runic lynx
Is it possible to remove glasses from the girl of this picture with DALL-E GPT? It generated this yesterday, I have the ID but everytime I ask it for a change, it sends a new whole picture. I also tried with generative AI from photoshop but it destroys the face of the girl.
frank nebula
For changes you want to use a VERY similar prompt. Generally the longer the prompt the more stable it will be. If you want it to be stable, you need to ask what the prompt was, and then ask it to send over a very minor modification of that prompt.
runic lynx
For changes you want to use a VERY similar prompt. Generally the longer the prom...
So should I ask the prompt it used for the image with ID=<id>?
frank nebula
So should I ask the prompt it used for the image with ID=<id>?
Yeah give it a try
runic lynx
Yeah give it a try
Well, it says it can't gather prompts on previously generated images...
frank nebula
Insist that it can. Tell it to look for the json packet it sent to dall-e.
Oh you'll need to be in the same thread as the previously generated image.
runic lynx
Yeah I still haven't deleted the chat where the image has been generated
runic lynx
Insist that it can. Tell it to look for the json packet it sent to dall-e.
"I do not have direct access to the previous transactions or prompts sent to DALL-E. This goes for every sent or received element on the platform during our interactions". I hope I translated it well. This is what it said.
runic lynx
Are you in the same thread?
Well yeah, this has been generated yesterday, here it comes the part where it did
Oh I did it
It sent me the prompt
"Create a circular logo for a food blog, with a male and female character in the Apple memoji style, excitedly looking at a pizza in the center of the table, with mouths open and joyful expressions. The characters should be as in the previous design with reference image ID "rHfpw9xtcdPswyOy", but remove the glasses from the female character. Maintain everything else in the image as it is, with the minimal border and the pure gradient background"
Wait, this seems a prompt I asked later, to remove glasses
Not the one used to generate it
frank nebula
Ask it for all prompts sent in the thread and see if you can find it
runic lynx
It is struggling to give me the correct prompt. It is giving me prompts I asked later. I'm keep trying
runic lynx
Ask it for all prompts sent in the thread and see if you can find it
Oh, I'll try, I hope it can
analog cosmos
Oh, I'll try, I hope it can
Hey jack, question. Do you still have the original creation in a chat? if so you can see the exact prompt by clicking on it and going to the info
runic lynx
Hey jack, question. Do you still have the original creation in a chat? if so you...
Oh man, really thanks. Here it is:
Create a circular logo for a food blog, with a male and a female character portrayed more animatedly and lively as in the user-provided reference image. Both characters should have the style of Apple memoji with the female character having dark wavy hair, glasses, dark eyes, and resting her face on her hands, and the male character with dark hair, a stubble beard, and dark eyes. They should be looking towards the center at food items on a table, specifically at a slice of pizza and a hamburger. Include a colorful gradient background and make sure the table is visible, resembling the setup in the user's reference image.
Now what should I do?
analog cosmos
gimme a sec, i have the image of a tatto mixing freebird and a chicken being generated then ill tinker with it, lol
at first glance, removing the word 'glasses' for the girl should help, but its going to be really hard to reproduce an EXACT duplicate. im trying a few things on my end
runic lynx
This is the image it is talking about. My reference, as it said
Anyway, if I wanted to try, should I give it the image I want and then paste that prompt with no glasses?
analog cosmos
i havent tried generating the same pic multiple times in awhile, i got this one... And im trying a prompt for generating a similar but different one
runic lynx
Well, it is really good! I'm struggling due to the slowness and network errors it gives to me
analog cosmos
It messed that one up, lol, pizza hamburger
But yeah i feel something changed in the last few weeks as far as how much power it takes to generate pics. I know its harder on my computer than it used to be, but luckily ive got a pretty decent setup
i'm getting different pics now, which is good. If you are happy with the style, here's what im using...
Using dalle. Do not add, remove, or change any part of this prompt. Be sure to use a different seed or ID than has been used in this chat already.
Create a circular logo for a food blog, with a male and a female character portrayed more animatedly and lively as in the user-provided reference image. Both characters should have the style of Apple memoji with the female character having dark wavy hair, dark eyes, and resting her face on her hands, and the male character with dark hair, a stubble beard, and dark eyes. They should be looking towards the center at food items on a table, specifically at a slice of pizza and a hamburger. Include a colorful gradient background and make sure the table is visible, resembling the setup in the user's reference image.
runic lynx
Yeah, as you can see it completely changes the image
analog cosmos
correct, there's really no way around that
runic lynx
Wait, are you using GPT-4 turbo API? I'm using normal ChatGPT. I haven't tried with APIs
analog cosmos
nope, normal gpt4. i havent done any api usage
Now... i do have a suggestion. if that first one is 100% what you want, except for the glasses... throw that baby in photoshop and remove them, lol
runic lynx
Not really, the first one you sent I think is perfect. Also, I already used photoshop with generative AI to remove glasses... results were horrible lol
runic lynx
i havent tried generating the same pic multiple times in awhile, i got this one....
Anyway, I think this is really good. I'm not very happy about the hands, but it's ok
analog cosmos
yeah i just noticed something when i zoomed closer to yours... Her glasses go into her skin. making it a lot harder to photoshop out, lol.
runic lynx
Well I have never used photoshop in my life... is it quite harder for me for sure ahahah
frank nebula
Hey jack, question. Do you still have the original creation in a chat? if so you...
Wow had no idea it was in the meta data.
analog cosmos
Wow had no idea it was in the meta data.
Yep. a lot of my prompts i will use something along the lines of 'do not alter, change, add, remove, or translate anything...
Because it has a habit of doing so
runic lynx
i havent tried generating the same pic multiple times in awhile, i got this one....
Could you give me the prompt for this one and which image did it refer to?
analog cosmos
Could you give me the prompt for this one and which image did it refer to?
I used the same prompt for them all.
Using dalle. Do not add, remove, or change any part of this prompt. Be sure to use a different seed or ID than has been used in this chat already.
Create a circular logo for a food blog, with a male and a female character portrayed more animatedly and lively as in the user-provided reference image. Both characters should have the style of Apple memoji with the female character having dark wavy hair, dark eyes, and resting her face on her hands, and the male character with dark hair, a stubble beard, and dark eyes. They should be looking towards the center at food items on a table, specifically at a slice of pizza and a hamburger. Include a colorful gradient background and make sure the table is visible, resembling the setup in the user's reference image.
runic lynx
I used the same prompt for them all.
Using dalle. Do not add, remove, or change...
No images in input? Just prompt?