Construct an AI-Powered WhatsApp Sticker Generator with Python

Think about sending your individual customized memes or cartoons as a substitute of those from the web. So, remodel your selfies or pictures into enjoyable, stylized stickers utilizing OpenAI’s new GPT-Picture-1 mannequin. On this tutorial, we’ll construct a WhatsApp sticker generator in Python that applies numerous artwork types, together with caricature and Pixar-style filters, to your photographs. 

You’ll learn to arrange the OpenAI picture modifying API, seize or add photographs in Colab, outline humorous and humorous textual content classes, or use your individual textual content and course of three stickers in parallel utilizing a number of API keys for pace. By the top, you’ll have a working sticker maker powered by GPT-Picture-1 and customized textual content prompts.

Why GPT-Picture-1?

We evaluated a number of cutting-edge image-generation fashions, together with Gemini 2.0 Flash, Flux, and Phoenix, on the Leonardo.ai platform. Specifically, all these fashions struggled with rendering textual content and expressions appropriately. As an example:

• Google’s Gemini 2.0 picture API usually produces misspelled or jumbled phrases even when given precise directions. For Instance, with Gemini, the precise textual content seems to be like ‘Massive Sale Immediately!’ and we get outputs like “Massive Sale Todai” or random gibberish. 
• Flux delivers excessive picture high quality typically, however customers report that it “shortly launched little errors” into any textual content it renders. Flux additionally makes tiny spelling errors or garbled letters, particularly because the textual content size will increase. Flux additionally defaults to very related face generations, i.e, “all faces are trying the identical” except closely constrained.
• Phoenix is optimized for constancy and immediate adherence, however like most diffusion fashions, it nonetheless views textual content visually and may introduce errors. We discovered that Phoenix might generate a sticker with the proper wording solely sporadically, and it tended to repeat the identical default face for a given immediate.

Collectively, these limitations led us to develop GPT-Picture-1. In contrast to the above fashions, GPT-Picture-1 incorporates a specialised immediate pipeline that explicitly enforces right textual content and expression adjustments.

Learn extra: Find out how to run the Flux mannequin?

How GPT-Picture-1 Powers Picture Enhancing

GPT-Picture-1 is OpenAI’s flagship multimodal mannequin. It creates and edits photographs from textual content and picture prompts to generate high-quality picture outputs. Basically, we are able to instruct GPT-Picture-1 to use an edit to a supply picture primarily based on a textual content immediate. In our case, we use the pictures. Edit the API endpoint with GPT-Picture-1 to use enjoyable and humorous filtering, and overlay textual content to a photograph enter to create stickers. 

The immediate is rigorously constructed to implement a sticker-friendly output (1024×1024 PNG). Then GPT-Picture-1 primarily turns into the AI-powered sticker creator, the place it is going to change the looks of the topic within the picture and add hilarious textual content.

# Arrange OpenAI shoppers for every API key (to run parallel requests) shoppers = [OpenAI(api_key=key) for key in API_KEYS]

So, for that, we create one OpenAI consumer per API key. With three keys, we are able to make three simultaneous API calls. This multi-key, multi-thread method makes use of ThreadPoolExecutor. It lets us generate 3 stickers in parallel for every run. Because the code prints, it makes use of “3 API keys for SIMULTANEOUS era”, dramatically dashing up the sticker creation..

Step-by-Step Information

The concept of making your individual AI sticker generator could sound advanced, however this information will enable you to simplify your complete course of. You’ll start with the surroundings preparation in Google Colab, then we’ll evaluate the API, perceive classes of phrases, validate textual content, generate completely different inventive types, and at last generate stickers in parallel. Every half is accompanied by code snippets and explanations so you may observe alongside simply. Now, let’s proceed to code:

Putting in and Operating on Colab

To generate stickers, we’ve bought to have the fitting setup! This mission will use Python libraries PIL and rembg for fundamental picture processing, and google-genai will probably be used to be used within the Colab occasion. Step one is the set up the dependencies immediately in your Colab pocket book.

!pip set up --upgrade google-genai pillow rembg !pip set up --upgrade onnxruntime !pip set up python-dotenv

OpenAI Integration and API Keys

After set up, import the modules and arrange API keys. The script creates one OpenAI consumer per API key. This lets the code distribute image-edit requests throughout a number of keys in parallel. The consumer listing is then utilized by the sticker-generation features.

