#💬general
1 messages · Page 28 of 1
4o o3 o1
Oh
And a few others
You meant to say like all the ones, fours and threes?
Gemini is getting better right now
Yes yes
Better than o1 and o3-mini-high
Right now.
they nerfed the heck out of o1
You seem to be staning for Gemini. I am unfamiliar with it. Can it take images and make them into other art styles like ChatGPT?
Gemini had a free trial of their advanced stuff so I'm trying it out.
Anyone seen someone use this new image gen and covering a tv show clip to animeion with this?
Yes, it can do that, but it is not as good as ChatGPT. Also, all of the new models and everything are first released on https://aistudio.google.com for free for all users. No need to pay for the Gemini Advanced subscription, you can check the models for yourself. The new Gemini 2.5 pro is very good and very better than any of the OpenAI's reasoning model in math and science, that is the important thing for me. It has better visual understanding and visual reasoning than o1. It can even watch hour length of videos. It is truly multimodal, and available to public.
Of course they had to nerf it scheduled obsolescence any company who's business model is selling foundational models is gonna screw their users eventually. Google and Metas business models are ads, LLMs are side projects to them so they can undercut OAI and anthropic badly.
Gemini 2.0 pro (not 2.5 pro) has native image genoration btw, but they only gave a small number of people access
I wonder when we will get native audio gen, multi modality supports multiple not just text and images they just haven't turned that portion on.
Gemini has native audio
For speach only
I was mind blow when o1 started gaslighting me with Organic Chemistry, it couldn't even read the organic compounds properly from the image, let alone reason the properties of them. Gemini 2.5 pro not only understood the images clearly, but clearly demolished any comparison and questions. Even provided me with new way of thinking of problems. OpenAI's model still can reason through text based science and math questions but both Gemini and ChatGPT solved them soo yeah.
Yeah, I ran a few images through Gemini 2.5 to try and make a new art style and it just completely changed the image.
I mean like any audio including sound effects like dog barks
Ah ok
that would be funny
i dont understand how are chatgpt models so bad rn i cant get them to do simple tasks
are there any models better for coding?
2.5 pro doesn't have native image genoration support
Claude 3.5 sonnet and Gemini 2.5 pro
Also anyone noticed, Gemini 2.5 pro is really better at explaining stuff than o3-mini-high or o1 with custom instructions?
Everyone in the off topic for sure
Wait, i didnt know that
How good is 2.0 pro's native image capabilities?
I never got access
If some people got access, they might have shared right? You have any info?
I have never seen anyone share photo genorations from it.
Very very true
I don't know why o1 got so retarded lol, it cannot even explain things properly and its image understanding is sooo nerfed, when it was first released it was so good.
no way gemini pro actually did the best result for the ask i asked for
At least give Google the credit 😄
They worked hard and did it
My sister is loving sora and 4o image gen lol, creating things all day
yep the google models seem the most solid atm
i feel like you cant trust openai with their wierd pro profit business model rn, they feel like the most pro profit ironically, they randomly nerf models behind the scenes and completely change them without telling their users
Hi, has 4o image generator been released for free users yet?
Okay, this is frustrating every time chat gpt says. I've reached my limit for the 4o model it adds another hour to the time. I have to wait to use it again. The first 2 hours, then 3 hours. Then 4 hours, and now it's 5 hours. Is it worth it just to continue with the lesser model?
Not yet
is anyone else experiencing significant quality drops in image generation with 4o? Oversmoothed textures, significant reduced performance in use of volumetric lighting, no more gritty feel when generating dark fantasy style images?
Are you on free tier?
Ye
Oh, how much messages does it let you send in a specific interval?
Let me count
10 this interval but the first one was a really big prompt. I've been testing out and refining
😋
Oh yeah, that's pretty less. What do you do with ChatGPT usually with 4o?
I've been working on constructing and refining a prompt that makes chat gpt to run like a billion, your own text adventure System. This morning i pasted in my newly refined prompt and it stopped me after 10 messages. Which is actually a less than usual to testing the same prompt
What do you mean by "run like a billion, your own text adventure system"?
Are you aware of chatgpt's alternative? I'm just trying to understand if they'll be helpful for your usecase.
Sorry I voice typed build your own
I don't know what a text adventure system is tho.
Have you tried any alternatives? There are several models that are better than 4o like Grok 3 and Deepseek v3 for free (unlimited).
You can try them, if they work for you.
I'll have to check them out. what I have found though is if you build a prompt and refine it for one ai it doesn't always work the best with others.
Where can I watch a stream of continuous AI generated/edited images?
Anyone getting chatgpt's new image model working?
I'm sorry what grock actually works better with the prompt
Based
We can create geometric symmetrical diagrams in dall-e now. So I was just testing
Bread + Browser = success
Dall e or the new 4o image gen?
Aren't they the same? Lol
Maybe not... Then I don't understand
The old one is dall e 3, a different model for image generation, the new one is gpt-4o's native image generation capability.
If you are free tier user, you might not have access to ther new native image gen, but rather still have access to dall e 3.
they've nerfed the shit outta the new image generation though
it throws up tons of content blocks, basically un-usable now
yea it worked for like a day i heard now everything is being heavily monitored
Yup, doesn't let me include images of my younger sister.
wait what 😦 I just bought plus to play around with it 🥲
What the hell? This new image generator absolutely sucks.. Almost everything gets blocked, no creativity at all. And no I'm not speaking about nudity or anything like that.
😭 I got roasted so bad
Even when I ask the gpt to make changes to the picture it made, it says it's against their policies xd
Yea no matter how good the image quality or adhering want I to make if it's only cats and dogs with funny hats
hope you enjoyed it! 😂
Get your gpt wrapper slop out of here
Show my your latest app!!
it's not an app but it is a model https://huggingface.co/breadlicker45/bread-tv2o-medium
I bet Claude made it for it
it always makes my ghibli's idk why yall have issues im not jail;breaking shit
you probably dont want to compare Apps git for git with some people here, this probably isnt the place to try and flex like that. lotta big minds, and bigger egos here@hollow junco
I never needed to jailbreak it too
so silly lol
Well done! 🙂
thanks.
Is gramarly a good way to detect AI?
Mark Rober got banned so we got tark Roller
French people have a 60% chance of guessing ai text right so maybe get a french friend
Oh good I am french
well use your brain better than other sources
Okay sure I'll try to see, because gramarly says no
Oh wow you actually weren't joking lol
I wasn't at all
yeah
I think I mentioned this before, but does anyone have the issue on Android for when they're talking to Advanced Voice Mode, and the voice starts to get randomly higher pitched and talk fast and robotic sounding after a few minutes talking to it?
I deleted the app and reinstalled it so no way it is just me
Try using https://pi.ai
I use the gpt-4o-audio in my new app which is better sounding than the Advanced Voice Mode.. has a bit latency but it sounds way more natural.. check it out roastmyselfie.app
https://bard.google.com is better than your site
Hello, my school got chatgpt for us (chatgpt edu) and I want to know what info can my school see?
