Python
CGPT SDK Docs Python
ChainGPT AI NFT Generator SDK Documentation
The ChainGPT AI NFT Generator SDK provides a comprehensive suite of tools to create, manage, and prepare NFTs for minting. This includes enhancing prompts, generating images (synchronously and asynchronously), tracking generation progress, retrieving minting metadata, and querying supported blockchains and contract ABIs.
Table of Contents
Installation
Quick Start
SDK Initialization
NFT Service Overview
NFT Service Methods
enhance_prompt()
generate_image()
generate_nft_queue()
get_progress()
mint_nft_metadata()
get_chains()
get_abi()
Data Models & Types
Enums Reference
Error Handling
Complete Example
Best Practices
New Features
Installation
Install the ChainGPT SDK via pip:
pip install chaingpt
Or add to your requirements.txt
:
chaingpt>=1.1.4
For development with environment variables:
pip install python-dotenv # Optional for .env file support
Quick Start
Here's a minimal example to get you started:
import asyncio
from chaingpt.client import ChainGPTClient
from chaingpt.models import GenerateImageRequestModel, NFTImageModel
async def quick_example():
client = ChainGPTClient(api_key="YOUR_API_KEY")
# Generate an image synchronously
request = GenerateImageRequestModel(
prompt="a dragon in space",
model=NFTImageModel.VELOGEN,
height=512,
width=512
)
response = await client.nft.generate_image(request)
# Save the image
with open("dragon.jpg", "wb") as f:
f.write(bytes(response.data.data))
await client.close()
if __name__ == "__main__":
asyncio.run(quick_example())
SDK Initialization
Basic Initialization
import asyncio
from chaingpt.client import ChainGPTClient
from chaingpt.exceptions import ChainGPTError
# Basic initialization
client = ChainGPTClient(api_key="YOUR_API_KEY")
# With environment variable
import os
API_KEY = os.getenv("CHAINGPT_API_KEY")
client = ChainGPTClient(api_key=API_KEY)
Required Imports
from chaingpt.client import ChainGPTClient
from chaingpt.models import (
EnhancePromptRequestModel,
GenerateImageRequestModel,
GenerateNFTQueueRequestModel,
MintNFTRequestModel,
TraitModel,
TraitValueItemModel,
NFTImageModel,
ImageEnhanceOption
)
from chaingpt.exceptions import ChainGPTError, APIError
NFT Service Overview
Access the NFT service through the client:
nft_service = client.nft
The NFT service provides seven main methods:
enhance_prompt(): Improve prompts for better image generation
generate_image(): Create images synchronously (fast preview)
generate_nft_queue(): Queue asynchronous NFT generation (high quality)
get_progress(): Track queued generation progress
mint_nft_metadata(): Generate metadata for blockchain minting
get_chains(): List supported blockchain networks
get_abi(): Get smart contract ABI
NFT Service Methods
enhance_prompt()
Enhances or refines a text prompt for better image generation results.
Method Signature:
async def enhance_prompt(self, request_data: EnhancePromptRequestModel) -> EnhancePromptResponseModel
Parameters:
request_data
EnhancePromptRequestModel
Yes
Request containing the prompt to enhance
Request Model (EnhancePromptRequestModel
):
prompt
str
Yes
The text prompt you want to enhance
Response Model (EnhancePromptResponseModel
):
statusCode
int
HTTP status code (200 for success)
message
str
Response message
data.enhancedPrompt
str
The enhanced version of your prompt
Example:
async def enhance_example():
enhance_request = EnhancePromptRequestModel(prompt="a dragon in space")
response = await client.nft.enhance_prompt(enhance_request)
if response.statusCode == 200:
print(f"Original: a dragon in space")
print(f"Enhanced: {response.data.enhancedPrompt}")
# Output might be: "A majestic dragon soaring through the cosmic void,
# surrounded by stars and nebulae, highly detailed, fantasy art"
generate_image()
Generates an image synchronously based on the given prompt and parameters. Best for quick previews and testing.
