and other properties of physics. Crazy right?
One of my most famous eye prompts came down to a series of testing, and multiple prompts starting off very simply and becoming more and more complex. Here’s how it started and ended.
- Closeup of eye –ar 16:9.
- Macro photography of eye –ar 16:9.
- Macro photography of eye zoomed into iris –ar 16:9.
- Macro photography of eye zoomed into iris with consistent melanin –ar 16:9.
- Macro photography of eye zoomed into iris with consistent melanin and circular contraction folds –ar 16:9.
…and just went from there into a series of very specific anatomy-based prompts that yielded the following results:
As you can see, the problem arises in the fact that this prompt cannot be used for other subjects, as it is specific to the biology of the eye itself. In fact, you could continue taking this prompt to new levels by specifying light physics to adjust the reflection and glare in the eyes, as well as providing specific information on how the whites of the eyes should look and the eyelashes.
Because of the above, you can see advanced prompting comes down to a few different systems and processes, and it’s not an easy task that may require tons of research on your end.
My 3 part prompting system and process is as follows
Start as Simply as Possible:
Understand what the typical result of your initial prompt is. This gives you a baseline to work from and helps identify what needs to be refined.
Structuring Your Prompt:
Begin to include more specific information relevant to the picture you are generating. This may include details like the quality of light, physics, color, scientific names, biology, anatomy, and even the description of angles.
Balance Specificity with Generality:
Be specific enough for the software to understand your requirements but general enough that it does not start to over-compensate with inferences. This balance is tricky but crucial. It’s about guiding the AI without restricting its creative capabilities.
List of factors that influence your prompt
Lighting Conditions:
The direction, intensity, and color of light sources can dramatically change the appearance of the subject and setting.
Physical Setting:
Indoor or outdoor environments, urban or rural landscapes, and specific locations like forests, beaches, or city streets.
Time of Day:
Different times (dawn, midday, dusk, night) can affect the mood and lighting of the image.
Weather Conditions:
Elements like rain, snow, fog, or clear skies can impact the atmosphere and lighting.
Perspective and Angles:
The angle from which the subject is viewed, including aerial, eye-level, or worm’s-eye views.
Color Schemes:
The overall color palette, which can set the tone and mood of the image.
Textures and Materials:
The surface qualities of objects in the image, such as rough, smooth, reflective, or matte.
Cultural and Historical Context:
Elements that indicate a specific era or cultural setting, like clothing, architecture, or technology.
Emotional Tone:
Conveying feelings such as happiness, sadness, mystery, or tension.
Character Attributes:
For images with people or creatures, details like age, gender, expressions, and posture.
Scientific and Biological Details:
Accurate representation of natural phenomena, anatomy, or ecological systems.
Artistic Style:
The desired artistic approach, such as realistic, impressionistic, surreal, or abstract.
Composition Elements:
The arrangement of objects and elements within the frame, including the rule of thirds, symmetry, or asymmetry.
Depth and Focus:
Depth of field, focus points, and how these elements guide the viewer’s attention.
Movement and Dynamism:
Suggesting motion or stillness in the scene, which can affect the image’s perceived energy.
Interactions and Relationships:
How different elements or characters in the image interact or relate to each other.
Scale and Proportion:
The size of objects or characters in relation to each other and their environment.
Symbolic Elements:
Incorporation of symbols or metaphors to convey deeper meanings.
Technological Aspects:
For futuristic or sci-fi themed images, details of technology, machinery, and innovation.
Fantastical Elements:
For fantasy-themed images, elements like magic, mythical creatures, or unreal landscapes.
Photoshopping results when we have to
When our prompts are almost perfect but not quite there we result to Adobe Photoshop to help us. There are many case uses in which we have used Photoshop on the pictures here at Stock MJ and here are a few reasons why.
- We often pull images into Photoshop to lens-correct the subject on images where the picture focal length has made the subjects face too wide or narrow but everything else in the photo is above par.
- We also pull images into Photoshop to correct the saturation, hue, contrast and other image properties to help make images more photo-realistic and natural.
- Lastly, at times we pull images into Photoshop when there is something in the frame that AI has added for unknown reasons if the picture is otherwise perfect.
Conclusion about prompting
All in all, it’s not hard to type words into Midjourney and get results – we all know this. The hard part is taking the time out to research your subject, understand the nuances of how your prompt changes when variables are introduced and then refining them to get the best output possible, while then possibly introducing Photoshop to refine the results even more.
All in all, I would love to provide everyone who follows Stock MJ with my prompts, the issue is that they change from picture to picture, scenario to scenario, subject to subject and they are continually evolving as new versions of Mdjourney are released.
That is why we have done the work for you…but it is a lot of work!
People also ask about Midjourney
Are there alternatives to Midjourney?
Yes, there are alternatives to Midjourney that also provide great image results through natural language prompts provided to an AI art generator. Some of those include:
- DALL-E 3
- Stable Diffusion XL and SDXL Turbo
- Adobe Firefly
- Bing Image Creator
- Leonardo.Ai
- Blue Willow by LimeWire
- Fotor AI Image Generator
- NightCafe
Comments are closed.