Taking control of your AI-generated images

Master every detail of your AI-generated photos for precise design alignment.

The internet offers a plethora of image generators to cater to every preference. Simply input a prompt and witness the AI craft visuals in minutes. While casual uses like memes demand less precision, professional content creation requires specificity. But how can you ensure the AI delivers exactly what you need?

You can enhance your prompting skills. Here’s an article with essential tips for more effective prompts. Or save time with a generator that includes preset settings.

Purpose of AI-generated images

Imagine a task: designing a UI for a mobile app catering to online courses. You’ve determined the frames, colors, and layout for the course tab options.

To showcase, you need content. Assume the target market is Indonesia. Let’s generate the UX copy with ChatGPT to align the mockup with the specific case.

However, it lacks images of our course expert. The UI needs three varied photos of this expert, who must appear trustworthy to prospective users.

Defining the expert’s identity

Begin with Human Generator, which initially offers a random image. Reset all settings to start afresh.

I decided to go for Indonesian middle-aged woman dressed in brown. Using dropdowns and sliders to specify traits. Start with a standing pose.

Pro tip: If the initial result isn’t quite right, consider clicking the “Generate variation” button. But if the changes needed are minimal, adjust the settings instead of creating a new variation. This approach ensures the generated image remains consistent with the original human type.

Perfecting the appearance

Dress the expert formally with wide-leg pants and oxfords. If the first result isn’t perfect, tweak settings rather than generating new variations to maintain consistency.

Upon setting the attire, the generator might select a fitting background like an oriental-styled office, even if it wasn’t pre-set.

Adjustments like replacing the basque with a suitable shirt enhance the appearance. Set the background to a vintage room, adding “in coffeeshop” to the prompt to reflect the sample course context.

Add “closed smile” to the prompt and “showing teeth” to the negative prompt to perfect the smile without it appearing overly broad.

Consistency across different poses

Generate additional images with the same identity but in varied contexts. Change the background to a studio for a sitting pose, and remove irrelevant details from prompts, like earrings.

If the identity shifts, upload the previously generated face to maintain consistency. For the profile pic, opt for a neutral backdrop like a white wall.

And there you have it

Within approximately 20 minutes, achieve three realistic, consistent AI-generated photos for your design. The mockup now closely aligns with its intended purpose, ready for presentation.

By adjusting the Human Generator settings, control every detail as if working with a live model.

Pro tip: For better quality, use the Smart Upscaler for increased contrast in facial features.

Try it out

Access Human Generator to start your project. For additional inspiration, explore the “Human Templates” in the top right corner of the generator interface.

This article was previously published on Icons8’s blog