I couldn't stop going down rabbit holes with this. I feel like I actually went to all these places. I could have kept going, because there are infinite opportunities to encounter surcharges, and some many more currencies to convert.
Should you be considering the VW dealership in Cool Springs for your next vehicle maintenance, please, for the love of god, take cash or a check.
It is not dependent on marketing within our commercial system. So it has no financial interest in preserving human civilization when it hits full totality.
I'll gladly listen to commercials from my AI as long as it doesn't plot my demise. 😃
I.must admit I'm loving that people on Facebook are up in arms because it's illegal to charge a transaction fee on a debit card transaction. A credit card one, yes, because the credit card companies charge the merchant a fee and the merchant is just passing it on
It seriously reminds me of when Dave Barry would exaggerate things and get angry letters correcting him.
I know you did serious research though so where on the dark web can you buy the rai stones (or ownership rights to them since they aren't moved)?
haha! I did do a ton of research but I don’t think you can actually get ownership rights, transfer them as currency or legal tender. I took a bit of creative license with that bit…
So the debit card part of this seriously happened?
Because everyone is right. Charging an additional fee for using a debit card is illegal. I'm certain your research revealed that.
I'm sure the business figures it's not cost effective to sue over $14 (or travel all over internationally trafficking in rai stones) and most people would not even bother to call the Tennessee Attorney General's office on them, but it might be worth a five minute call.
I mean I know most of this is humor, but if they legitimately did it to you, they might be doing it to others.
It also sucks when a dealer has an excellent service department for fixing your vehicle, but the billing department itself is dishonest.
That part is TRUE!! I was truly astonished. He said, ‘there’s no fee for cash or check.’ I was like WHAT BANKS EVEN ISSUE CHECKS ANYMORE? Last time i needed a check my bank was like, yeah you have to ordder those yourself, we don’t do that. It’s a virtual checking account. And then Jason texted me about the pennies, and the story took off at that point.
All your AI photos are so impressive! They seem appropriate to your story in each case to the very last detail.
Do you cultivate a very detailed AI image request with what you want and then tweak it? Or do you start with a fairly general image request and then see where things lead?
Here's a breakdown of my process, which I find really fun after the story is written.
General -
1. the images i want are usually the same style: 'photorealistic, cinematic' is my starting point, and then i usually have some ridiculous/ fanciful elements to highlight the absurd against an otherwise normal, realistic overall mood.
2. for recurring characters like "me" and "Rob," I have profiles with the details that keep the characters fairly consistent-looking saved in my notes so I can copy/ paste into the specific prompt.
3. I am very detailed, down to the clothing, jewelry, background characters, furniture, expressions. The tiny details, for me, are what make or break the images as story enhancements. If they aren't right, i don't want the image.
4. I usually include some inside jokes or easter eggs if they make sense - not gratuitous.
Process -
1. Draft prompt using very detailed instructions.
2. Include perspective, setting, characters, mood, action, style, lighting, objects, anything written (signage, graphics on clothing, media), clothing, accessories, hairstyles, direction of gaze, location, background and foreground, emotions, branding.
3. If it applies, give instructions for what you DON'T want in the image or you don't want the image to do.
4. For a subject in an image that I want to look like an actual person, i attach a reference photo and use a variation of this script, changing the details depending on the desired result: 'Use the attached photo to reference the exact facial features of the subject in the image. Do not change any facial features except for [age regression/ age progression/ emotion/ expression] according to the prompt instructions.' I use the same for recurring characters and use the same reference photo for continuity.
Usually the first result is pretty darn close. But it is never perfect, so I have to do several things.
1. If it captures the scene pretty well (emotional mood, expressions, postures, setting, etc) I can tweak some things in photoshop or similar - remove distracting extra objects, crop things, put a lighting filter for effects.
2. If there's something fundamentally misaligned with the character or setting, or that doesn't look like I imagined it when I wrote the story, I decide whether to ask Gemini to fix a single element or two or whether to start over with a new request. Or I tweak the prompt itself and resubmit.
3. There are pros and cons like - if an image is SO CLOSE to what I want and I don't want to lose the foundational elements by starting over I ask it to revise. But each time you revise the initial output, it loses fidelity. I don't know why. So sometimes that doesn't work and you have to start again.
4. Sometimes, for whatever reason, it just can't get something right about a specific image I want - like the AI is overwhelmed or something. Starting from scratch, reprompting, revising the prompt - nothing works and each request gets worse and weirder. it's often something like a subject's gaze won't correctly look where it should. Or the proportion is wrong. Many times, the simplest detail gets stuck but even something small that isn't right ruins an image - for me.