Method Signature:
async def generate_image(self, request_data: GenerateImageRequestModel) -> GenerateImageResponseModel
Parameters:
request_data
GenerateImageRequestModel
Yes
Image generation parameters
Request Model (GenerateImageRequestModel
):
prompt
str
Yes
-
Text prompt for image generation
model
NFTImageModel
Yes
-
AI model to use (see Enums)
height
int
Yes
-
Image height in pixels
width
int
Yes
-
Image width in pixels
steps
int
No
Model default
Number of refinement passes for image generation. Higher values can yield more detailed images at the cost of longer generation time. Each model supports a range: for example, velogen supports 1–4 steps (default ~2), while NebulaForge XL and VisionaryForge support up to 50 (default ~25). If not provided, a model-specific default is used.
enhance
ImageEnhanceOption
No
ORIGINAL
Image enhancement level
style
str
No
None
Artistic style (e.g., "fantasy_art", "cyberpunk")
traits
List[TraitModel]
No
None
List of traits to apply
image
HttpUrl
No
None
Reference image URL for image-to-image generation
isCharacterPreserve
bool
No
False
Preserve character features from reference image
Response Model (GenerateImageResponseModel
):
data.type
str
Data type (always "Buffer")
data.data
List[int]
Raw image bytes as integer array
Image-to-Image Generation:
from pydantic import HttpUrl
# Generate image based on a reference image
request = GenerateImageRequestModel(
prompt="a dragon in space",
model=NFTImageModel.VELOGEN,
height=512,
width=512,
image=HttpUrl("https://example.com/reference-image.jpg"),
isCharacterPreserve=True # Preserve character features from reference
)
response = await client.nft.generate_image(request)
Example:
async def generate_image_example():
request = GenerateImageRequestModel(
prompt="a majestic cyberpunk dragon",
model=NFTImageModel.VELOGEN, # Fast model for sync generation
height=512,
width=512,
steps=3,
enhance=ImageEnhanceOption.ENHANCE_1X,
style="cyberpunk"
)
response = await client.nft.generate_image(request)
# Save the image
image_bytes = bytes(response.data.data)
with open("generated_image.png", "wb") as f:
f.write(image_bytes)
print(f"Image generated: {len(image_bytes)} bytes")
Image-to-Image Generation:
from pydantic import HttpUrl
# Generate image based on a reference image
request = GenerateImageRequestModel(
prompt="a dragon in space",
model=NFTImageModel.VELOGEN,
height=512,
width=512,
image=HttpUrl("https://example.com/reference-image.jpg"),
isCharacterPreserve=True # Preserve character features
)
response = await client.nft.generate_image(request)
Using Traits:
from chaingpt.models import TraitModel, TraitValueItemModel
traits = [
TraitModel(
trait_type="Background",
value=[
TraitValueItemModel(value="night_sky", ratio=70),
TraitValueItemModel(value="lava_pit", ratio=30)
]
),
TraitModel(
trait_type="Color",
value=[
TraitValueItemModel(value="blue", ratio=50),
TraitValueItemModel(value="red", ratio=50)
]
)
]
request = GenerateImageRequestModel(
prompt="a dragon",
model=NFTImageModel.VELOGEN,
height=512,
width=512,
traits=traits
)
generate_nft_queue()
Initiates an asynchronous NFT image generation job. Use this for high-quality images or when you don't need immediate results.
Method Signature:
async def generate_nft_queue(self, request_data: GenerateNFTQueueRequestModel) -> GenerateNFTQueueResponseModel
Parameters:
request_data
GenerateNFTQueueRequestModel
Yes
NFT generation parameters
Request Model (GenerateNFTQueueRequestModel
):
walletAddress
str
Yes
Blockchain wallet address for the NFT
prompt
str
Yes
Text prompt for image generation
model
NFTImageModel
Yes
AI model to use
height
int
Yes
Image height in pixels
width
int
Yes
Image width in pixels
chainId
int
Yes
Blockchain network ID (see get_chains())
amount
int
Yes
Number of NFTs to generate in this batch
steps
int
No
Generation steps
enhance
ImageEnhanceOption
No
Enhancement level
style
str
No
Artistic style
traits
List[TraitModel]
No
List of traits
image
HttpUrl
No
Reference image URL for image-to-image generation
isCharacterPreserve
bool
No
Preserve character features from reference image
Response Model (GenerateNFTQueueResponseModel
):
statusCode
int
HTTP status code (201 = successfully queued)
message
str
Response message
data.collectionId
str
Unique ID for tracking this generation job
Example:
async def generate_nft_queue_example():
request = GenerateNFTQueueRequestModel(
walletAddress="0x742d35Cc6635C0532925a3b8D8842A36fF20fd40",
prompt="A mystical phoenix rising from cosmic ashes",
model=NFTImageModel.NEBULA_FORGE_XL, # High-quality model
height=1024,
width=1024,
steps=30,
enhance=ImageEnhanceOption.ENHANCE_1X,
chainId=56, # BSC Mainnet
amount=1,
style="cinematic"
)
response = await client.nft.generate_nft_queue(request)
if response.statusCode == 201:
print(f"NFT queued successfully!")
