One Thing I Love: Learning
I know, it might sound a little strange.
But years ago, I realized that learning is part of my “life’s purpose,” and I’ve been embracing it ever since.
Thankfully, AI came along and made learning not only more enjoyable but also a bit faster.
While experimenting with a few image-generation platforms, I picked up a couple of tips that can help new users appreciate Leonardo AI’s suite of tools.
1. Pay Attention to Aspect Ratios
Each Leonardo AI model specifies the dimensions it was trained on.
For example:
- Dreamweaver 7: 640 x 832.
- Leonardo Diffusion XL: 1024 x 768.
Try to stick as close to these dimensions as possible. If you don’t, you might end up with odd results, like a head stacked on top of another or a random leg thrown in (trust me, I’ve been there!).
That said, don’t hesitate to experiment with different ratios. Sometimes, you can get a better result.
For example:
Switching from 640 x 832 to 832 x 640 might surprise you, even though it’s not the recommended setting. So, go ahead and explore!
2. Lean into Guidance and Prompt Magic
The left-hand section of Leonardo AI is one of your best tools for improving your renders. Let’s break down two key features: Prompt Magic and Guidance, both of which determine how effectively your prompts shape the results.
Prompt Magic
Prompt Magic enhances Leonardo’s ability to interpret your input. Previously, Leonardo relied more on its own discretion, which often produced good results.
With Prompt Magic, you gain more control over the output while still allowing some flexibility.
- V2: Available for Free members.
- V3: A paid upgrade that offers even better precision.
Guidance
Guidance adjusts how strictly Leonardo follows your prompt.
- The default is 7, which balances creativity and adherence.
- For more control, try raising it to 8 or 9.
Note:
Higher levels can sometimes produce anomalies, but they may also bring you closer to your vision. Don’t hesitate to test different settings to find the sweet spot!
Pro Tip:
The default setting is 0.3, but increasing it to 0.75 often delivers sharper and more accurate images. Experiment to find what works best for your needs!
3. Use RAW Mode
RAW Mode is exclusive to Alchemy, but it’s incredibly useful for handling longer and more complex prompts.
When your prompt gets too detailed, Leonardo may stray from your intended result. RAW Mode acts as a stronger version of Guidance, keeping the focus tighter and helping you produce sharper, more precise renders.
4. Use Tiling for Seamless Print-on-Demand Patterns
If you’re into creating graphics for print-on-demand products, the Tiling feature is essential. When enabled, it helps you generate seamless patterns perfect for designs like fabrics, wallpapers, or merchandise.
Want to learn how? Here’s a quick tutorial on using Tiling in Leonardo AI to elevate your designs!
5. Experiment with Perspectives
Changing perspectives can completely transform the mood of your render. Here are two unique options to explore:
- Dutch Angle Perspective: Creates an off-center feel. For example, a bench in the woods might appear lonely, eerie, or slightly unsettling.
- Underwater Perspective: Captures both worlds above and below water, offering a serene and vibrant atmosphere.
Experimenting with different perspectives allows you to better convey the message or emotion you want in your designs.
6. Experiment with Colors
Colors play a huge role in capturing attention and setting the mood for your designs.
Example:
- A beach scene with a sunset palette evokes warmth and calmness, creating a rich, intense vibe.
- The same scene in pastel tones feels softer and more subtle, offering a cooler, gentler look.
The contrast between the richness of sunset hues and the softness of pastels changes how viewers perceive the image. Experimenting with colors can help you evoke the exact emotions you want.
7. Reduce Color Bleeding with Details
Color bleeding is still a problem in Leonardo and it used to drive me nuts until I learned one simple lesson: I had to be specific.
For example, here’s what happened
Prompt: a woman in a purple cape walking in a lush forest
Refining Your Prompt for Better Results
The initial render was stunning, filled with purple tones everywhere. While beautiful, it wasn’t exactly what I envisioned.
So, I made my prompt more specific:
Prompt:
A woman in a purple cape walking through a lush forest with thick brown trees, vibrant green grass, and a patch of white and yellow daisies.
Being more detailed helped guide the output closer to what I had in mind.
Adjusting for Better Color Balance
Now the woman’s purple cape stands out, and the overwhelming purple in the flora is toned down. The result feels more natural and balanced.
If you encounter color bleeding, review your prompt and add stronger, more specific details to guide the output. A well-crafted prompt can make all the difference!
8. The Power of a Single Word
When creating images, my goal is to evoke emotion and set a specific mood.
The fascinating part about Leonardo—or any AI image generator—is how a single word can dramatically change the tone of an image.
For example:
Here’s an image of a prince looking directly at the camera. Changing just one word in the prompt can shift the mood entirely.
Quick Note
I noticed that this prince somewhat resembles the artist Prince. It wasn’t my original intention, but I’m actually happy with how it turned out!
The image has a stately, regal feel, almost like an official portrait.
