add doubao ark api node for seedance i2v and seedream t/i2i
This commit is contained in:
AhBumm
@@ -7,6 +7,7 @@ from .nodes4tuzi import (
|
||||
LoadVideoFromUrlComfyIO,
|
||||
)
|
||||
from .nodes4hypr import HyprLab_Image_API_Node
|
||||
from .nodes4doubao import seedance_api_node, seedream_api_node
|
||||
|
||||
# Exporting the node classes for ComfyUI to discover
|
||||
NODE_CLASS_MAPPINGS = {
|
||||
@@ -33,6 +34,8 @@ NODE_CLASS_MAPPINGS = {
|
||||
"load_video_from_url": LoadVideoFromUrlVHS,
|
||||
"load_video_from_url_comfy_core": LoadVideoFromUrlComfyIO,
|
||||
"hyprlab_image_api_node": HyprLab_Image_API_Node,
|
||||
"doubao_seedance_api_node": seedance_api_node,
|
||||
"doubao_seedream_api_node": seedream_api_node,
|
||||
}
|
||||
|
||||
NODE_DISPLAY_NAME_MAPPINGS = {
|
||||
@@ -59,4 +62,6 @@ NODE_DISPLAY_NAME_MAPPINGS = {
|
||||
"BillBum_Modified_GPTImage1_API_Node": "Custom GPTImage1 API Node",
|
||||
"BillBum_Modified_Flux_API_with_imgInput": "Custom Flux API Node",
|
||||
"hyprlab_image_api_node": "HyprLab ImageGen API Node",
|
||||
"doubao_seedance_api_node": "Doubao Seedance VideoGen API Node",
|
||||
"doubao_seedream_api_node": "Doubao Seedream ImageGen API Node",
|
||||
}
|
||||
+327
@@ -0,0 +1,327 @@
|
||||
import io
|
||||
from PIL import Image
|
||||
import numpy as np
|
||||
import torch
|
||||
import requests
|
||||
import base64
|
||||
import time
|
||||
import json
|
||||
import tenacity
|
||||
import math
|
||||
from comfy.utils import common_upscale
|
||||
|
||||
|
||||
## DataType Conversion Functions
|
||||
def tensor2pil(image):
|
||||
return Image.fromarray(np.clip(255. * image.cpu().numpy().squeeze(), 0, 255).astype(np.uint8))
|
||||
|
||||
def pil2tensor(image):
|
||||
return torch.from_numpy(np.array(image).astype(np.float32) / 255.0).unsqueeze(0)
|
||||
|
||||
def tensor2ndarray(image):
|
||||
return np.clip(255. * image.cpu().numpy().squeeze(), 0, 255).astype(np.uint8)
|
||||
|
||||
def ndarray2tensor(image):
|
||||
return torch.from_numpy(image.astype(np.float32) / 255.0).unsqueeze(0)
|
||||
|
||||
|
||||
## Node Classes
|
||||
class seedance_api_node:
|
||||
|
||||
def __init__(self):
|
||||
pass
|
||||
|
||||
@classmethod
|
||||
def INPUT_TYPES(s):
|
||||
return {
|
||||
"required": {
|
||||
"model": ("STRING", {"default": "doubao-seedance-1-5-pro-251215"}),
|
||||
"prompt": ("STRING", {"forceInput": True}),
|
||||
"seed": ("INT", {"default": -1, "min": -1, "max": 0xffffffff}),
|
||||
"api_url": ("STRING", {"default": "https://ark.cn-beijing.volces.com/api/v3/contents/generations/tasks"}),
|
||||
"api_key": ("STRING", {"default": "Input_your_API_key_here..."}),
|
||||
"resolution": (["480p", "720p", "1080p"], {"default":"480p"}),
|
||||
"ratio": (["16:9", "4:3", "1:1", "3:4", "9:16", "21:9", "adaptive"], {"default":"adaptive"}),
|
||||
"duration": ("INT", {"default":5, "min":1, "max":12, "step":1}),
|
||||
"camerafixed": (["true", "false"], {"default":"false"}),
|
||||
"watermark": (["true", "false"], {"default":"false"})
|
||||
},
|
||||
"optional": {
|
||||
"first_frame": ("IMAGE",),
|
||||
"last_frame": ("IMAGE",),