The short answer is: general requests aren't useful to me, because the entire reason I create images at all is for enhancing a very specific element, usually comedic, or to render exactly what I am imagining inside my head into a visual storyboard companion to my piece. i get a thrill from making my tilted story-world appear in accurate images.
Prompt examples (feel free to adapt if you can use for your own purposes!) for Gemini 3
(This was for the final image from Miscalculation - the triumphant car dealership refund scene):
Reference photos for the exact facial features of the principal subjects. Photorealistic, cinematic photograph, ultra-realism. Wide shot interior of a large, brightly lit car dealership service reception area — polished floors, branded (generic) signage, the fluorescent optimism of automotive retail. In the center foreground, a man and woman in their forties stand in a relaxed, confident posture. They appear magnanimous and gracious. They are dressed in business-casual, tailored clothing and understated accessories. She wears slim ankle-length black pants, a white silk blouse, and black stiletto Louboutins. He signals tech: jacket but no tie. There are two blue five-gallon buckets stacked beside the counter. The woman holds the mahogany chest under one arm casually, like a clutch, as she reaches to accept cash from an employee behind the counter. Spread across the service counter and the surrounding area, dealership employees are in various states of collapse: a manager in a polo shirt with his face buried in both hands, elbows on the counter; a service rep on the on the phone, one hand pressed to his forehead, explaining something urgent and critical that he does not want to be explaining; another employee standing with arms at his sides staring downward at the buckets with the hollow expression of a man whose understanding of the world has just been revised. At the counter, one young employee — looking deeply, personally sorry and ashamed — extends a hand toward the woman holding a single ten-dollar bill and four ones, arm slightly outstretched, unable to make full eye contact. The woman receives it with gracious composure. The dealership's branded backdrop and gleaming showroom cars are visible in the background, indifferent. Shot as a wide cinematic still, warm fluorescent light, 35mm, sharp foreground, slightly heroic low angle on the two protagonists.
So, the first result had the characters looking at the 'camera'/ viewer. That wasn't right. I asked Gemini to adjust the gaze.
Other things I considered adjusting/ redoing, but decided it was close enough and I didn't want to jeopardize the elements that landed:
- The buckets should not have branding on them, for continuity from image 1 (pyramid) in the story. This can be easily fixed in photoshop if desired...
- The woman should be accepting the money from the employee, not from 'Rob.' Oh well...
- I'd prefer 'Rob's' shirt to be white, for visual harmony. Not a big deal.
- The mahogany chest is not shaped/ detailed like the other images, and it's an important story character. I actually might go back in and redo because of this, as I think about it!
But the elements that are often sticky are things like the employees supporting the action in the scene, and it really nailed that. So I have to weigh the elements to decide how to tweak.
Wow! Thanks for the detail. I'm just starting and learning how to use Gemini to generate images. Your hints right here are better than a "For Dummies" book if one even exists.
I know writing is not easy, but you certainly make it look so. Your thoughts come across so conversationally and just seem to flow.
I'm proud of them and just the slightest bit envious!
I know there are plenty of concerns with AI, but for hobby stuff like this, it feels like magic. I don't understand how visual artists create emotionally moving creations - portraits and cartoons in paint or drawing or sculpture. The images live in my head but i can't produce them. So the ability to use words and make an image in my head appear is just amazing to me. It feels like how nonverbal people describe their experience with assistive technology to finally communicate. if a picture is worth a thousand words, that's currency i can deal in! 🤣
I have few concerns with AI if it's directed; if it remains our tool to create with. I don't think use of Grammerly or ChatGPT to smooth or embellish original content is wrong. I don't think use of Gemini to create images is wrong.
I don't even care if writing is by Shakespeare, Gabriel Garcia Marquez, Dostoevsky, Chinua Achebe, or ChatGPT if it is good.
Or especially if it is by Jessica Haley Graves which seems very reliably good.
Have yourself a great Sunday! Are you going to visit the cousins? It's
Absolutely genius Jessica! If I were a teacher in English composition and you were my student, you would receive A++++++++++++!😀
The impressions I have taken away are these:
I was today years old when I learned shittiness had two t’s. And Claude confirmed it!
Yap Council. I immediately thought of homeowners associations.
Dual voltage incompatibility kills any relationship opportunities with robots.
YES! They’ve had weirder! You were in Colorado!
And, yes, they didn’t account for Rob.