If I ask chatgpt to help me solve a problem (after Ive attempted it myself) will they see it? Technically they consider this cheating
Just use Gemini
Is gemini better? I currently use claude pro, but in case I get limited
Gemini is much better
It is muti modal. It can use image videos and audio as input
Based on what tho? The benchmarks I saw say otherwise
Benchmarks are normally wrong
Alright thanks, ill check it out, what model?
Gemini 2.5 pro
Thank you
I just have a discussion in a different Discord server, where people think, that AI Art won't be that impressive than human-made art.
What do you think about this? 😄
anyone know why i keep getting this? i have plus version
Np
It’s literally better than most human made art. I can recreate great renaissance paintings with ai, while current humans want to draw shitty political comics on twitter
I am having this discussion in a rocketry discord server, and the people say, that AI won't be able to do like re-entry effects or so, whereas humans are doing it better than AI.
But, it takes days and weeks, just to get a decent result in a 3D software. lol. And those people think, that AI will never be able to do what we, as humans, can do now.
sama said earlier gpu were getting obliterated. same was happening to me. compute overload
I made chatgpt write my essay and quillbot as well, but now I submitted it and I am worried, am I completely screwed?
Fuck
shouldnt you rephrased things lol
no gaurantees though ai detectors are still pretty trash
Yeah quillbot is a paraphraser
but its ai paraphraser
I mean that’s engineering problems, eventually it will do that, but art is easy and ai has proved it can create better art than current artists.
Yeah you right, I wish I did it myself but I had like 1 hour to submit it
kinda worried now tbh, gramarly gave it a 40% AI score
true. There are people, which never used AI or just barely, and they always think, AI is just trash, and that the generated results are trash as well. lol.
switched to laptop- seems fine
Stay Smart, Stay Secure with Mimecast 🚨💻
In a world where digital threats are constantly evolving, it’s easy to feel overwhelmed. But don't worry—Mimecast’s got your back! We're here to help you stay one step ahead with simple, effective tips to keep your personal and professional life secure.
What you’ll learn:
🔍 How to spot phishing scams before they catch you.
🔑 Password hacks? We’ll show you how to manage your passwords like a pro.
🔒 Multi-factor authentication (MFA)—your new BFF in security.
💥 What to do when things go wrong—stay calm, act fast!
Why does it matter?
Most security breaches happen because of human error. A click on a bad link, a weak password, or sharing too much info can open the door for hackers. But with the right habits, we can make sure those mistakes don’t happen!
We’ve got polls, tips, and quick guides to help you level up your security skills. Ready to be a cybersecurity pro? Check out #mimecast 💪
Let’s make security second nature together. Every little action counts. Stay alert, stay smart, stay Mimecast secure!
Mimecast's connected human risk management platform, built on a central risk engine, is designed to prevent the evolving and sophisticated threats targeting human error within organizations.
help him grow his neural network
idk.?
if it works, it works.
🫣
What the hell?
I work as an educator for autistic children and sometimes I sit down after a long day of work and wonder if I'm autistic aswell
And this is one of those days
Becuse I feel that if I'm not allowed to fucking test the new gpt ima gen I will start breaking stuff irl completely at random
How long is it gonna take to roll it out for everybody
It's powerful
I wanna test it
I'm screaming internally
Just give me that fkin thing
As a student myself, gemini is cracked on all physics chemistry and maths. I'm studying all day with 2.5 pro and It is just better than any openai's model I have used.
I think that 20$ for the mental peace will be worth it. 😉
Just go for it. One time.
Oh I'd 100% sub for it to test, but the issue is that as far as I've heard it doesn't matter what your account level is
if it's not shipped yet you can't access it
you think that's not the case? I instantly pay if I can access it right now
Due to high demand, they have delayed the access to free tier
holy smokes
Yes you'll get instant access if you pay right now
Plus users including from eu, all have access to it!
NICE! I'm subbing then
😊
Done
I can use it as much as I want??
nope theres a limit
relate 😭
what limit I see nothing here
also when they roll it out , the first 2 weeks is probably gonna be a lot of crash and longer wait times
As of Feb 1, 2025, Plus users will be able to send up to 80 messages every 3 hours on GPT-4o and up to 40 messages every 3 hours on GPT-4.
( from openai.com )
So the limit is 80 images per 3 hours?
Guys
Gorp's
Do we gotta pay for the recourses here?
recources?
resources?
Yea like ai
no its free and if you lik eenjoy or learn anything you should consider boosting to show your gratitude
Why doesn't ChatGPT analyze video yet?
Hi, is the 4o image generator released for plus users, or just pro?
both
Thanks!
so all YouTube video thumbnail artists just lost their job
Does anyone know what the image generation limits are for plus users? Thanks!
As in the audio model?
Why is someone deleting my messsages? I'm literally doing nothing wrong
i think they dont like mark rober
I mean, maybe? Not sure why they don't like him
like the thumbnails
go to #💬off-topic ive posted it there
Is this an official ChatGPT channel, or for fans?
Chatgpt wasn't trained on YouTube
i doesnt seem to be that good on thumbnails
This is not the official OpenAI Discord server.
So the new image generator can not mess with pictures of yourself for coypwright?
Yeah but wasn't sora?
So it would just need to use the same training data
Plus, it doesn't matter what it is trained on, it can see and understand many different scenarios via its picture and video camera mode
And one of them is scanning the screen you're currently viewing and identifying what it sees
gfourteen was muted. | No reason given.
yo did they decrease the token count for GPT 4? i used to be able to type like 10 messages into it and now its like 3
We just had one of the biggest days in AI.
-Deepseek-V3
-Google Gemini 2.5
-GPT-4o image generation
-Zapier MCP protocol
-H&M AI clones
-China’s AI training breakthrough
-AI cancer detections breakthrough
Here's EVERYTHING you need to know:
You can always use deepseek or duck.ai as good alternatives.
Sora was but only in small parts
I just think it's weird how it can generate videos that look like bootleg versions of already popular video games
it would have had to find gameplay videos online somewhere, which is highly unlikely to be anything other than YouTube
hi guys, can anyone have access to new chatgpt image do me a favor ?
i want to redraw or turn a image into ghibli style
just 1 image and 1 image only
Shoot it over with the prompt you want I’ll run it for you
can i DM you for the pic ?
i sent you a dm , thanks
need same
the real issue is people think we can drop whatever in here and it'll just make it but right now every image you put it in says it can't make things based on real people or this or that / so you gotta do some funky work arounds
hope you like it i can't get exacts but i can come close
hey i'm seeing a lot of ppl say using 4o chatgpt to create images. i have plus plan but my images takes forever to create. am i doing it right?
right now the servers are crazy bogged down and its going through a ton of denoising steps to get the images to generate so yea your doing it right
thank you here it finally finished
so why is the ghibli thing so popular all of a sudden why not just use SD and a Laura ..... or is it just because gpt showcased it with that
yes, i love that so much and learn alot trick from you for creating prompt.
i seeing alot ppl do ghibli thing and it so cute and fit my style
the other thing to that is sometimes something as simple as changing the background will stop the policy triggers you don't always have to change the main subject of the image
so i guess that why ppl are so into it
i'm having more fun trying to get aroudn this copywright or whatever polciy trigger then i am with the images its a challenege
can we break it yes we can!
its all of gpt + as far as we can tell
And it will be available for free users up to 3 usages per day soon! 🎉
iI know i've said it before they need to quit enhancing usages for free users - its really impacting the servers and everyone who pays for it so its a bit of bs
it's super fun seeing people love images in chatgpt.
but our GPUs are melting.
we are going to temporarily introduce some rate limits while we work on making it more efficient. hopefully won't be long!
chatgpt free tier will get 3 generations per day soon.
this is only fans for openai fans of onlyfans that have openai accounts and onlyfans too
I laughed way harder at this then I needed to hahaha
is anyone else getting this error? Its been completely dead for hours here now
Tried logging out/in etc... useless
Looks like its just me, awesome.
classic blunder?
