ChatGPT mein ab tasveer bhi ban sakti hai!
Aajkal toh AI ka zamana hai, aur ismein ek nayi cheez aa gayi hai – ChatGPT ab tasveerein bhi bana sakta hai! OpenAI ne apne GPT-4o model mein image generation capability add kar di hai. Matlab ab aapko sirf text likhna hai, aur ChatGPT usse related tasveer bana dega. Yeh image generator bas decorative nahi hai, balki useful bhi hai. Isse aap accurate text rendering, high prompt adherence, character consistency, aur image editing capability bhi pa sakte hain.
Pehle bhi tha, lekin ab aur better!
Pehle bhi ChatGPT images generate kar sakta tha, lekin woh basic tha. Ab isme character consistency aur text generation kaafi improve ho gaya hai. Company ka kehna hai ki ab yeh language models ki primary capability ban jayegi.
Kya naya hai?
Yeh naya image generator online images aur text ke joint distribution pe train kiya gaya hai. Iska matlab hai ki yeh model samajhta hai ki images language se kaise relate karte hain aur ek dusre se kaise relate karte hain. Isliye, ab character consistency aur bhi better hai, aur aap ek hi character ki multiple images bina kisi problem ke generate kar sakte hain.
Text bhi likh sakta hai!
Iske alawa, yeh images mein text bhi accurately likh sakta hai. Matlab signboards, restaurant menus, aur whiteboard pe likha hua text bhi yeh accurately generate kar sakta hai. Aap ek image input mein de sakte hain, aur chatbot use different styles mein recreate kar sakta hai aur usmein edits bhi kar sakta hai.
Multi-turn generation bhi!
ChatGPT ab multi-turn generation bhi offer karega. Aap chatbot ko prompts dekar generated image mein changes aur additions karne ke liye keh sakte hain, aur woh bina dusre elements ko change kiye output ko refine kar sakta hai. OpenAI ka kehna hai ki yeh model ek image mein 10-20 different objects ko handle kar sakta hai aur unhe accurately add kar sakta hai.
Kis kis ke liye available hai?
Yeh features currently ChatGPT Plus, Team, aur Pro subscribers ke liye available hain. Pehle yeh free tier ke liye bhi available tha, lekin high request volume ki wajah se, free tier ke liye rollout delay kar diya gaya hai.
Safety ka bhi dhyaan!
Safety ke liye, OpenAI saari AI-generated images ke metadata mein C2PA information add kar raha hai taki unhe asli images se alag pehchana ja sake. AI firm ne ek internal search tool bhi banaya hai jisse verify kiya ja sakta hai ki image company ke model se generate hui hai ya nahi. Iske alawa, company harmful content jaise ki child sexual abuse material aur sexual deepfakes wale image requests ko block karta hai. Jab users real people ki images edit karte hain, toh company ne restrictions bhi lagayi hain.
Ghibli style bhi!
Kai users ne social media pe Ghibli-styled recreations of their images aur popular memes share kiye hain jo GPT-4o se generate kiye gaye hain. OpenAI CEO Sam Altman ne bhi apni profile picture X pe Ghibli-style rendition mein change kar di hai.