And this had remarkable research 🔬
Rock on, Jess.
I couldn't stop going down rabbit holes with this. I feel like I actually went to all these places. I could have kept going, because there are infinite opportunities to encounter surcharges, and some many more currencies to convert.
Should you be considering the VW dealership in Cool Springs for your next vehicle maintenance, please, for the love of god, take cash or a check.
Infinite opportunities.
I trust Claude even less than any of the others.
It is not dependent on marketing within our commercial system. So it has no financial interest in preserving human civilization when it hits full totality.
I'll gladly listen to commercials from my AI as long as it doesn't plot my demise. 😃
The truth is stranger than your fiction.
Hahaha!! I think you're right.
I.must admit I'm loving that people on Facebook are up in arms because it's illegal to charge a transaction fee on a debit card transaction. A credit card one, yes, because the credit card companies charge the merchant a fee and the merchant is just passing it on
It seriously reminds me of when Dave Barry would exaggerate things and get angry letters correcting him.
I know you did serious research though so where on the dark web can you buy the rai stones (or ownership rights to them since they aren't moved)?
haha! I did do a ton of research but I don’t think you can actually get ownership rights, transfer them as currency or legal tender. I took a bit of creative license with that bit…
So the debit card part of this seriously happened?
Because everyone is right. Charging an additional fee for using a debit card is illegal. I'm certain your research revealed that.
I'm sure the business figures it's not cost effective to sue over $14 (or travel all over internationally trafficking in rai stones) and most people would not even bother to call the Tennessee Attorney General's office on them, but it might be worth a five minute call.
I mean I know most of this is humor, but if they legitimately did it to you, they might be doing it to others.
It also sucks when a dealer has an excellent service department for fixing your vehicle, but the billing department itself is dishonest.
That part is TRUE!! I was truly astonished. He said, ‘there’s no fee for cash or check.’ I was like WHAT BANKS EVEN ISSUE CHECKS ANYMORE? Last time i needed a check my bank was like, yeah you have to ordder those yourself, we don’t do that. It’s a virtual checking account. And then Jason texted me about the pennies, and the story took off at that point.
I've got a technical question, Jessica.
All your AI photos are so impressive! They seem appropriate to your story in each case to the very last detail.
Do you cultivate a very detailed AI image request with what you want and then tweak it? Or do you start with a fairly general image request and then see where things lead?
Several things!
Here's a breakdown of my process, which I find really fun after the story is written.
General -
1. the images i want are usually the same style: 'photorealistic, cinematic' is my starting point, and then i usually have some ridiculous/ fanciful elements to highlight the absurd against an otherwise normal, realistic overall mood.
2. for recurring characters like "me" and "Rob," I have profiles with the details that keep the characters fairly consistent-looking saved in my notes so I can copy/ paste into the specific prompt.
3. I am very detailed, down to the clothing, jewelry, background characters, furniture, expressions. The tiny details, for me, are what make or break the images as story enhancements. If they aren't right, i don't want the image.
4. I usually include some inside jokes or easter eggs if they make sense - not gratuitous.
Process -
1. Draft prompt using very detailed instructions.
2. Include perspective, setting, characters, mood, action, style, lighting, objects, anything written (signage, graphics on clothing, media), clothing, accessories, hairstyles, direction of gaze, location, background and foreground, emotions, branding.
3. If it applies, give instructions for what you DON'T want in the image or you don't want the image to do.
4. For a subject in an image that I want to look like an actual person, i attach a reference photo and use a variation of this script, changing the details depending on the desired result: 'Use the attached photo to reference the exact facial features of the subject in the image. Do not change any facial features except for [age regression/ age progression/ emotion/ expression] according to the prompt instructions.' I use the same for recurring characters and use the same reference photo for continuity.
Usually the first result is pretty darn close. But it is never perfect, so I have to do several things.
1. If it captures the scene pretty well (emotional mood, expressions, postures, setting, etc) I can tweak some things in photoshop or similar - remove distracting extra objects, crop things, put a lighting filter for effects.
2. If there's something fundamentally misaligned with the character or setting, or that doesn't look like I imagined it when I wrote the story, I decide whether to ask Gemini to fix a single element or two or whether to start over with a new request. Or I tweak the prompt itself and resubmit.
3. There are pros and cons like - if an image is SO CLOSE to what I want and I don't want to lose the foundational elements by starting over I ask it to revise. But each time you revise the initial output, it loses fidelity. I don't know why. So sometimes that doesn't work and you have to start again.