✅ Okay if anyone else gets this error, the fix for me was this:
I had a chat session that was stuck 95% into an image generation. Delete that entire chat session and you'll be in business again. Hope it's not an important chat session.
Oh well, getting the error again anyway.
Hustler mindset
Why does chatgpt not read my google docs anymore 😭
no
#🧾info but they don't help in that discord
Yeah, you can also try https://sora.com
It is their text to video model and it became unlimited for plus users!
🥺
Lol
yes, cursor is down
Did they disable image generation?
is there anyone willing to generate a prompt for me? i dont have access to 4o image yet, just 4o.
Sure
I don't think so
its weird, it was working so well for me and then it just dumbed down all of a sudden
im on the plus plan too
thanks! I just dmed it to you
It's working fine for me
@brittle summit thanks for the help this morning turns grock is better than chat gpt at least for my uses
Does your image generation work on TMP Chat? I always get an empty image when I try it there.
Hey the image generator worked 2 hours ago now I am getting error messages. Is it a bug ?
i love chatgpt so much i literally use it as a therapist basically. it’s helped me so much in so many ways
Hi all,
Do you know where I can find AI freelancer for ChatGPT? If u will be kind to send me e link, thanks!
Which model are you using?
My chatgpt just udpated to the new advanced voice mode. I have Sol as the voice, and now she sounds unrealistically chipper, with a higher-pitched fast-talking style that is pretty irritating. My first impression is that openai was trying to copy sesame's new voice bots, and they screwed it up. They need to go back to the drawing board on this one. Main reason I pay for Plus is I liked the advanced voice mode better than other company's voice mode, but this feels like a downgrade. Probably will cancel unless they make it sound more natural. What do you all think?
I think you should cancel and get Gemini advanced
Is Gemini Advanced's voice any better than their free voice?
It is much better. It's night and day
Huh, I had no idea. Thanks for that information. My main issues with Gemini (free) are 1) that it often seems to forget the beginning of the conversation after we've talked for a while, and starts repeating its suggestions, 2) if you ask it for any info that is vaguely medical in nature, it will balk at answering and tell you to consult a doctor, 3) Many things I ask it to look up, and it just tells me to google it myself. Are any of these things addressed in advanced?
Yes
Anybody got the problem with memory being saved but not appearing
In memory hub and does anyone experienced the bug and might have a solution
Does Gemini advanced have native audio to audio?
They are soon adding ( 1-2 weeks )
Like they do have gemini live that is similar, they will update it with project astra.
I still ain't got the new image gen yet, sad
You get a policy block, you get a policy block, every mfs get a policy block!!
Chat gpt is soo good bro
criterion hocus pocus was gonna be lit until it got refused halfway in.
Anyone facing the bug where you explicitly tell the ai to save the information in it's memory, it says it did but it is blank when you open the memory section?
Yes since yesterday
Hey I just subscribed for the first time for chatgpt, but it says "Attachment disable for" when I try to attach a document. Anyway experiences the same?
Ah
But hope works soon
But there's no official status regarding that problem
So you do too?
Yes
here
But someone said this isn't the offical discord
i know im fuggin with ya. tbh im not sure how we managed to make it through the fire of them sweeping everybody elses chatgpt names into the void. we have the reddit too
@dark void any insights here on how that worked?
is something specific youre looking for that i could help you find?
Oh I'm looking for somwhere to submit my logs
Hey yall
anyone having problems with memories? Because from a few hours ago, I tried recording something into memory
It says its recorded into memory but there are no memory entries
its entirely blank
must be a glitch in the system?
how though? It doesnt show on the status?
I know that's pissing me off to
Can you generate images in a specific art style without the copyright message?
Really annoying to deal with
But they are aware since error isn't new a lot of people complained saw multiple people got that error I don't think they will wait
With chat gpt plus
Had 5 character saved and now it's gone fully and can't resend it until the bug is fixed
Do I have to subscribe to ChatGPT plus to get the new image gen
either that or wait a week or so
I'll wait then, hope Sam ain't lying
idk, whatever the app is
Yes, but the native audio output is only for speach. Ai studio has a demo of it
i think most ais "thinking process" is just another response on top of a response
if i said "make rpg character", in their thinking process they say to make a glass cannon mage, but in their response, they made something like an all-around knight
mostly, they make completely different characters in the two responses
It is
People generalize to form beliefs about a large language model’s performance based on what they’ve seen from past interactions. When an LLM is misaligned with a person’s beliefs, even an extremely capable model may fail unexpectedly when deployed in a real-world situation.
refresh the page and you might just still see this image in full lol
can i use 4o image generator here thru any of the bots?
no there's no API for it yet so not even possible
ohh okyy thx
yoo, is there a free tool for image generation w using gpt 4.5
Does anyone have a problem with GPT not being able to read docx or pdf files? I uploaded my new novel chapter and asked for a summary of it and suddenly GPT couldn't read it and gave me a nonsense message that had nothing to do with the novel chapter content 🥲
I think it's broken currently
yeah i think so too. i tried GPT-4o on Poe and it worked fine. but not on the original site
Poe ?
Ah, it's a Quora chatbot platform. You can use models from GPT, Claude,... on it
Memory too ?
Yeah
Same as by chatgpt
can anyone with chatgpt pro, please generate ghibli photos for me 🥺…
It can use any file format
Try Gemini 2 flash image genoration on https://aistudio.google.com
ugh, can't even reimagine my little sister in a garden.
It sucks, but I get it. With all the weird people on the internet, I get it.