print(f"Collection ID: {response.data.collectionId}")
return response.data.collectionId
else:
print(f"Failed to queue: {response.message}")
return None
get_progress()
Checks the status and progress of a queued NFT generation job.
Method Signature:
async def get_progress(self, collection_id: str) -> NFTProgressResponseModel
Parameters:
collection_id
str
Yes
Collection ID from generate_nft_queue()
response
Response Model (NFTProgressResponseModel
):
statusCode
int
HTTP status code
message
str
Response message
data.generated
bool
True
if generation is complete
data.images
List[str]
IPFS URLs (only when generated=True
)
data.progress
int
Progress percentage (only when processing)
Example:
async def track_progress_example(collection_id: str):
max_attempts = 20
attempt = 0
while attempt < max_attempts:
response = await client.nft.get_progress(collection_id)
print(f"Status: {response.data.generated}")
if response.data.progress:
print(f"Progress: {response.data.progress}%")
if response.data.generated:
print("Generation completed!")
if response.data.images:
print(f"IPFS URLs: {response.data.images}")
return True
await asyncio.sleep(15) # Wait 15 seconds
attempt += 1
print("Timeout reached")
return False
mint_nft_metadata()
Generates the metadata required for minting NFTs on a blockchain. Call this after NFT generation is complete.
Method Signature:
async def mint_nft_metadata(self, request_data: MintNFTRequestModel) -> MintNFTResponseModel
Parameters:
request_data
MintNFTRequestModel
Yes
Metadata generation parameters
Request Model (MintNFTRequestModel
):
collectionId
str
Yes
ID of completed NFT generation job
name
str
Yes
Name of the NFT
description
str
Yes
Description of the NFT
symbol
str
Yes
Symbol for the NFT collection
ids
List[int]
Yes
List of NFT IDs to mint (e.g., [1]
for single NFT)
Response Model (MintNFTResponseModel
):
statusCode
int
HTTP status code
message
str
Response message
data._id
str
MongoDB document ID
data._id
str
MongoDB document ID
data.name
str
NFT name
data.description
str
NFT description
data.image
str
IPFS URL for the NFT image
data.attributes
List[Dict]
NFT attributes/traits
data.collectionId
str
Collection ID
data.transaction
Any
Transaction details (if applicable)
data.__v
int
MongoDB version field
data.createdAt
str
Creation timestamp
data.updatedAt
str
Last update timestamp
data.__v
int
MongoDB version field
data.createdAt
str
Creation timestamp
data.updatedAt
str
Last update timestamp
Example:
async def mint_metadata_example(collection_id: str):
request = MintNFTRequestModel(
collectionId=collection_id,
name="Cosmic Phoenix #1",
description="A stunning AI-generated phoenix from the ChainGPT NFT Forge.",
symbol="CPHX",
ids=[1] # Single NFT
)
response = await client.nft.mint_nft_metadata(request)
if response.statusCode in [200, 201]:
print("Metadata generated successfully:")
print(f" ID: {response.data._id}")
print(f" Name: {response.data.name or 'Not provided'}")
print(f" Image: {response.data.image or 'Not provided'}")
print(f" Attributes: {response.data.attributes or 'None'}")
print(f" Collection ID: {response.data.collectionId or 'Not provided'}")
print(f" Created At: {response.data.createdAt or 'Not provided'}")
# This metadata can now be used with smart contracts
return response.data
else:
print(f"Failed to generate metadata: {response.message}")
return None
get_chains()
Retrieves the list of supported blockchain networks for NFT operations.