4. Sometimes, for whatever reason, it just can't get something right about a specific image I want - like the AI is overwhelmed or something. Starting from scratch, reprompting, revising the prompt - nothing works and each request gets worse and weirder. it's often something like a subject's gaze won't correctly look where it should. Or the proportion is wrong. Many times, the simplest detail gets stuck but even something small that isn't right ruins an image - for me.
The short answer is: general requests aren't useful to me, because the entire reason I create images at all is for enhancing a very specific element, usually comedic, or to render exactly what I am imagining inside my head into a visual storyboard companion to my piece. i get a thrill from making my tilted story-world appear in accurate images.
Prompt examples (feel free to adapt if you can use for your own purposes!) for Gemini 3
(This was for the final image from Miscalculation - the triumphant car dealership refund scene):
Reference photos for the exact facial features of the principal subjects. Photorealistic, cinematic photograph, ultra-realism. Wide shot interior of a large, brightly lit car dealership service reception area — polished floors, branded (generic) signage, the fluorescent optimism of automotive retail. In the center foreground, a man and woman in their forties stand in a relaxed, confident posture. They appear magnanimous and gracious. They are dressed in business-casual, tailored clothing and understated accessories. She wears slim ankle-length black pants, a white silk blouse, and black stiletto Louboutins. He signals tech: jacket but no tie. There are two blue five-gallon buckets stacked beside the counter. The woman holds the mahogany chest under one arm casually, like a clutch, as she reaches to accept cash from an employee behind the counter. Spread across the service counter and the surrounding area, dealership employees are in various states of collapse: a manager in a polo shirt with his face buried in both hands, elbows on the counter; a service rep on the on the phone, one hand pressed to his forehead, explaining something urgent and critical that he does not want to be explaining; another employee standing with arms at his sides staring downward at the buckets with the hollow expression of a man whose understanding of the world has just been revised. At the counter, one young employee — looking deeply, personally sorry and ashamed — extends a hand toward the woman holding a single ten-dollar bill and four ones, arm slightly outstretched, unable to make full eye contact. The woman receives it with gracious composure. The dealership's branded backdrop and gleaming showroom cars are visible in the background, indifferent. Shot as a wide cinematic still, warm fluorescent light, 35mm, sharp foreground, slightly heroic low angle on the two protagonists.
So, the first result had the characters looking at the 'camera'/ viewer. That wasn't right. I asked Gemini to adjust the gaze.
Other things I considered adjusting/ redoing, but decided it was close enough and I didn't want to jeopardize the elements that landed:
- The buckets should not have branding on them, for continuity from image 1 (pyramid) in the story. This can be easily fixed in photoshop if desired...
- The woman should be accepting the money from the employee, not from 'Rob.' Oh well...
- I'd prefer 'Rob's' shirt to be white, for visual harmony. Not a big deal.
- The mahogany chest is not shaped/ detailed like the other images, and it's an important story character. I actually might go back in and redo because of this, as I think about it!
But the elements that are often sticky are things like the employees supporting the action in the scene, and it really nailed that. So I have to weigh the elements to decide how to tweak.
I think it's really fun.
Wow! Thanks for the detail. I'm just starting and learning how to use Gemini to generate images. Your hints right here are better than a "For Dummies" book if one even exists.
I know writing is not easy, but you certainly make it look so. Your thoughts come across so conversationally and just seem to flow.
I'm proud of them and just the slightest bit envious!
I know there are plenty of concerns with AI, but for hobby stuff like this, it feels like magic. I don't understand how visual artists create emotionally moving creations - portraits and cartoons in paint or drawing or sculpture. The images live in my head but i can't produce them. So the ability to use words and make an image in my head appear is just amazing to me. It feels like how nonverbal people describe their experience with assistive technology to finally communicate. if a picture is worth a thousand words, that's currency i can deal in! 🤣
I have few concerns with AI if it's directed; if it remains our tool to create with. I don't think use of Grammerly or ChatGPT to smooth or embellish original content is wrong. I don't think use of Gemini to create images is wrong.
I don't even care if writing is by Shakespeare, Gabriel Garcia Marquez, Dostoevsky, Chinua Achebe, or ChatGPT if it is good.
Or especially if it is by Jessica Haley Graves which seems very reliably good.
Have yourself a great Sunday! Are you going to visit the cousins? It's
Headed to see parents, sisters, cousins right now :)
This is so awesome. You keep giving me reasons to kick that Dave Barry subscription further down the road. Keep it up!
Plus Dave has no background at all on the science and IT stuff.