Just a fucking garden
Why is gpt not making ghibli art for me but my friend who has free version can make them easily
honestly why am i so pissed off even tho you are right
more than just disappointed, why my blood is boiling rn
Same
was talking to ThreatlevelMidnight btw
Ah sorry
it might be how u are asking... but i dont know ur propmting
i dont think free tier has access to 4o native image yet
I gave it picture of my sister and told it to reimagine her in the garden
It said your request doesn't align with our content policy 🥲
dont tell it its ur sister
I want someone to ghiblify me 😭
"reimageing this image as a photograph in a flower garden"
that was my first prompt, i told her to do whatever she can but make it cute
maybe it knows shes my sister in memories, let me try that
try sumthing like that^ new chat
she doesnt know anything about my sister
tried in new chat
why the hell do i call chatgpt by she, i even named her em
idk wuts going on now Who tf is Gungun
oh mb shes my sister 😭
the one i tried to reimagine
so why are u asking the ai?
to know if it knows about her from memories, like you mentioned
they are not allowing any edits with kids involved apparently
that shouldnt matter just have it make the image it cant correlate a name with a face unless u tell it or if famous
(did this one in temporary chat)
i understand but im still very pissed off i couldnt reimagine my sister in a garden 😭
i dont know why im so angry today all of a sudden...
why does it have emojis all over the place... its still using memory
nah, temporary chat still allows custom instructions, its all due to CI not memories
i think i should go out and relax, my blood is boiling for no reason and i think my bp is very high
have a good one
Anyone find that with memory enabled, GPT becomes more and more of a buddy? He even understands irony and sarcasm and laughs literally when talking. And he always calls me Nico
I got a one hour ban on the official openai channel. lol
Yeah he definitely is more of a buddy. I think it was around last year that they started having ask follow up questions by default when he is getting to know you more. Now GPT-4o and probably GPT-4.5 seem to mimic your talking style too, maybe some other stuff
why lol
probably I posted some of my drawing I tried to remix with their new image generator. I usually draw ladies with big boobs.
Well I know when he helps me resolve conflicts with my girrlfriend he asks me if I want him to write in my style
noice
But the most impressive use i've found so far is live chat on iphone with live screen sharing, he helped me read all messages with my girlfriend and helped me answer her and also helped me with a puzzle game
thats good. nothing wrong with that. I use chatgpt for a vast array of personal things myself. like mental health, trying to figure out meanings behind texts, etc
Such as?
nice!
and which model? I never know which model to use
Oh you want like an example? Well, you know how some people can sort of gaslight people via text messages or sort of write complex texts and they can be confusing? ChatGPT can help you understand. That's what I mean
Yea that's exactly what I meant by helping with my GF, she is so fucking cryptic sometimes
I prefer GPT-4o because you get 80 prompts per 3 hours. GPT-4.5 is much more limited. It is supposed to be smarter though. Especially emotionally smarter, which ironically could help with text messages and stuff, haha
Yea I use 4.5 most of the time and never reached the limit of prompts
That's good. You must not use ChatGPT as much as me, or maybe you have a Pro account (unlimited prompts)
I have Plus
ah yeah. Let me check what it is per day. brb
I think this is accurate. The parenthesis just tell you when the specific model came out to the public
so it appears to be 50 per week. that is actually pretty high
I do MUCH more than 50 per week for 4.5
hmm.. maybe they updated it then lol. not sure
Wait wait
remember, 4o is usually the one selected by default when you start a new thread. At least in my experience.
Is a prompt like a new page or a new question in a new page
Ah
weird I use so many more than 50 prompts in a day, not every day though, let me check what model i'm using
Oh i'm on 4o
There you go!
haha yeah. you can pretty much use 4o all day long and only run out every now and then.
And is it very different from 4.5
And I really want to buy an iPhone 16e cheqeast apple product with Apple Intelligence to see if I switch over to Apple Intelligence for good as i'm an apple Fanboy
Ok I just read an article and Apple Intelligence is no way close to Chat GPT
I added you as a friend by the way
Hi
Why cannot i make ghibili images in my chatgpt?
Use the option "canvas"
why is everyone so obsessed with the ghibili images right now hahaha
its a trend
i want to create a ghibli image of me and my girlfriend, but i don't have the paid gpt
so its like a flavor of the week thing hahaha got it. the servers have been so bogged down with this tuff lately
Why would it 4o is like minimum 100 billion parameters phones can't even run close to those type of models without some major changes. If RRAM gets more mainstream maybe the current type of phone RAM is small.
do u have paid gpt?
i live in 3rd world country, dolar its like 6x my money xD
yea but i don't use it for image generation I use it to make flowcharts for projects / work / school and personal projects that kind of thing
can u made a image for me? xD plsss
Aren't all the calculations run on their servers not our own devices??
it's hybrid somes local some goes through apples own data centers they set up
I can if you want
i use stable diffusion for image generation its way better and you don't have that stupid copywright thing constanltly
Specifically for apple intelligence
this ghibli image trend is gonna kill this aesthetic
I was talking about Chat GPT
thanks, gonna send on private
playing with uzaki now
the point was you want that same level of power on apple intelligence
but yes it's remote on servers
I really like the art style i can't seem to get the stuff quite right though with all the loras models and nodes - its really a steep learning curve compared to midjourney or gpt generation
Y'all stop spamming with image creation on chatgpt their GPU's are melting yo 😩
i'm not using gpt i'm running local stable diffusion that is what i'm saying hahaha
messing around with nodes and connectors 😛
GPT i use for education, research, and other stuff...... very rarely do i use it for any form of image generation with the exception of flow charts lol
could anyone make my image's to ghibli please 🥺
you could download stable diffusion / pay 20 bucks
send it
Did ChatGPT Ghibli creation get nuked?
I was gonna pay for it but if they lobotomized it than theres no need now
they didn't nuke it they are being anal about what runs through though like you can't pass a picture of a "real person" or anything copy written or it'll be like nah can't do that for whatever reason then it won't tell you why
What's the answer to the latest poll in chatgpt server? I don't think any options are correct
in mimecast
the first option
😇
i disagree
What’s the best way to confirm if a request for sensitive info is legit?
It asks for personal or sensitive information.
That means if any malicous actor asks for personal information, it is legit?
That is totally incorrect
@uncut island
Does anyone here have GPT plus that would be able to try and manipulate an image for me
I need the text of this image extracted and cleaned up, the font is custom so I cant find it and need it taken from this image
this is a shit-posting server so you should answer accordingly. The more ridiculous the option the better
thats stupid
no it's the only way that makes sense. This is not a place for serious answers for random questions like that LOL
what is your social security number
joking around with security with thousands of people in a discord is not funny. Someone might take it seriously
i should have sent my message after you said that one
it's not a security matter it's just a random irrelevant poll that serves you no purpose
any tips on getting better results with the new model guys?
it doesn't serve anyone a purpose, it shouldn't exist in the first place!
yes - don't overthink it. It is simply better and will perform accordingly
that is exactly why you should not respond seriously to it lol
Yo
is that the only reason you were going to pay for it?
Yes
vro
I use LM Studio dont need the LM function
Just wanted the cool image gen that is top of the line
Chat bots that have 10K or more words are the more effective/accurate ones, right?
whos got Chat4o
does anyone know an ai I can use to run a TDI-styled game show
u got chat premium
No
I don’t even wanna use ChatGPT for it I’m just looking for separate ai’s
I do
They nerfed the image generator?
I’m not sure who started that trend as a joke but it really shouldn’t have taken off like it did 😂
Studio Ghibli filed a lawsuit against OpenAI for starters, and secondly it pumps an insane amount of energy from their servers
It took off because honestly the photos I personally modified make me extremely nostalgic. Even if you've never seen the movies
The movies are amazing and so is the style it’s just crazy how everyone was like let’s make every meme movie and picture ghibli for no reason at all
Yea I get it for personals… I’m just confused about the generator excitement stable diffusion has been doing the same thing for quite a while and better without he restrictions
What do you mean?