API_KEYS = [ # 3 API keys             "API KEY 1",              "API KEY 2",              "API KEY 3" ] """# Stickerverse """ import os import random import base64 import threading from concurrent.futures import ThreadPoolExecutor, as_completed from openai import OpenAI from PIL import Picture from io import BytesIO from rembg import take away from google.colab import information from IPython.show import show, Javascript from google.colab.output import eval_js import time shoppers = [OpenAI(api_key=key) for key in API_KEYS]

Picture add & digicam seize (logic)

Now the following step is to entry the digicam to seize a photograph or add a picture file. The capture_photo() makes use of JavaScript injected into Colab to open the webcam and return a captured picture.upload_image() makes use of Colab’s file add widget and verifies the uploaded file with PIL.

# Digicam seize by way of JS def capture_photo(filename="picture.jpg", high quality=0.9):     js_code = """     async perform takePhoto(high quality) {         const div = doc.createElement('div');         const video = doc.createElement('video');         const btn = doc.createElement('button');         btn.textContent="📸 Seize";         div.appendChild(video);         div.appendChild(btn);         doc.physique.appendChild(div);         const stream = await navigator.mediaDevices.getUserMedia({video: true});         video.srcObject = stream;         await video.play();         await new Promise(resolve => btn.onclick = resolve);         const canvas = doc.createElement('canvas');         canvas.width = video.videoWidth;         canvas.top = video.videoHeight;         canvas.getContext('second').drawImage(video, 0, 0);         stream.getTracks().forEach(monitor => monitor.cease());         div.take away();         return canvas.toDataURL('picture/jpeg', high quality);     }     """     show(Javascript(js_code))     information = eval_js("takePhoto(%f)" % high quality)     binary = base64.b64decode(information.cut up(',')[1])     with open(filename, 'wb') as f:         f.write(binary)     print(f"Saved: {filename}")     return filename # Picture add perform def upload_image():     print("Please add your picture file...")     uploaded = information.add()     if not uploaded:         print("No file uploaded!")         return None     filename = listing(uploaded.keys())[0]     print(f"Uploaded: {filename}")     # Validate if it is a picture     strive:         img = Picture.open(filename)         img.confirm()         print(f"📸 Picture verified: {img.format} {img.dimension}")         return filename     besides Exception as e:         print(f"Invalid picture file: {str(e)}")         return None # Interactive picture supply choice def select_image_source():     print("Select picture supply:")     print("1. Seize from digicam")     print("2. Add picture file")     whereas True:         strive:             selection = enter("Choose choice (1-2): ").strip()             if selection == "1":                 return "digicam"             elif selection == "2":                 return "add"             else:                 print("Invalid selection! Please enter 1 or 2.")         besides KeyboardInterrupt:             print("nGoodbye!")             return None

Output:

Examples of Classes and Phrases

Now we’ll create our completely different phrase classes to placed on our stickers. Due to this fact, we’ll use a PHRASE_CATEGORIES dictionary that accommodates many classes, resembling company, Bollywood, Hollywood, Tollywood, sports activities, memes, and others. When a class is chosen, the code randomly selects three distinctive phrases for the three sticker types.

PHRASE_CATEGORIES = {     "company": [         "Another meeting? May the force be with you!",         "Monday blues activated!",         "This could have been an email, boss!"     ],     "bollywood": [         "Mogambo khush hua!",         "Kitne aadmi the?",         "Picture abhi baaki hai mere dost!"     ],     "memes": [         "Bhagwan bharose!",         "Main thak gaya hoon!",         "Beta tumse na ho payega!"     ] }

Phrase Classes and Customized Textual content

The generator makes use of a dictionary of phrase classes. The person can both choose a class for random phrase choice or enter their very own customized phrase. There are additionally helper features for interactive choice, in addition to a easy perform to validate the size of a customized phrase.

def select_category_or_custom():     print("nChoose your sticker textual content choice:")     print("1. Choose from phrase class (random choice)")     print("2. Enter my very own customized phrase")     whereas True:         strive:             selection = enter("Select choice (1 or 2): ").strip()             if selection == "1":                 return "class"             elif selection == "2":                 return "customized"             else:                 print("Invalid selection! Please enter 1 or 2.")         besides KeyboardInterrupt:             print("nGoodbye!")             return None # NEW: Perform to get customized phrase from person def get_custom_phrase():     whereas True:         phrase = enter("nEnter your customized sticker textual content (2-50 characters): ").strip()         if len(phrase)  50:             print("Too lengthy! Please hold it below 50 characters.")             proceed         else:             print(f"Customized phrase accepted: '{phrase}'")             return phrase

For customized phrases, enter size is checked (2–50 characters) earlier than acceptance.

