Grok Imagine just Upped the Mobile Image Generation Game
I’ve been using Artificial Intelligence for image generation now for over a year. In the early days, it was very much hit and miss. It would imagine things that weren’t in the prompt. It would add extra fingers, it would forget fingers on the hand. When you zoomed in on the image and looked at the faces, they looked like monstrosities that were more alien than human. It would sometimes take me an hour or more just to get one satisfactory and usable image. It was frustrating to say the least.
There are still plenty of AI image generators out there that continue to have the same problems. Over time, OpenAI’s ChatGPT has gotten better as have other leading AI companies. The one I’ve been using most lately though is Grok AI by xAI. I have found it to be an excellent AI image generator for mobile or portrait images. It isn’t perfect, and I’ve yet to find an AI image generator that is, but it is very good for mobile use.
If you want to create more traditional image formats like 16:9 for presentations, you would ask Grok to generate the image in their general chat window. Imagine is for mobile use and for quick iteration of images since it features on infinite scroll to help you find the right image for your needs. Keep in mind that there are limits to the number of images generated by Grok Imagine based on your subscription level.
The Prompt Still Matters
I would encourage you to learn how to craft a good prompt. While Grok can do wonders with little in your prompt, the more precise you can be, the better the result. Things to consider when crafting a prompt include:
The Subject: What is the main subject of the image? If there is a person involved or more than one, be as precise as you can about them. Rather than type, “a king dreaming while on a bed”, type “a Babylonian king wearing ancient bed clothes, troubled while dreaming.”
The Context: Where is the subject? What is taking place around them? What does the background look like? Is it in focus or out of focus? Specificity here is also helpful for the AI.
The Camera: If you imagine being the one standing there taking the picture with the camera, that’s what it means to consider the camera in the shot. Where is the camera positioned? What kind of camera is it? You can even specify the f stops to use with the camera. Specificity here too can yield nice results.
Artistic Style: Here is where you can have some fun. Is it going to be a photo? A cinematic photo? Do you want it to be a drawing? A painting? An image in anime style? Every good image prompt is going to let the AI know what kind of artistic style you are going for. My experience with Grok is that it can do all kinds of styles well.
Simple Video Creation at a Click
Any image you generate with Grok Imagine can be briefly animated with a simple click on the play icon of the generated image on the bottom right hand side. At this time it seems to me to be more of a parlor trick but if you are looking for a short, 5 second video clip to include in your TikTok or YouTube Short feed, Grok Imagine can help. It is also possible to upload your own image and have Grok animate that image.
I imagine we will eventually see Grok, ChatGPT, and other LLMs get better at general video production. That day is not today. Services I’ve used like Runway that have more advanced video generation prompting are expensive to use and hit and miss. I’ve heard a lot of praise for Google Veo, currently Veo 3 at the time of writing, but haven’t had a chance to really put it through its paces. My access to Google Veo has been very limited due to the expensive cost of generating images. I imagine that cost will come down over time.
Copyright Concerns
Copyright law concerning AI generated images is still in its infancy. There is therefore some risk when using AI images in paid products or in live streams. The current law states that AI images cannot be copyrighted if they are used straight from the LLM without modification. Modifications can’t just be a minor tweak. It has to involve significant human interaction. Some suggest that a series of prompts modifying the image might qualify, for what you will need to keep a copy of the prompts you used to generate the image.
I would also suggest you take the time to do a Google Reverse Image search, use other services like Bing reverse image search, or Tineye.com. These services check your images against a database of billions of other images to see if there is a match. Some of these services can also show you duplicate or similar images. You don’t want to get caught using someone’s copyrighted work, especially if you are making money off of the image through the sale of a product like a book or magazine.
Image Generation is Only Getting Started! AI Image Generation is here to stay. Over the last 2-3 years it has been making rapid progress. It has now progressed to the point where decent images can be generated for use in sermons, books, and presentations. I expect that image generation will continue to improve going forward. Experts disagree on how fast AI will progress but there is no question that AI is here to stay.
Final Caution
As Christian leaders, we must guard against idolatry and falling into the trap of iconaltry. According to Wikipedia, an Icon “has now come to be closely associated with religious art.” Eastern Orthodox and Catholic churches use images as an aid in worship. In my opinion, we need to tread very lightly when generating pictures of Jesus. I believe I can state with one hundred percent accuracy that the popular recognition of Jesus as depicted in images, paintings, and sculptures is false. We have absolutely no idea what Jesus looked like other than, according to Isaiah 53, he would have been an average looking, plain Jewish man of his time.
When we generate images for our sermons, presentations, and books, let us regularly disavow the use of images to get people to move people emotionally to make decisions for God. This is especially true of displaying an image of Jesus Christ while making an appeal up front for people to give their life to God. If we are not careful, we can inadvertently distort the image of God in people’s minds. AI generation of images makes this danger even more real than previous times. If you ask it to generate images of Jesus, it will almost always spit out typical interpretations of what Jesus is considered to look like in popular Christian culture.
Finally, don’t be afraid to experiment with different LLMs and image generation services. Also, don’t settle for the first thing an AI generates if it is not suitable. Keep working on different ways of wording your prompts and maybe, you’ll stumble on the image you are looking for.
Joseph Duchesne is the creator of The Church AI Guy, a space where faith meets innovation while discussing the long-term impact of AI. A pastor, autodidact, and author of two books—The Last Crisis and Discover the One—he’s passionate about showing how Jesus-centered discipleship can thrive in a digital world. When he’s not experimenting with the latest tech, he’s reading theology, building church community, or spending time with his wife.