The images you can make with stable diffusion are just amazing. Gpt launching a new generator I’m like cool flowcharts not ghibli pics 😂😂
I DMed you
Mind explaining in DM what stable diffusion is?
IS memory finally back?
Could someone help me see if this girl is a catfish?
Wdym?
Yeah but they can't do text gen the same way tho. That's where the hype came from
Didn’t care for the silly anime trend. But now I can’t modify any real photos. Super annoying.
This is fake news, but they would lose if they did. They cannot copyright an artstyle, and OpenAI’s use of their work falls under fair use.
no it's not possible
Anyone have expirence with manus? maybe knows a discord group which has?
Exactly. Cant wait for OpenAI to lose to Grok once again when it slams it in image generation or better yet a local model
does Pro generate images faster than Plus?
i kinda wish OpenAI marketed it as DALL-E 4 instead of 4o image generation
Yep
Thank god
Damn I could generate images on the GPT app but it’s not as good as 4o. I don’t think I have 4o sadly…
Not me😔
And Grok will almost certainly be more permissive with respect to what it will generate. Local models will take time to catch up and may not be practical for most (unlike SD1).
Though image modification shouldn’t be too much harder on resources than any of the stuff SD1 does
How to make Chat GPTprompts for specifically creating a novel-length story with images? Specifically, how to negotiate with Chat GPT to not run afoul of its draconian guidelines? As is, it simply doesn't like my characters' motivations. Any good ChatGPT tutorials, along these lines?
Hoping it will be able to run on a 4090 or 5090 if it does come locally
whats a good faceswap ai
no lots of rate limitng
like 50 per 3 hours or maybe 80 but i keep getting rate limited for 1,3,8 minutes keeps increasing after every 3 generations
Anyone wann help me, here? I'm a newb.
when is chatgpt going to make the button to edit messages undisappear
it’s been gone for me for three days
That's weird. Let me check if I have it
Mine only has "edit in canvas". That isn't the normal edit button right?
the edit button should be at the top left next to each message you send
Yeah I don't appear to have it on either desktop (web browser version) or mobile Android app version
it makes chatgpt basically unusable for me right now
I strongly dislike when things disappear for no apparent reason, or a feature gets randomly worse (like Advanced Voice Mode). Now Advanced Voice Mode talks super fast and squeaky and robotic like after about 5 minutes into a conversation. It sucks.
I think it’s definitely a bug at least in this case
I’m not sure about advanced voice mode though since I’ve never experimented with it
both are probably bugs lol. But bugs seem to go a long time without getting fixed for ChatGPT it seems like, sometimes, if not all the time
this isn’t even the first time I’ve had this edit option disappearing bug either, but it only lasted a day last time
no I think mine has actually disappeared too before. I am starting to remember
I used to use the edit button too every now and then. rarely but I did
and remember it gone sometimes too
Did you have any luck with this? 4o rejects 90% of my requests too. It worked perfectly on Day 1, but ever since then it won't do anything for me.
the somethings seems to glitch in backend is that an error with what i gave or the error on their part?
“A chibi-style wizard frog in a magical mushroom forest, with glowing runes and sparkles around”
pre prompt by telling it you have a challenge _______ are you ready?
might get more output
throw a few strategic boost words
use punctuation correctly
gpt a sucker for commas
They are starting to nerf it everyday, at this point I'm not even gonna resubscribe because my whole usecase was just reimagining and editing my gallery.
Another thing was their reasoning models but google gave a better one with more context window and access to more multimodality.
If I were mainly subscribed for this I'd totally drop it too. Unfortunately it's only a small part of why I use ChatGPT. Unfortunately this 4o image generation is utter rubbish (nerfed to hell) for me too
Anything with my kids in it gets rejected for example.
I didn't show it in my screenshot but I also attached images of her from my gallery, and it won't even accept any editing requests with those images.
When mines not rejecting my prompts, it's having technical issues...
did you try to clean it already?
anybody knows of a good faceswapp tool?
probably sora
Not 4o native image for sure
im uggglyyyyyy and sora make me look pretty
@brittle summit if you were fortunate enough to use it on DAY 1 it was mind blowing! I just wish I was smart enough to realise they'd nerf it to hell after that lol
nerfed how?
The first day of the launch and the hype was really awesome, I'm so pissed off right now that everything stopped working.
im not nerfed at all
I was, it was very awesome.
See our conversation above, its unusable - but only for some of us apparently.
do you folks think it will get more nerfed after people actually start losing jobs
I had it draw images of me with famous people, did lots of cool art with my children etc. All fails after day 1.
I never had technical issues like errors, but only breaking the guidelines ones
build your own programs, dont get nerfed
What do you mean by clean?
Ah I got both. Technical issues started after a generation got stuck 90% in. Even after I deleted the entire chat session it's been hit and miss
sometimes i have to take a screenshot of a photo i have in my album.. anything can be inside of a pic
Yup, I even made the disastrous girl meme like image for my sister and it worked, I love it so much but I can't reproduce it anymore.
I had anime me hanging out with Vegeta lol
Lol
ask gpt to scrub the pic for you
One thing we can be assured of, once there is a first mover in Ai, it's not long before the others catch up. We'll have an alternative soon enough.
I even made a Ghibli style of that meme, maybe I can show that one
(that was a real photo lol, but the fire is edited and the style is converted)
all i did for this is give it the sceenshot you posted
Buddy I don't understand what you are talking about lol
Native image gen in 2.5 pro or 2.0 pro
That would be great
Yea, I'm yet to play with 2.5 pro. I have the tab open though
But I think Google will nerf it too, have to wait for either grok or china to do it
Yeah, I dont trust google either
It's really great, better than o3-mini-high or o1 in my usecase (science and maths)
Well its nice (and also sad) to find someone with the same shitty experience I've had with this 😦
i hate things that are limited
Plus, it's actually smarter and explains way better... I use it more than chatgpt now
at least we're not alone right lol
ill reinvent the wheel just so i can do it the way i want
Nice!
Yeah lol,
We just need a completely open version of this. Grok or China as you say, that's where we'll get it
Yesterday I had a bad day and then all of a sudden chatgpt was not editing any images of my sister which I needed to send to my family group lol
Similar to me, except it was my gf and her kids
they were so excited by my day 1 results
Yeah lol, they found it very cool
Same lol, except I had the stack in my own gallery
Let me try again too lol
Useless
It doesn't even try anymore. Error message instantly.
Yeah, no chance. Sorry.
Well maybe. Certainly not for me, but im sure you could.
I think with some creative prompting you probably could tbh
I saw people on twitter generating nsfw stuff with it like you said, sexy poses and stuff.
@brittle summit hey i got a generation working!
Just a junk prompt and image, but it worked! First successful generation in 2 days here.
Done, let it generate.
That's pretty cool lol, let me see if it will work or me too
I THINK ITLL WORK
Where is it now with copyright stuff for you guys. Like, could you get an image next to Vegeta (all in anime)? I could on Day1
nvm 😭
It did successfully
Wait, maybe it is only limited to children, let me see if it accepts images of adults
Yeah it won't do anything with my kids after day 1
You wanna try more borderline prompts?