|
||||
}
|
||||
}
|
||||
|
||||
RETURN_TYPES = ("STRING",)
|
||||
RETURN_NAMES = ("response_str",)
|
||||
FUNCTION = "create_seedance_task"
|
||||
CATEGORY = "BillBum/API Nodes"
|
||||
|
||||
def _poll_task_status(self, task_id, api_url, api_key, interval=1, max_attempts=500):
|
||||
headers = {
|
||||
"Authorization": f"Bearer {api_key}",
|
||||
"Content-Type": "application/json"
|
||||
}
|
||||
url = f"{api_url}/{task_id}"
|
||||
|
||||
for attempt in range(max_attempts):
|
||||
try:
|
||||
response = requests.get(url, headers=headers)
|
||||
if response.status_code == 200:
|
||||
task_data = response.json()
|
||||
status = task_data.get("status")
|
||||
response_text = json.dumps(task_data, indent=2, ensure_ascii=False)
|
||||
if status in ["succeeded", "failed", "cancelled"]:
|
||||
return response_text
|
||||
else:
|
||||
time.sleep(interval)
|
||||
else:
|
||||
return f"Failed to fetch task status. HTTP Status Code: {response.status_code}\nResponse: {response.text}"
|
||||
except Exception as e:
|
||||
return f"An exception occurred: {str(e)}"
|
||||
return "Polling timed out."
|
||||
|
||||
def _to_base64_url(self, image_tensor):
|
||||
pil_image = tensor2pil(image_tensor)
|
||||
buffered = io.BytesIO()
|
||||
pil_image.save(buffered, format="PNG")
|
||||
img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
|
||||
return f"data:image/png;base64,{img_str}"
|
||||
|
||||
def create_seedance_task(self, model, prompt, seed, api_url, api_key, resolution, ratio, duration, camerafixed, watermark, first_frame=None, last_frame=None):
|
||||
headers = {
|
||||
"Content-Type": "application/json",
|
||||
"Authorization": f"Bearer {api_key}"
|
||||
}
|
||||
|
||||
content = [{"type": "text", "text": prompt}]
|
||||
|
||||
if first_frame is not None:
|
||||
fsf_b64url = self._to_base64_url(first_frame)
|
||||
content.append({
|
||||
"type": "image_url",
|
||||
"image_url": {"url": fsf_b64url},
|
||||
"role": "first_frame"
|
||||
})
|
||||
|
||||
if last_frame is not None:
|
||||
lsf_b64url = self._to_base64_url(last_frame)
|
||||
content.append({
|
||||
"type": "image_url",
|
||||
"image_url": {"url": lsf_b64url},
|
||||
"role": "last_frame"
|
||||
})
|
||||
|
||||
data = {
|
||||
"model": model,
|
||||
"content": content,
|
||||
"ratio": ratio,
|
||||
"resolution": resolution,
|
||||
"camera_fixed": True if camerafixed == "true" else False,
|
||||
"watermark": True if watermark == "true" else False
|
||||
}
|
||||
|
||||
# 检查是否为文生视频 (T2V) 模式
|
||||
is_t2v = all(item.get("type") == "text" for item in content)
|
||||
|
||||
# 针对 1.5-pro 系列模型，经过测试传参 duration（无论在 body 还是 prompt 中）均会导致 400 错误
|
||||
# 官方 1.5 模型目前可能为固定时长，故直接忽略该参数以确保调用成功
|
||||
if "doubao-seedance-1-5-pro" in model:
|
||||
data["generate_audio"] = True
|
||||
# 不发送 duration 参数
|
||||
else:
|
||||
# 1.0 等旧版模型仍需发送 duration
|
||||
data["duration"] = duration
|
||||
|
||||
if seed != -1:
|
||||
data["seed"] = seed
|
||||
|
||||
try:
|
||||
response = requests.post(api_url, headers=headers, json=data)
|
||||
if response.status_code == 200:
|
||||
response_json = response.json()
|
||||