From Prince to Warlock
With a simple change, the mood shifts from royal to dark and dangerous. This warlock’s intense yellow eyes and the sharp edges of his clothing contrast sharply with the stateliness of the prince. The image now conveys a sense of danger and mystery.
Next, let’s look at the warrior.
9. Remove filler words
I love ChatGPT.
I really do.
Like many others, I use it religiously, but I can tell when prompts are straight from ChatGPT. Those prompts are full narratives about how it should convey the feelings of this and that…
When turning text into images, you don’t need definite articles or half the words ChatGPT gives you, because Leonardo AI, Midjourney, and all of those others don’t really care.
If you’re using AI to generate prompts, ensure that you keep to the essential terms.
As someone who admits to being verbose, I work to make my text-to-image prompts more economical.
Instead of saying, “Generate an illustration of a blah blah blah.”
I use something like, “ultra-detailed hand-drawn illustration, subject, something else,” and so on.
Now, does this mean I am not repetitive?
Nope.
In a longer prompt, Leonardo needs to be reminded to draw details, etc.
I try to keep my prompts to about 40 or 50 words, maximum, and for me, that’s saying something.
Fewer filler words help with creating cleaner prompts.
10. Scale it back sometimes.
This dovetails nicely with the removal of filler words for efficiency.
There have been times when I’ve come up with a complex prompt that goes awry.
It’s like “Jenga.”
One more brick out of place and down it goes.
That’s when I have to scale it down to the bare bones. That means starting again with one piece of the image, and then slowly working my way back until the prompt is stable enough to support another piece.
I also recommend using simpler terms in the process of scaling back.
For example, if I need a horse in action, I tend to use “horse galloping.”
Instead, I say “horses running.” A simpler term that’s just as effective.
11. Experiment with Photo Settings
For me, playing with photo settings was a game changer in creating the images I wanted.
I started by reading up on photography basics to understand focal length, aperture, and ISO settings. When I applied these in my prompts, I saw stunning results, but I ran into some challenges.
Photography is a craft that blends reality with illusion, and I don’t have much experience with it. That’s where I realized I needed some extra help.
So, I turned to my trusty tool to get the scenes I envisioned.
12. Play with Styles (No, not Harry)
Doing the same thing repeatedly is crazy and boring.
When you mix things up, you can get an amazing result and that’s where styles come into play.
You need to figure out how to combine, mix, and layer them into your prompts.
So what are styles?
Think of it like a signature.
You’ve heard Leonardo da Vinci. When you look at his work, it’s got a foggy atmosphere with some realism. That style was typical of the Renaissance period.
Salvador Dali had dreamlike surrealistic images while Andy Warhol made vibrant pop art images of cultural icons cool.
Oh, and graffiti is a style too.
There are plenty of art styles, including graffiti, feminist art, and so on, that you can reference in your prompts.
13. Be Agnostic When Editing
In Leonardo Canvas Editor, you can edit any image, whether it was generated on the platform or not.
- If you’re editing an image created with Leonardo, stick with the same fine-tuned model used for the original image.
- For images not generated on the platform (like photos), I recommend using Stable Diffusion 1.5 or 2.1 models.
Why?
Other models may interpret your prompt in ways that don’t match your image. For example, if you created an image in Absolute Reality, you wouldn’t edit it with a 3D Animation model, right?
Stable Diffusion 1.5 and 2.1 are “agnostic.” They focus on your image and make adjustments that stay true to the original.
14. Raid the Community Feed
14. Explore the Community Feed
The Leonardo AI community feed is a goldmine for learning and inspiration. You can find a wealth of ideas, tips, and examples that can help deepen your understanding and improve your work.
You can explore recent and top prompts from other designers, see how their images were generated, and remix them to fit your own style.
I can’t count how many times I’ve hit a creative block, checked the community feed, and ended up with a much better render. It’s a fantastic way to learn and improve your prompt engineering skills.
15. Join Discord
I’d heard of Discord but never used it until I got into AI generation. Now, I’m part of several communities, and I visit whenever I want to refresh my skills or get inspired.
Beyond the community tab, the Discord server has been incredibly helpful. Many talented digital artists share great work and are eager to help with any questions.
I also love the masterclass threads that have taught me a lot about better prompting.
Now, this is not comprehensive, by any means.
That’s my list—though it’s not exhaustive.
Even after using Leonardo AI so many times, I remind myself that I’m not alone in this. Prompting can feel isolating, especially when dealing with a temperamental bot.
But you’re not alone. With a little effort, small tweaks, or a shift in perspective (literal or figurative), you’ll find yourself getting much closer to the results you’re aiming for.
Annet
If you want to spend time in an interesting and unusual way, visit the website https://undressapp.ai/en. This service helps to create realistic images of girls with the ability to change their appearance and clothes as you wish. The platform is easy to use – even beginners will quickly figure it out. The advantages include high-quality graphics, a variety of styles and complete protection of personal data. An excellent option for creative and exciting leisure.