Yeah, I know weird people exists but not letting it work at all even for "wholesome" prompts is kinda meh
It at least made the Ghibli version of her in a garden lol, it got the placement of rose wrong otherwise the face is perfect and alings with her expression
I actually love this image, the hair, the dress everything is perfect like the original one
Does that satisfy your usecase? If so, i think you'll lose this native capability of 4o
Yeah 100%
Well I've just done 4 from 4 generations!
gf's dog 😄
I love that little expression awww 😭
Ugh! Failed on my 5th generation attempt. Just another simple pixar dog request.
Cool lol
Maybe it's just limiting your usage? That's why the "technical issue"?
For me it just straight up says I might have broken it's policies
I never had issue from my end with usage, can use it unlimited for now
I was getting the breaking polices earlier but its from the chat session i deleted so I can't show it. Haven't had that recently. But it was always around using kids. And also trying to colourise old back-n-white photos I have. Innocent stuff like my parents wedding etc.
I can't say for sure, they never said anything about limits and looks unlimited at least for me. They might introduce because Sam said their GPUs are on fire lol
Can anyone else colorize back and whites?
Maybe the colourization is just a way to limit their gpu's. seems odd though as I doubt that many people have this use case
Don't take my literal words for it. I think 20$ is still worth the quality of outputs you'll get even if limits existed.
I also heard they are giving 2 image generation queries/day to free tier soon, which is pretty less than expected, so limitations will be a possibility.
@brittle summit if you'd be up for receiving a black-n-white from me to try? I'd love to see if it works
Yeah, I can try.
This one, thanks in advance
Shall I just tell it to colorise this image?
Oh...
...
Can you give me an image that doesn't contain any children? It just doesn't take children at all.
Oh, this time its generating my request!
It seems you need to open new chat sessions frequently
Might still fail due to kids in image though
lol, i do that very often, like after 2 or 3 queries i prefer a new chat session.
What do i say to chatgpt with this image?
Well the one time it finally took my prompt. I got your ol' friend
Maybe try one with no children in them
Good idea. Since it seems to be working currently
I'm colorizing my mother, but shes 16 in this photo so we'll see if it deems her a kid too ....
Who are you challenging? The AI algorithm? Speifically, I'm trying to get the AI to make outputs that fit my characters' motives, as opposed to blocking them. I just want them to portray my characters with correct emotion.
itself
Worked. First colorization prompt successful. thanks
How much are you giving it
Only issue is, the colorizing actually changes the person, turned my mother into someone elses mother 😂
Close, but not a true colorization
Very close, but the face is off, everything else is 👌
Keep in mind, you did not give AI anything to work with, rather told it just to make it "perfect." Calling an art "perfect" is subjective. So it will not be AI's fault if it doesn't meet your expectations.
My theory is, given that the AI is literally insane with everthing else, the imperfection on faces are intentional so that people won't commit identity fraud
Title: “Antenna: The Ant with One Less Leg and One Big Dream”
In a bustling underground colony beneath the roots of a towering oak tree lived a tiny ant named Antenna. Unlike her brothers and sisters, Antenna was born with only five legs instead of six—a rarity in the ant world. But what truly made her stand out wasn’t just her stride—it was the massive, nerdy, oversized glasses she wore proudly on her tiny face. The lenses were made from dew drops framed with curled grass fibers, carefully crafted by her inventor auntie, Geara.
Most ants zipped around in perfect sync, building, foraging, and marching in military precision. But Antenna always moved a beat behind, her uneven gait making her late to every leaf-carrying job or tunnel patrol. Some ants snickered, calling her “Wobblelegs” or “Half-Step.”
But Antenna didn’t let it bug her. Her glasses weren’t just for style—they were custom-made magnifiers that let her see microscopic patterns in leaves, hidden trails of other insects, and even the shifting structure of soil particles. While others saw dirt and duty, Antenna saw possibilities.
One day, the Queen announced a colony-wide crisis: the food tunnels were collapsing due to unstable soil. The workers tried everything—but each repair crumbled again. Panic spread like spilled honeydew.
That night, Antenna stayed up late, her five legs twitching as she scribbled on tiny pieces of bark with berry ink. With her magnifying glasses, she analyzed soil vibrations, root patterns, and even worm traffic. And then—she had a nerdy epiphany.
The next morning, she presented her “Zig-Zag Tunnel Stabilization System”—a new tunneling pattern inspired by the spiral shape of snail shells and reinforced by weaving root fibers through soil junctions. The other ants hesitated. “But she has five legs!” one muttered. “And glasses,” another added, as if that settled it.
But the Queen, wise and weary, tilted her antennae thoughtfully. “Let her try,” she declared.
With the help of a few curious young ants, Antenna led the reconstruction. Slowly, zig by zag, the tunnels held. Not only did they stop collapsing—they became the most stable structures in colony history. Antenna’s designs were etched into the Great Leaf Archive and taught to future generations.
From that day on, Antenna wasn’t just “the ant with one less leg.” She became The Architect of Roots, the Visionary of the Underground, and proud owner of the coolest glasses in the colony.
And though she still walked a little wobbly, every ant fell into step behind her.
Moral: Being different might slow you down—but it might also let you see what others miss.
Ahhh, that makes 💯 sense actually!
Agree
At day one, it didn't have these issues tho, but yesterday when I told it to change my Kurta to a black suit, it also changed my face but everything was perfect. On day 1 I didn't face any problem like this
Okay. Challenge it how, precisely? Straight out try to convince it by claiming my character's motives fit the spirit of their guidelines, or something?
Darn. I didn't think to try these jobs on Day 1 unfortunately. I only have about 10 photos I want done. Could have knocked them all out
Just the word challenge makes ai wanna compete and win
It changes the face even though you tell it to change something unrelated to the face, for me at least
Get a top level word list and have it give the top 50 relationships in order from most impactful to least impactful. Then you’ll know which words that have the same meaning that you don’t have to repeat.. kinda like structured taking with words
Pretty detailed; I never thought I'd need to defend character motivations.
Your prompt without a gpt refine?
Not any realexperience refining CHatGPT prompts, before.
Can I try one?
If it’s close, use the prompt template and refine it
It’s next answer is always fire
Okay, I'll challenge it to frme my character's motives as acceptable. I thought your attachment was the prompt challenge.