Phrase Validation and Spelling Guardrails

def validate_and_correct_spelling(textual content):     spelling_prompt = f"""     Please verify the spelling and grammar of the next textual content and return ONLY the corrected model.     Don't add explanations, feedback, or change the which means.     Textual content to verify: "{textual content}"     """     response = shoppers[0].chat.completions.create(         mannequin="gpt-4o-mini",         messages=[{"role": "user", "content": spelling_prompt}],         max_tokens=100,         temperature=0.1     )     corrected_text = response.decisions[0].message.content material.strip()     return corrected_text

Now we’ll create a pattern build_prompt perform to arrange some basic-level directions for the agent. Additionally observe build_prompt() calls the spelling validator, after which embeds the corrected textual content into the strict sticker immediate:

# Concise Immediate Builder with Spelling Validation def build_prompt(textual content, style_variant):     corrected_text = validate_and_correct_spelling(textual content)     base_prompt = f"""     Create a HIGH-QUALITY WhatsApp sticker in {style_variant} fashion.     OUTPUT:     - 1024x1024 clear PNG with 8px white border     - Topic centered, balanced composition, sharp particulars     - Protect authentic facial identification and proportions     - Match expression to sentiment of textual content: '{corrected_text}'     TEXT:     - Use EXACT textual content: '{corrected_text}' (no adjustments, no emojis)     - Daring comedian font with black define, high-contrast colours     - Place textual content in empty area (high/backside), by no means protecting the face     RULES:     - No hallucinated parts or ornamental glyphs     - No cropping of head/face or textual content     - Keep real looking however expressive look     - Guarantee consistency throughout stickers     """     return base_prompt.strip()

Model Variants: Caricature vs Pixar

The three fashion templates stay in STYLE_VARIANTS.  The primary two are caricature transformations and the final is a Pixar-esque 3D look. These strings will get despatched immediately into the immediate builder and dictate the visible fashion.

STYLE_VARIANTS = [
    "Transform into detailed caricature with slightly exaggerated facial features...",
    "Transform into expressive caricature with enhanced personality features...",
    "Transform into high-quality Pixar-style 3D animated character..."
]

Producing Stickers in Parallel

The actual energy of the mission is the parallel sticker era. The sticker era is completed in parallel with threading all three on the similar time, utilizing separate API keys, so wait instances are dramatically diminished.