Method Signature:
async def get_chains(self, test_net: Optional[bool] = None) -> GetChainsResponseModel
Parameters:
test_net
bool
No
None
True
for testnets, False
for mainnets, None
for all
Response Model (GetChainsResponseModel
):
statusCode
int
HTTP status code
message
str
Response message
data.chains
List[ChainInfoModel]
List of supported chains
Chain Info Model (ChainInfoModel
):
chainId
int
Numerical chain ID
chainName
str
Chain name (e.g., "Ethereum Mainnet")
network
str
Network type ("mainnet", "testnet")
networkType
str
Additional network classification
Example:
async def get_chains_example():
# Get mainnet chains
mainnet_response = await client.nft.get_chains(test_net=False)
print("Mainnet Chains:")
for chain in mainnet_response.data.chains[:5]:
print(f" ID: {chain.chainId}, Name: {chain.chainName}")
# Get testnet chains
testnet_response = await client.nft.get_chains(test_net=True)
print("\nTestnet Chains:")
for chain in testnet_response.data.chains[:5]:
print(f" ID: {chain.chainId}, Name: {chain.chainName}")
# Common chain IDs:
# 1 = Ethereum Mainnet
# 56 = BSC Mainnet
# 137 = Polygon Mainnet
# 97 = BSC Testnet
# 11155111 = Ethereum Sepolia
get_abi()
Provides the ABI (Application Binary Interface) of the ChainGPT NFT smart contract for direct blockchain interaction.
Method Signature:
async def get_abi(self) -> ContractABIResponseModel
Parameters: None
Response Model (ContractABIResponseModel
):
statusCode
int
HTTP status code
message
str
Response message
data
List[Dict]
Contract ABI as JSON objects
Example:
async def get_abi_example():
response = await client.nft.get_abi()
if response.statusCode == 200:
abi = response.data
print(f"Contract ABI contains {len(abi)} functions/events")
# Use with web3.py
from web3 import Web3
w3 = Web3(Web3.HTTPProvider("YOUR_RPC_URL"))
contract = w3.eth.contract(
address="CONTRACT_ADDRESS",
abi=abi
)
return abi
else:
print(f"Failed to get ABI: {response.message}")
return None
Data Models & Types
TraitModel
Defines NFT traits/attributes for generation:
class TraitModel(BaseModel):
trait_type: str # Name of the trait (e.g., "Background", "Color")
value: List[TraitValueItemModel] # Possible values with ratios
class TraitValueItemModel(BaseModel):
value: str # Specific trait value (e.g., "night_sky")
ratio: int # Rarity/probability (higher = more common)
Error Models
All methods can raise the following exceptions:
from chaingpt.exceptions import (
ChainGPTError, # Base exception
APIError, # API-specific errors
AuthenticationError, # Invalid API key (401)
ValidationError, # Bad request (400)
InsufficientCreditsError, # No credits (402/403)
RateLimitError, # Too many requests (429)
NotFoundError, # Endpoint not found (404)
ServerError, # Server errors (5xx)
TimeoutError # Request timeout
)
Enums Reference
NFTImageModel
Available AI models for image generation:
VELOGEN
"velogen"
Fast generation
Quick previews, testing
NEBULA_FORGE_XL
"nebula_forge_xl"
High quality
Final NFTs, detailed art
VISIONARY_FORGE
"VisionaryForge"
Balanced
General purpose
DALE3
"Dale3"
Alternative model
Experimental
ImageEnhanceOption
Image enhancement levels:
ORIGINAL
"original"
No enhancement
ENHANCE_1X
"1x"
1x enhancement
ENHANCE_2X
"2x"
2x enhancement
Error Handling
Basic Error Handling
from chaingpt.exceptions import ChainGPTError, APIError
async def safe_nft_generation():
try:
request = GenerateImageRequestModel(
prompt="a dragon",
model=NFTImageModel.VELOGEN,
height=512,
width=512
)
response = await client.nft.generate_image(request)
return response
except AuthenticationError:
print("Invalid API key!")