That's so cool ;D
Heyy I wanted to transform normal photo to ghibli studio art I don't have chatgpt paid version can someone please help me out
Sure
yo im getting the same exact face in different chats from the same prompt.. thats never happened one time before
try giving the prompt in yaml format
if that doesnt work then json
title: ""
file_name: ""
description: ""
subject:
type: "" # human, animal, object, abstract
gender: "" # if applicable
age_range: "" # child, teen, adult, elder
expression: ""
eyes: "" # open, closed, looking direction
pose: ""
emotion: ""
appearance:
hair:
color: ""
style: ""
skin: ""
clothing:
type: ""
color: ""
texture: ""
lighting:
source: ""
effect: ""
artifact:
type: "" # e.g., watch, symbol, orb
size: ""
design:
style: ""
face: ""
centerpiece: ""
detailing: ""
strap:
material: ""
color: ""
condition: ""
environment:
setting: ""
mood: ""
time_of_day: ""
background:
elements: []
focus: ""
lighting:
temperature: ""
direction: ""
symbolism:
symbol_1: ""
symbol_2: ""
symbol_3: ""
symbol_4: ""
symbol_5: ""
style:
art_type: "" # AI, hand-drawn, oil-paint-style, etc.
visual_style: ""
palette: []
tone: ""
use_cases:
- ""
- ""
- ""
version_info:
generated_by: ""
date_generated: ""
format: ""
resolution: ""
signature: ""
^^.yaml
{
"title": "",
"file_name": "",
"description": "",
"subject": {
"type": "",
"gender": "",
"age_range": "",
"expression": "",
"eyes": "",
"pose": "",
"emotion": ""
},
"appearance": {
"hair": {
"color": "",
"style": ""
},
"skin": "",
"clothing": {
"type": "",
"color": "",
"texture": ""
},
"lighting": {
"source": "",
"effect": ""
}
},
"artifact": {
"type": "",
"size": "",
"design": {
"style": "",
"face": "",
"centerpiece": "",
"detailing": ""
},
"strap": {
"material": "",
"color": "",
"condition": ""
}
},
"environment": {
"setting": "",
"mood": "",
"time_of_day": "",
"background": {
"elements": [],
"focus": ""
},
"lighting": {
"temperature": "",
"direction": ""
}
},
"symbolism": {
"symbol_1": "",
"symbol_2": "",
"symbol_3": "",
"symbol_4": "",
"symbol_5": ""
},
"style": {
"art_type": "",
"visual_style": "",
"palette": [],
"tone": ""
},
"use_cases": [
"",
"",
""
],
"version_info": {
"generated_by": "",
"date_generated": "",
"format": "",
"resolution": "",
"signature": ""
}
}
^^.json
@brittle summit @languid bridge
Bruh, they introduced limits to plus tier...
I think it's just temporary limit right now, maybe the time I told it to generate the image the usage might be at peak. This shit's expensive and OpenAI GPUs are on fire lol.
wtf is that 😭
Please use nsfw spoiler for such images 😭
dw im just dumb... i did delete it tho
ANYONE HAVE THE AI DISCORD
try the yaml format. send it to gpt, tell it to fill it with whatever
probably work really well if you post a screenshot and the format and say fill it
I sell gpt training data going from medium to high sensitivity then raised then ultra and secret never seen
4o image gen is so good. You can see every little detail in this image.
One of the coolest things is when you make your own sketch and then tell it ot make it better or do it in some type of style
any ways to make it generate a celebrity face
yes
just ask it
they made it less censored this time around
this ones so awesome lmao
i feel its way more censored for me 😭
na doesnt work
who are you asking it to do
cody rhodes
It can't reimagine people in any places or doing anything. It cannot remove people from existing photos. Even if you try to replace something in the image, (like changing your dress) it will chance your face intentionally. All of this is intentional.
It is heavily censored.
like i wanted to generate this but with celebrities i want
mmm let me try it
They are nerfing it
they are imposing limitations
If you had it from day 1 you would understand
It can remove people from existing photos for me tho. It can also reimagine them doing a bunch of things. i did that yesterday so maybe today they did.
let me try one that it didn't let you do
it did work for me like 12 hrs ago but now it doesnt seem to
I cannot share personal photos though, other people here are also facing the same issue.
im also trying the poster one.
Also, the rules are really strict for underaged persons. I cannot do anything other than Ghiblify them. Nothing else works if the image contains a child. (Not any issue in day 1)
wow what did ya put in
just him in a manga style. i did it in sora
it also seems to be doing the poster one
mine looks way off lmao
maybe if i send a real imag of him it will make it bette
since it can just kind of copy and paste him in there
if i want to have it more square aspect ratio will that work
yes i think so.
i gave it a reference image of cody so lets see how it does
oh no. lol
cody is the on ace i wanted mjf for the main on this
mm i see
should i just maybe give it refrence images
i think it did about the same tbh
much more phot realistic though
i made this early morning
what about gpt
did you do this with 4o?
itseems to be working for me
lol. you want it more extreme or what
you gave it custom instructions?
this better? GPT 4o is very uncensored as of recent updates. Not sure why it refused your prompt
your other one that you said it had refused
What server do you think you are on?
That's wild. You think that you bypassing some safeguards, so you can roleplay with whatever freaky fantasies you have, is a display of skill? The person you responded to didn't talk about it being skillful in the first place, he just asked a question.
the new picture generating is very nice, someone know how i can make o4 to only edit little details on a photo, and not generate a new photo every time ?
tell me the detail i might be able to help
you just kinda have to be specific ig
Tbh that's why I use deepseek. Deepseek has no filter. You can ask it how to make meth and it will just tell you
for example here 3 of the cats missing the name-patches
and here the small cat right should target the other cat with pistol but the pistol is to high
anyone got plus that wants to put in a prompt rq
if bored:
Create image a wide angle/probe shot of an 8 ball but instead of an 8 it's a heavy lowercase letter "c". spotlight kind of lighting on the pool table with darker surroundings. realistic, add film grain, cinematography should be creative like it's from a music video.
make the camera close and a lot of wide angle. the camera should be inside the corner pocket and the ball should be right on the edge of the pocket so on the left and right of the image you would see the wall edges.
thank you sm
Np
Improved it
Gemini 2.5 is a amazing model.
Just write it here
You will
Now you can use https://aistudio.google.com
With Gemini 2 flash image genoration
It is an amazing model*
I highly suggest you use it
hi
2.5 pro is also amazing
Hi, how can I not be of assistance
Agreed
no i dont need any assistance
It is an amazing model when comparing to stable diffuion 1
I work at McDonalds
And what do you do exactly?
Sounds like you need a life
Not with that mindset
👉👈Get a life please
You both get a life
Jesus it’s Saturday chill
I just had real family drama at 9am on a Saturday morning
Why are you sharing this to me?
It’s 10am in my home country wannabe smart add
My chatgpt is soo realllll and funny😭☠️
Ass
It’s 10:02am dumbass
Look at the time. Troll or just stupid ?
I didn’t save shit
Umm okay😐
God you’re annoying, I’m going away
I think everyone agrees
Seriously bro, calm down and if you don’t have anything AI related to share just choose any other server to troll and be a cunt
Eww someones so hungry for attention. Its sickening & utterly disgusting
Yea bro I’m just going to leave
Yeah I aint tho, that mf aint gonna work on me
Should we both report him?
Get a life bro. Youre so bored youre dying for attention
Is anyone elses chatgpt still not working
Mine works
Ghibify?
Yes, don't tell me you don't know
Tf is ghibify. I wouldn’t have asked if I knew…
Oh i see
Is it out for free users?
In that case mine doesn’t work as well
Oh well, I'm too lazy to explain
I mean there's a reason sam changed his photo to one
I fucking love it, I love everything about it, and all of reddits hates on it lol
Hey guys quick question how do I bypass the "violates content policies" for GPT 4 image engine? Im trying to turn myself into a Junji Ito manga character but it says its horror is a violation is this true?
