Mastering Chats New Image Generator Pt1

Perfecting Prompts, Managing Duplicates, and Unlocking Your Creative Potential

Good morning friends, followers, subscribers, Substact, Earth and the Universe!

How are you all doing today?

Well, I’m inspired to give you my exponentially growing expertise on all things #writingwithAI, and now image generation with AI…

So, I will be sharing how I’m using Chat GPT as a writing collaborator, and now as an image generator.

First of all, I’ve been writing posts, blogs, and books for the past decade or so, starting on Facebook (where I was cancelled…thank you Fedbook…); where I built a page yes, called Shift Happens, when it was cancelled; it had some 77777 (or so) followers… and I lost all of Facebook and Twitter at the time,… so that was rough.

After playing around with Wordpress, Medium, and writing 3 books taken from my best Facebook and blog posts, and more called Musings of a Natural Philosopher, all available on Amazon, self-published, self-edited (with help from my partner), and even started translating them into French, but never published yet…

Experimenting with various Image Generators

I started playing with image generators about 2 months ago, just dabbling at first, and then I got the idea for my Eye of the Beholder epic adventure series and ran with it, writing a dozen chapters, and having Chat create image prompts like this one for the cover, about a month ago…

"A breathtaking book cover for a modern adventure-thriller novel. In the foreground, two protagonists—a rugged, sharp-eyed man (Sam Watson) and a fiery, red headed determined woman (Lisa MacNeil)—stand at the edge of an ancient ruin, staring into the unknown. A small futuristic spherical AI drone (David) with a glowing core hovers nearby (3-4 inches in diameter). Behind them, the Sphinx, Machu Picchu, and Stonehenge fade into the mist, symbolizing their global search for ancient artifacts. In Sam’s hand, a gleaming staff crackling with energy (Merlin’s Staff), while Lisa holds a crystalline Holy Grail pulsing with ethereal light. Above them to the corner, stormy clouds and in their midsts and the shadowy figures of the BuilderBear Group loom in the distance, symbolizing danger at every turn. The overall tone is mystical, cinematic, high-stakes, and thought-provoking—a mix of Indiana Jones, The Da Vinci Code, and ancient futuristic secrets. The colors should be deep blues, glowing golds, and rich, shadowy hues to evoke mystery, power, and awakening."

As this was quite the long Prompt, I asked Chat, please shorten…and here was the new Prompt:

A breathtaking book cover for a modern adventure-thriller novel. In the foreground, two protagonists—a rugged, sharp-eyed military man dressed in Kakis holding a staff and a fiery, red headed woman in Cargo Shorts holding the Holy Grail chalice—stand at the edge of the Egyptian Sphinx. A small (3-4i nch in diameter) spherical AI Orb hovers next to them. Background Fog.

It actually created different prompts for different image generators.

Here are the different Image Generators it recommended and to be used for what purpose:

📢 Final Tips:

✔ Choose MidJourney for art-heavy covers with a dramatic feel.
✔ Choose DALL·E 3 for flexibility and high-detail concept art.
✔ Choose Adobe Firefly for a polished, industry-ready book cover.
✔ Consider a final edit in Photoshop/Canva to perfect details!

I won’t be sharing the prompts as it is not relevant for this article, and it is already getting long…

---

So, would you like to see some of the results?

These images were generated with Dream Up (another image generator)…

This last one was the one I used finally…For Chapter 1…

Here are some of Firefly’s Creations, definitely more futuristic:

One of my main observations was that using random image generation tools, it was more difficult keeping consistency with the images, and especially characters.

One of the main advantages of using the same LLM you use for writing and Image Generation was the consistency of the characters, and style.

Another HUGE advantage was that you can take its last image and make small (or larger) modifications straight from that image. This makes for a continuity of images that keep improving. You must remember that for the free version, you can make 1, 2 maybe 3 complex images per day, and then you need to wait 24 hours…

THUS THE IMPORTANCE OF A GOOD PROMPT FROM THE GET GO!

Especially if you are on a tight schedule!

On a personal level, I am on my third (yukk) and final (yeh) rereading of my first novella, and still want to add about 20 images, or one per chapter, including a full body image of all main characters (about a dozen) in the book, so besides 2-3 more iterations for book Cover 1 left, I have some 30 more images to create, and we can figure some of them will need refining…so 50, maybe 20 or more days for this… so, plan well and try to get at least 1-2 images per day if you are looking for 20-30 images for your own work!

My First Foray with Chat GPTs image generator

I wanted to add an image per chapter, as I am sharing them on Substack, at least about half of the first book will be available for free, and the rest behind a small paywall…and I’m up to 4 chapters available now.