except ValidationError as e:
print(f"Invalid request: {e.message}")
if e.field:
print(f"Problem with field: {e.field}")
except InsufficientCreditsError:
print("Not enough credits in your account")
except RateLimitError as e:
print(f"Rate limited. Retry after: {e.retry_after} seconds")
except APIError as e:
print(f"API error {e.status_code}: {e.message}")
except ChainGPTError as e:
print(f"General error: {e.message}")
Specific Error Scenarios
# Handle generation failures
async def robust_generation(collection_id: str):
try:
while True:
progress = await client.nft.get_progress(collection_id)
if progress.data.generated:
return progress.data.images
elif progress.statusCode >= 400:
raise ChainGPTError("Generation failed")
await asyncio.sleep(10)
except ChainGPTError as e:
print(f"Generation error: {e}")
return None
Complete Example
Here's a comprehensive example showing the full NFT creation workflow:
import asyncio
import os
from chaingpt.client import ChainGPTClient
from chaingpt.models import *
from chaingpt.exceptions import ChainGPTError
async def complete_nft_workflow():
"""Complete NFT generation workflow from prompt to metadata."""
# Initialize client
client = ChainGPTClient(api_key=os.getenv("CHAINGPT_API_KEY"))
try:
# Step 1: Enhance the prompt
print("1. Enhancing prompt...")
enhance_request = EnhancePromptRequestModel(
prompt="a mystical dragon guardian"
)
enhanced = await client.nft.enhance_prompt(enhance_request)
enhanced_prompt = enhanced.data.enhancedPrompt
print(f"Enhanced: {enhanced_prompt}")
# Step 2: Generate a quick preview
print("\n2. Generating preview...")
preview_request = GenerateImageRequestModel(
prompt=enhanced_prompt,
model=NFTImageModel.VELOGEN,
height=512,
width=512,
enhance=ImageEnhanceOption.ORIGINAL
)
preview = await client.nft.generate_image(preview_request)
# Save preview
with open("preview.jpg", "wb") as f:
f.write(bytes(preview.data.data))
print("Preview saved as preview.jpg")
# Step 3: Queue high-quality NFT generation
print("\n3. Queueing NFT generation...")
nft_request = GenerateNFTQueueRequestModel(
walletAddress="0x742d35Cc6635C0532925a3b8D8842A36fF20fd40",
prompt=enhanced_prompt,
model=NFTImageModel.NEBULA_FORGE_XL,
height=1024,
width=1024,
steps=30,
enhance=ImageEnhanceOption.ENHANCE_1X,
chainId=56, # BSC Mainnet
amount=1,
style="fantasy-art",
traits=[
TraitModel(
trait_type="Element",
value=[
TraitValueItemModel(value="fire", ratio=60),
TraitValueItemModel(value="ice", ratio=40)
]
)
]
)
queue_response = await client.nft.generate_nft_queue(nft_request)
collection_id = queue_response.data.collectionId
print(f"Queued with ID: {collection_id}")
# Step 4: Track progress
print("\n4. Tracking progress...")
max_attempts = 20
for attempt in range(max_attempts):
progress = await client.nft.get_progress(collection_id)
print(f"Attempt {attempt + 1}: Generated: {progress.data.generated} - Progress: {progress.data.progress}%")
if progress.data.generated:
print(f"Completed! IPFS URLs: {progress.data.images}")
break
await asyncio.sleep(15)
else:
raise ChainGPTError("Generation timeout")
# Step 5: Generate metadata
print("\n5. Generating metadata...")
metadata_request = MintNFTRequestModel(
collectionId=collection_id,
name="Mystical Dragon Guardian #1",
description="An AI-generated mystical dragon guardian, protector of ancient realms.",
symbol="MDG",
ids=[1]
)
metadata = await client.nft.mint_nft_metadata(metadata_request)
print("Metadata generated:")
print(f" ID: {metadata.data._id}")
print(f" Name: {metadata.data.name or 'Not provided'}")
print(f" Image: {metadata.data.image or 'Not provided'}")
print(f" Attributes: {metadata.data.attributes or 'None'}")
# Step 6: Get chain info for minting
print("\n6. Getting chain information...")
chains = await client.nft.get_chains(test_net=False)
bsc_mainnet = next((c for c in chains.data.chains if c.chainId == 56), None)
if bsc_mainnet:
print(f"Target chain: {bsc_mainnet.chainName} (ID: {bsc_mainnet.chainId})")
# Step 7: Get contract ABI
print("\n7. Getting contract ABI...")
abi_response = await client.nft.get_abi()
print(f"Contract ABI received (first few elements): {abi_response.data[:2]}")
print("\n✅ NFT workflow completed successfully!")