# Generate single sticker utilizing OpenAI GPT-image-1 with particular consumer (WITH TIMING) def generate_single_sticker(input_path, output_path, textual content, style_variant, client_idx):     strive:         start_time = time.time()         thread_id = threading.current_thread().identify         print(f"[START] Thread-{thread_id}: API-{client_idx+1} producing {style_variant[:30]}... at {time.strftime('%H:%M:%S', time.localtime(start_time))}")         immediate = build_prompt(textual content, style_variant)         end result = shoppers[client_idx].photographs.edit(             mannequin="gpt-image-1",             picture=[open(input_path, "rb")],             immediate=immediate,             # input_fidelity="excessive"             high quality = 'medium'         )         image_base64 = end result.information[0].b64_json         image_bytes = base64.b64decode(image_base64)         with open(output_path, "wb") as f:             f.write(image_bytes)         end_time = time.time()         period = end_time - start_time         style_type = "Caricature" if "caricature" in style_variant.decrease() else "Pixar"         print(f"[DONE] Thread-{thread_id}: {style_type} saved as {output_path} | Period: {period:.2f}s | Textual content: '{textual content[:30]}...'")         return True     besides Exception as e:         print(f"[ERROR] API-{client_idx+1} failed: {str(e)}")         return False # NEW: Create stickers with customized phrase (all 3 types use the identical customized textual content) def create_custom_stickers_parallel(photo_file, custom_text):     print(f"nCreating 3 stickers together with your customized phrase: '{custom_text}'")     print("   • Model 1: Caricature #1")     print("   • Model 2: Caricature #2")     print("   • Model 3: Pixar Animation")     # Map futures to their information     tasks_info = {}     with ThreadPoolExecutor(max_workers=3, thread_name_prefix="CustomSticker") as executor:         start_time = time.time()         print(f"n[PARALLEL START] Submitting 3 API calls SIMULTANEOUSLY at {time.strftime('%H:%M:%S', time.localtime(start_time))}")         # Submit ALL duties without delay (non-blocking) - all utilizing the identical customized textual content         for idx, style_variant in enumerate(STYLE_VARIANTS):             output_name = f"custom_sticker_{idx+1}.png"             future = executor.submit(generate_single_sticker, photo_file, output_name, custom_text, style_variant, idx)             tasks_info[future] = {                 'output_name': output_name,                 'textual content': custom_text,                 'style_variant': style_variant,                 'client_idx': idx,                 'submit_time': time.time()             }         print("All 3 API requests submitted! Processing as they full...")         accomplished = 0         completion_times = []         # Course of outcomes as they full         for future in as_completed(tasks_info.keys(), timeout=180):             strive:                 success = future.end result()                 task_info = tasks_info[future]                 if success:                     accomplished += 1                     completion_time = time.time()                     completion_times.append(completion_time)                     period = completion_time - task_info['submit_time']                     style_type = "Caricature" if "caricature" in task_info['style_variant'].decrease() else "Pixar"                     print(f"[{completed}/3] {style_type} accomplished: {task_info['output_name']} "                           f"(API-{task_info['client_idx']+1}, {period:.1f}s)")                 else:                     print(f"Failed: {task_info['output_name']}")             besides Exception as e:                 task_info = tasks_info[future]                 print(f"Error with {task_info['output_name']} (API-{task_info['client_idx']+1}): {str(e)}")         total_time = time.time() - start_time         print(f"n [FINAL RESULT] {accomplished}/3 customized stickers accomplished in {total_time:.1f} seconds!") # UPDATED: Create 3 stickers in  PARALLEL (utilizing as_completed) def create_category_stickers_parallel(photo_file, class):     if class not in PHRASE_CATEGORIES:         print(f" Class '{class}' not discovered! Out there: {listing(PHRASE_CATEGORIES.keys())}")         return     # Select 3 distinctive phrases for 3 stickers     chosen_phrases = random.pattern(PHRASE_CATEGORIESBeginner, 3)     print(f" Chosen phrases for {class.title()} class:")     for i, phrase in enumerate(chosen_phrases, 1):         style_type = "Caricature" if i  1:             fastest_completion = min(completion_times) - start_time             print(f"Parallel effectivity: Quickest completion in {fastest_completion:.1f}s")

Right here, generate_single_sticker() builds the immediate and calls the pictures. edit endpoint utilizing the required client_idx. The parallel features create a ThreadPoolExecutor with max_workers=3, submit the three duties, and course of outcomes with as_completed. This lets the script log every completed sticker shortly. Furthermore, we are able to additionally view the logs to see what is going on for every thread (time, what was it caricature or Pixar fashion).

Principal execution block 

On the backside of the script, the __main__ guard defaults to working sticker_from_camera(). Nonetheless, you may agree/uncomment as desired to run interactive_menu(), create_all_category_stickers() or different features.

# Principal execution if __name__ == "__main__":     sticker_from_camera()

Output:

Output Picture:

For the entire model of this WhatsApp sticker generator code, go to this GitHub repository.

Conclusion

On this tutorial, we’ve walked via establishing GPT-Picture-1 calls, developing an prolonged immediate for stickers, capturing or importing photographs, choosing amusing phrases or customized textual content, and working 3 fashion variants concurrently. In only a few hundred strains of code, this mission converts your photos into some comic-styled stickers.

By merely combining OpenAI’s imaginative and prescient mannequin with some artistic immediate engineering and multi-threading, you may generate enjoyable, customized stickers in seconds. And the end result will probably be an AI-based WhatsApp sticker generator that may produce immediately shareable stickers with a single click on to any of your folks and teams. Now strive it to your personal picture and your favourite joke!

Ceaselessly Requested Questions