task_id = response_json.get("id", "")
|
||||
if task_id:
|
||||
return (self._poll_task_status(task_id, api_url, api_key),)
|
||||
else:
|
||||
return (f"Task ID not found. Response: {response.text}",)
|
||||
else:
|
||||
return (f"Failed to create task. HTTP {response.status_code}\nResponse: {response.text}",)
|
||||
except Exception as e:
|
||||
return (f"An exception occurred: {str(e)}",)
|
||||
|
||||
|
||||
class seedream_api_node:
|
||||
|
||||
def __init__(self):
|
||||
pass
|
||||
|
||||
@classmethod
|
||||
def INPUT_TYPES(s):
|
||||
return {
|
||||
"required": {
|
||||
"model": ("STRING", {"default": "doubao-seedream-4-0-250828"}),
|
||||
"prompt": ("STRING", {"forceInput": True}),
|
||||
"size": (["1K","2K","4K"], {"default": "1K"}),
|
||||
"api_url": ("STRING", {"default": "https://ark.cn-beijing.volces.com/api/v3/images/generations"}),
|
||||
"api_key": ("STRING", {"default": "Input_your_API_key_here..."}),
|
||||
"story_mode": (
|
||||
["disabled", "auto"],
|
||||
{
|
||||
"default": "disabled",
|
||||
"description": "Enable Story Mode for generating images with consistent elements across multiple generations. (Only for Seedream 4.0 and later models)"
|
||||
},
|
||||
),
|
||||
},
|
||||
"optional": {
|
||||
"image": ("IMAGE",),
|
||||
}
|
||||
}
|
||||
|
||||
RETURN_TYPES = ("IMAGE", "STRING")
|
||||
RETURN_NAMES = ("image", "response_str")
|
||||
FUNCTION = "generate_image"
|
||||
CATEGORY = "BillBum/API Nodes"
|
||||
|
||||
def _to_base64_url_from_input(self, img_input):
|
||||
# Accept torch tensor (single or batch), PIL Image or numpy array
|
||||
# If torch tensor batch: iterate and return list of data URLs
|
||||
urls = []
|
||||
if isinstance(img_input, torch.Tensor):
|
||||
imgs = img_input
|
||||
if imgs.dim() == 3:
|
||||
imgs = imgs.unsqueeze(0)
|
||||
|
||||
# downscale large inputs to a reasonable size
|
||||
samples = imgs.movedim(-1, 1)
|
||||
total = int(1536 * 1024)
|
||||
scale_by = math.sqrt(total / (samples.shape[3] * samples.shape[2]))
|
||||
if scale_by < 1:
|
||||
width = round(samples.shape[3] * scale_by)
|
||||
height = round(samples.shape[2] * scale_by)
|
||||
s = common_upscale(samples, width, height, "lanczos", "disabled")
|
||||
imgs = s.movedim(1, -1)
|
||||
|
||||
for idx in range(imgs.shape[0]):
|
||||
pil_image = tensor2pil(imgs[idx])
|
||||
buffered = io.BytesIO()
|
||||
pil_image.save(buffered, format="PNG")
|
||||
img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
|
||||
urls.append(f"data:image/png;base64,{img_str}")
|
||||
return urls
|
||||
|
||||
# single PIL or numpy image
|
||||
if isinstance(img_input, Image.Image):
|
||||
pil_image = img_input
|
||||
else:
|
||||
try:
|
||||
pil_image = Image.fromarray(np.array(img_input))
|
||||
except Exception:
|
||||
raise TypeError("Unsupported IMAGE input type")
|
||||
|
||||
buffered = io.BytesIO()
|
||||
pil_image.save(buffered, format="PNG")
|
||||
img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
|
||||
return [f"data:image/png;base64,{img_str}"]
|
||||
|
||||
def _decode_b64_to_tensor(self, b64_string):
|
||||
