Welcome to Episode 2 of our mini-series.
EVERY. PROJECT. NEEDS. GRAPHICS!
You may think of a podcast as an audio format, but it still needs cover art. YouTube is video, but it still benefits from a thumbnail image. Every project needs graphics.
Two pain points are often: where do we get good images, and what size image should we use? In this lesson, I’ll give you a quick intro to finding royalty-free images on Pexels, and how to get the right size template for any project on Canva.
I’ll also give a quick intro to AI Art generation with Midjourney, demonstrating image prompting, style reference, and character reference. Enjoy the video below, and subscribe to the mini-series playlist on Youtube to make sure you don’t miss a lesson!
Royalty Free Images from Pexels
There are a handful of places where you can find royalty-free stock photography, such as Unsplash and Pixabay, but Pexels is my favorite.
Some cool features not mentioned in the video are (1) the filters, where you can filter for images based on (2) tall, wide, or square orientation, as well as (3) color.
AI Art Generation with Midjourney
With Midjourney, all you need to do is describe the image you want, and let AI do the rest. You’ll get some random-ish results at first, but you can use an iterative process to refine the results to get something you can actually use.
For example, when I prompt Midjourney with “a buddha statue, 4k wallpaper, retro cyberpunk, dark synthwave, blade runner” I get something like this:
I kind of like the 3rd one, but I want to mix it up and see if I get something that I love, so I’ll request some strong variations on image three by clicking the “vary strong” button on that image.
This will give me four more images similar to the one selected, and I can keep requesting variations or go in a new direction.
You can accelerate the iterative process by typing “–r 3” at the end of your prompt, such as “a buddha statue –r 3” which will give you twelve images to start with, instead of just four.
Image Prompt
Besides prompting the AI with written instructions, you can include an image prompt as well, by uploading an image to Midjourney. The “image prompt” just takes your written instructions and the provided image and just smashes them together to create a new set of images.
For example, if I give the prompt “a Buddha” and also the image prompt Steamboat Willie:
I get something like this:
Style Prompt
Another way of using images in Midjourney is “style prompt.” This causes the AI to try to create what you’ve described, but in the style of the image provided. Here’s “a Buddha” with Steamboat Willie style:
Character Prompt
If you want to take a different direction, you can provide an image for use as a “character prompt.” This will try to keep a consistent character/subject throughout the images, but in my experience it leads to some pretty weird results. Here’s “a Buddha” but with Steamboat Willie as the character (not the style):
Prompt Chains and Combos
To get a more consistent series of images, such as for use in a presentation, you can take your AI generated image, and then use that image as a style prompt for the next AI generated image. You can also use multiple style prompt images at once. For example, here’s an image, generated from a text prompt and two style prompt images (both of which were generated by previous text prompts):
You can see how the previously generated images are influencing the style of the current output for a consistent “feel” throughout the presentation.
Right-size Templates with Canva
I wish I could have back all the time that I’ve spent in my career Googling something like “what size image for [insert social media platform] this year.” If you don’t know what “safe zones” are for social media posts, I envy you.
Fortunately, you don’t have to know how many pixels tall or wide an image needs to be for any particular use, because chances are, Canva has you covered. There are lots of awesome tools on Canva’s pro tier, but even just being able to select the proper size layout in the free tier is worthwhile.
Beyond that, using Canva is all about drag-and-drop, playing with gradients, transparencies, and text effects.
The only way to learn Canva is to just jump in, have fun, experiment, and find your own creative style.
Tie It All Together
Royalty-free stock photos from Pexels, AI generated art from Midjourney, and the tools and templates in Canva, can all add up to a fun and efficient combination for making graphic assets for any project.