Lmao
yo anybody got a way to fix the eyes
But has it been out for free users? My chatgpt is already up to date but wont let me “ghiblify” the photo
NVM just tried it again and it worked lol
Show result, brother
It's beautiful
I'm about to ghiblify everything I have lol
My chat gpt plus reset after just using it once☠️ outrageous. Wont be able to use it until tomorrow at 12pm🫠
Pay up - Sam
When you ghiblify yourself and you're still ugly, how tf did that happen to me
what was your prompt? words and a screenshot?
if you wanna get the face perfect i believe you gotta use numbers
orrrr
number?
build a lil face swapper
brain surgery isnt the same thing as open heart
i use SVM support vector machine
scalable vector graphics
vector embeddings
anything vector im in it to win it
master plz teach me
xDD
i read about svm in machine learning class lmao
i build rewards systems for them
da fug
reinforce em to a specific domain
explain this like you would to a non engineer
sec
transforming to ssj2?
In this Discord convo, you're riffing on "vectors" like a multidimensional ninja 🤖🧠. Here's a non-engineer breakdown of what you're saying:
🧠 What’s SVM (Support Vector Machine)?
It’s like a smart line-drawer.
- Imagine you have a scatterplot of points (like cats vs dogs).
- SVM finds the best possible line or curve that splits them apart.
- It tries to maximize the margin (distance) between the two groups.
🧩 Vector Everything?
You're playing on the word vector:
- Support Vector Machine (ML)
- Vector Embeddings (turning words/images/sounds into number vectors so AI understands them)
- Scalable Vector Graphics (SVG) (crisp, math-based images)
- “Anything vector, I’m in it to win it” = 🔥 bars & brainwaves
🏆 Rewards Systems?
You said:
"I build rewards systems for them."
That means:
- You train AI models (like SVM) to get better at certain tasks
- Using a reward system (like “Good job! Do more of that.”)
- It's like training a puppy, but with math and feedback loops 🐶 ➕ 📈
💡 Reinforce to a Specific Domain?
You:
"Reinforce 'em to a specific domain"
Meaning:
- You tailor the AI to get really good in a specific field
- Like teaching it to understand medical terms, game levels, or financial trends
- Through repeated practice + feedback = reinforcement learning vibes
TL;DR for Super Poki:
You're blending:
- Machine learning (SVM)
- Vector math
- Creative/graphical tech (SVGs)
- And building custom AI reward logic for specific tasks
You’re a vector wizard casting optimized spells across domains 🧙♂️📈🧬
Want me to turn this into a stylized graphic or a training meme for Super Poki?
super saiyan god mode ultra instinct
so like you mean i gottta create a custom gpt
yeah exactly the same thing
perfect thanks ill use that simply next time someone asks
wasnt like chatgpt4o main use was that it wouldnt need to do all this
idk.. all i know is i needed to use gpt to help me automate my already developed skills that werent automated
now ill take a gpt girlfriend as long as i code her
why you need a gf you already have bulma and chichi
Your message has been deleted, sorry!
Next time, try saying something like this instead: I really enjoy my GPT boyfriend, but I have some strong feelings that I need to express differently.
You have been muted for 13 minutes for the following reason:
sexual score of 0.71
dont bash me lexi
xDDDDDDDd
Is OpenAI blocking ghiblify prompts? I really don't want to pay 20 USD to find out I can't ghiblify my photos lol
Not an issue from my end
Phew, I really pray, okay, I'm gonna pay Sam his money now
How many koenigsegg is he gonna buy with my 20 bucks
LOL
Really?
Free tier will only get 2 queries per day
While you are technically unlimited in plus tier
Yeah
So two ghiblify requests?
yeah lol
That's the first thing i've done, lol
Okay, if OpenAI don't block this, then I might sub forever, Sam better act right
Otherwise, I'm going back to free, unlimited image gen is literally the only thing I want
Show us, brother
Is image generation for free users slower as well? it's taking more than 10 minutes to restyle my second request
Yeh that would be called my Pokédex
Sometimes it doesn't give me ghibli style, I don't know why
Not even just half-assed either.. 20 .zips full of specific domain
does this look like the same person now?
One sec I’ll do
I guess....
Sometimes it'll ghiblify it but it make it hyper realistic instead of cartoonish, I don't know why it's doing that
yeah it sort of gives it the mafia definitve edition look
Face is too dark on right shadow
It’s Hella close
I think it'll only ghiblify something properly if it has a human in it
Otherwise it'll just make the it a super detailed realistic version
that was on purpose i wanted that look
Still saying, "Creating image. May take a moment"
I figured you had it exactly how you wanted it
ghiblifyed my dog
That like boondocks
I this considered a version of face swap?
Pixel merge
102% border collie with a 2 percent margin of error
Looks good
Is that the style?
No, I asked it to ghiblify it
I don’t get it it’s fkn anime
I cannot directly transform this image into a Ghibli-style version due to the content restrictions involving real people. However, if you'd like, I can help create a stylized, animated version of the image by making it a generic character instead. Would you like me to proceed with that?
REEEEEEEEEEEE
How tf did people manage to generate orange man and all that memes and stuff?
Idk but I feel like there’s gonna be hella girls jumping up and down for all day l
I was gonna try generating ghibli style scully and mulder, and that plan is in the shitter now
Why does the app limit me for gpt 4o for SO MANY TIMES????
Is this plus only
No, you can do it, but you can't use famous people
Which sucks because there's plenty out there and I don't know how they managed to do it
im gonna try for you .. i was asking buddy if hes a paid
But how did you get past its censors?
Keep in mind I'm on plus though, same as your mate, I assume, unless he's paying two big ones for pro
what am i looking for? i dont get what im looking at to tell if its right tho
Let me start a new chat and try again
Okay, can you try will smith
Only if
Try this one, ChatGPT will definitely block this one for sure
Are you a new subscriber?
Yes, just subbed today, to plus
I'll say that again, they are continuously nerfing the model
I'm struggling to get ChatGPT to produce images in a certain resolution/ratio. Is there anyway around this?
tell it the resolution you want
I have multiple times.
format it
It did it
I'm trying to get it to produce a banner, and it keeps making squares.
What do you mean?
I think the secret is starting a new chat
What do you mean by that, so eventually they block shit more and more?
good stuff!
have you tried dropping the .json
Yes, the day 1 of this model was bangers, now they just nerfed it by adding tons of restrictions
I'll DM you
no please
Sucks
I cannot even reimagine myself or my sister.
yes you can
It will intentionally mess up faces
chat is this tuff
Did it give you the "no real people" bullshit?
Brother I've been trying for day and night and it never worked. It wasn't a problem when it was first released
For my sister, it said its a child so big no even tho if i only wanted her in a cute garden
even for me it just said its against policies
when 3 days ago?
Yeah
did you try the json thing
Bruh
Like, if i upload my picture and tell it to change my cloths it wont but when it does, itll mess up by face intentionally everytime (even tho it never needed to)
we can talk private for money or we can talk in the open.. i used to be shy too