So for Chapter 2, the BBG and Architects of Control chapter I used this prompt:

"Ultra-detailed digital illustration of a sinister private boardroom in LockSteed Bartons Headquarters. The room is shrouded in thick cigar smoke and exudes an atmosphere of quiet malice. Dark mahogany walls lined with aged war memorabilia create a foreboding backdrop. At the head of the long, polished table sits John Tomkins, the elusive and calculating leader of the BuilderBear Group, his eyes cold and intent. Across from him, reclining in a worn leather chair, is Michael Bloomfield, the ruthless CEO of LockSteed Bartons and the Pentagon’s favored drone manufacturer. The lighting is eerie and subdued, emphasizing shadows and the subtle gleam of danger, capturing the tense, malevolent energy of the scene."

And shorter one:

" Ultra-detailed digital illustration of a sinister private boardroom with thick cigar smoke and dark mahogany walls adorned with war memorabilia. John Tomkins, cold and calculating, sits at the head of a table while Michael Bloomfield ruthless CEO of LockSteed Bartons reclines in a leather chair under moody, eerie lighting."

The result was quite good from the get go, I was again, impressed!

Chat actually gave me a couple more images offering slight variations, but I decided to keep this one for Chapter 2…

And for Chapter 3 and the remembrance scene from Irak 2003, I used this prompt:

Create an image for my book, I would like an image of Sam in fatigues during the Iraq war 2003, desert background, you can be inspired by this scene, so perhaps an explosion close by

Here is the first image it created:

So, as Sam was a sniper… I added:

Could you update the image with Sam carrying a sniper rifle, and make it landscape not square 9x16

And the final result and the one I used for Chapter 3:

After a pretty good start, I started encountering some issues, with Chapter 4’s image…

Here are a few versions of an image I wanted and here is the initial prompt I used:

A dimly lit urban alley at night, two figures (Sam and Lisa) huddled close together. Hovering between them is a newly awakened, futuristic AI drone (David), sleek and black with subtle glowing lights. The drone’s sensors and lights hint at its emerging consciousness, as faint lines of digital energy connect it to the heroes’ smartphones. In the background, the looming silhouettes of tall buildings blend with a ghostly outline of Machu Picchu—otherworldly ley lines weaving through the scene, merging modern grit with ancient mysticism.

And in order, what it came up with and my corrections:

So for those who have read the first chapters, this was definitely NOT what I was looking for. I would not want to meet whatever that thing is in any alley! No sir E!

So my next correction prompt was:

OK, please model the drone after this one, it is smaller, also please bring the humans much closer into focus, the female is a red head, and make the alley darker, and not as wide, the drone is right next to them, communicating with them. Less light...

Which I then shared this image, which it created previously that was perfect for David’s first form (prior to sentience, and integrating into the Atlantean Orb of Influence)… Yes I go there, make sure you read it from the beginning…

And the next image it produced was:

CHATS RESPONSE: Here's the updated image with the smaller, sleeker drone, a darker and narrower alley, and Lisa’s vibrant red hair. The scene has a more intimate and intense atmosphere, with the AI drone close to the heroes, communicating through digital energy lines. Let me know if you want any more tweaks!

So getting better but still wasn’t quite what I was looking for…

Especially as Sam (on the left) is NOT A ROBOT…

My next prompt was:

OK, closer, the male is human, in his 40s a little scruffy, not a robot... make the size of the image square, and the humans a little closer... The mark on the wall should ressemble Machu Pichu more. They are not communicating with the ley lines, so you can cut these lines out, and focus on the wall machu pichu. thanks

As you can see, I am always polite, and use Chat like a friend!

So, next image was better again:

CHAT’S RESPONSE: Here's the refined image with the changes you requested: the square format, Sam as a scruffy human male in his 40s, and the glowing Machu Picchu symbol etched into the wall. The alley is darker and narrower, with the compact drone hovering right next to them. Let me know if you want any further tweaks!

I had some issues with this one, even though I do use it on occasion for promotional purposes…

One last modification... could you make the humans wearing more traditional city clothes? Make her hair a little longer... no guns... and added this (as it couldn’t generate anymore image at the time…could you make Machu Picchu look like Machu Picchu, not the pyramid but the mountain and village? Can you dress the 2 characters in city dress, casual, no guns, they are beginning their adventure, this is more the look at the end.. maybe add a little bit of trash and graffiti...thanks

And CHAT returned with:

So to keep the article from becoming to wordy, I won’t put CHATs response every time but here was my new prompt:

Could you put them deeper in the alley, like their talking, take off the graffiti on the side... make her a little taller, wearing pants, take the gun off of the guy... thanks

So as you can see, it’s starting to converge to what I was imagining in the first place… but what the hell is that guy doing, just sitting there?? So…

Yeah, please take out that guy sitting there... Maybe have the girl facing us a little more...

And I finally got the one I used for Chapter 4 on Substack…

---

I hope you are enjoying this little rundown of my thought process while creating prompts, and adjusting them as I go.

As the post is already longer than I originally anticipated, I will have to continue in Part 2, so as not to swamp you with too much info, when we can explore the two covers I have had done!

Farewell for now, Part 2 should be coming in the next few days, like Sunday or Monday!

Hope you enjoy your weekend and this post!