return metadata.data
except ChainGPTError as e:
print(f"❌ Error: {e}")
return None
finally:
await client.close()
# Run the complete workflow
if __name__ == "__main__":
result = asyncio.run(complete_nft_workflow())
Best Practices
1. Resource Management
Always close the client when done:
async def proper_usage():
client = ChainGPTClient(api_key="YOUR_API_KEY")
try:
# Your NFT operations here
pass
finally:
await client.close()
# Or use async context manager if available
async with ChainGPTClient(api_key="YOUR_API_KEY") as client:
# Operations here
pass
2. Error Handling
Handle specific errors appropriately:
async def robust_operation():
try:
# Your operation
pass
except AuthenticationError:
# Handle auth issues - check API key
pass
except InsufficientCreditsError:
# Handle billing issues
pass
except RateLimitError as e:
# Handle rate limiting - wait and retry
await asyncio.sleep(e.retry_after or 60)
except ValidationError as e:
# Handle validation errors - fix request
print(f"Fix field: {e.field}")
3. Progress Tracking
Use reasonable polling intervals:
async def track_with_backoff(collection_id: str):
wait_time = 10 # Start with 10 seconds
max_wait = 60 # Max 60 seconds between polls
while True:
progress = await client.nft.get_progress(collection_id)
if progress.data.generated:
return progress.data.images
await asyncio.sleep(wait_time)
wait_time = min(wait_time * 1.2, max_wait) # Exponential backoff
4. Model Selection
Choose the right model for your use case:
# For quick previews and testing
preview_model = NFTImageModel.VELOGEN
# For final high-quality NFTs
production_model = NFTImageModel.NEBULA_FORGE_XL
# Adjust steps based on model and quality needs
quick_steps = 10 # For previews
quality_steps = 30 # For final NFTs
5. Prompt Engineering
Create effective prompts:
# Good prompt structure
prompt = "a majestic dragon, fantasy art, highly detailed, vibrant colors, mystical atmosphere"
# Use enhancement for better results
enhanced = await client.nft.enhance_prompt(
EnhancePromptRequestModel(prompt=prompt)
)
final_prompt = enhanced.data.enhancedPrompt
New Features
Image-to-Image Generation
The NFT service now supports image-to-image generation, allowing you to create variations based on reference images:
from pydantic import HttpUrl
# Generate image based on a reference image
request = GenerateImageRequestModel(
prompt="a dragon in space",
model=NFTImageModel.VELOGEN,
height=512,
width=512,
image=HttpUrl("https://example.com/reference-image.jpg"),
isCharacterPreserve=True # Preserve character features from reference
)
response = await client.nft.generate_image(request)
Character Preservation
When using image-to-image generation, you can preserve character features from the reference image:
# Preserve character features
request = GenerateImageRequestModel(
prompt="a dragon in a cyberpunk city",
model=NFTImageModel.NEBULA_FORGE_XL,
height=1024,
width=1024,
image=HttpUrl("https://example.com/dragon-reference.jpg"),
isCharacterPreserve=True # Maintains dragon's appearance
)
Enhanced Traits System
The traits system has been enhanced with better support for NFT collections:
traits = [
TraitModel(
trait_type="Background",
value=[
TraitValueItemModel(value="night_sky", ratio=70),
TraitValueItemModel(value="lava_pit", ratio=30)
]
),
TraitModel(
trait_type="Accessory",
value=[
TraitValueItemModel(value="crown", ratio=20),
TraitValueItemModel(value="sword", ratio=40),
TraitValueItemModel(value="none", ratio=40)
]
)
]
Updated Chain Support
The chain information now includes more detailed network information:
chains = await client.nft.get_chains(test_net=False)
for chain in chains.data.chains:
print(f"Chain: {chain.chainName}")
print(f" ID: {chain.chainId}")
print(f" Network: {chain.network}")
print(f" Type: {chain.networkType}")
Improved Response Structures
All API responses now follow a consistent structure with proper error handling:
# All responses include statusCode, message, and data
response = await client.nft.enhance_prompt(request)
if response.statusCode == 200:
enhanced_prompt = response.data.enhancedPrompt
else:
print(f"Error: {response.message}")
This ensures better error handling and consistent data access patterns across all NFT service methods.
Last updated
Was this helpful?