if b64_string.startswith(("data:image/png;base64,", "data:image/jpeg;base64,", "data:image/webp;base64,")):
|
||||
b64 = b64_string.split(",", 1)[1]
|
||||
else:
|
||||
b64 = b64_string
|
||||
image_data = base64.b64decode(b64)
|
||||
image = Image.open(io.BytesIO(image_data))
|
||||
return pil2tensor(image)
|
||||
|
||||
@tenacity.retry(wait=tenacity.wait_exponential(multiplier=1.25, min=2, max=20), stop=tenacity.stop_after_attempt(3), reraise=True)
|
||||
def generate_image(self, model, prompt, size, api_url, api_key, story_mode, image=None):
|
||||
headers = {
|
||||
"Content-Type": "application/json",
|
||||
"Authorization": f"Bearer {api_key}"
|
||||
}
|
||||
|
||||
if image is not None and model.startswith("doubao-seedream-4-5") and size == "1K":
|
||||
size = "2K"
|
||||
|
||||
payload = {
|
||||
"model": model,
|
||||
"prompt": prompt,
|
||||
"response_format": "b64_json",
|
||||
"size": size,
|
||||
"stream": False,
|
||||
"watermark": False,
|
||||
}
|
||||
|
||||
if story_mode == "disabled" and model.startswith("doubao-seedream-4"):
|
||||
payload["sequential_image_generation"] = "disabled"
|
||||
|
||||
if story_mode == "auto" and model.startswith("doubao-seedream-4"):
|
||||
payload["sequential_image_generation"] = "auto"
|
||||
|
||||
encoded_imgs = []
|
||||
if image is not None:
|
||||
try:
|
||||
encoded_imgs = self._to_base64_url_from_input(image)
|
||||
except Exception as e:
|
||||
return (None, f"Error encoding input image: {e}")
|
||||
|
||||
if encoded_imgs:
|
||||
if len(encoded_imgs) == 1:
|
||||
payload["image"] = encoded_imgs[0]
|
||||
else:
|
||||
payload["image"] = encoded_imgs
|
||||
|
||||
try:
|
||||
response = requests.post(api_url, headers=headers, json=payload)
|
||||
response.raise_for_status()
|
||||
except Exception as e:
|
||||
debug_info = {
|
||||
"request_payload": payload,
|
||||
"error_message": str(e),
|
||||
"response_text": getattr(response, 'text', 'No response text available')
|
||||
}
|
||||
pretty_debug = json.dumps(debug_info, indent=2, ensure_ascii=False)
|
||||
return (None, f"Request failed:\n{pretty_debug}")
|
||||
|
||||
try:
|
||||
response_json = response.json()
|
||||
except Exception as e:
|
||||
debug_info = {
|
||||
"request_payload": payload,
|
||||
"error_message": str(e),
|
||||
"raw_response": getattr(response,'text',str(response))
|
||||
}
|
||||
pretty_debug = json.dumps(debug_info, indent=2, ensure_ascii=False)
|
||||
return (None, f"Failed to parse JSON response:\n{pretty_debug}")
|
||||
|
||||
images_output = []
|
||||
data_list = response_json.get("data", [])
|
||||
if not data_list and "b64_json" in response_json:
|
||||
data_list = [response_json]
|
||||
|
||||
for item in data_list:
|
||||
b64_str = item.get("b64_json")
|
||||
if not b64_str:
|
||||
continue
|
||||
try:
|
||||
img_tensor = self._decode_b64_to_tensor(b64_str)
|
||||
images_output.append(img_tensor)
|
||||
except Exception as e:
|
||||
print(f"Failed decoding an image from response: {e}")
|
||||
|
||||
debug_info = {
|
||||
"request_payload": payload,
|
||||
"response": response_json
|
||||
}
|
||||
pretty = json.dumps(debug_info, indent=2, ensure_ascii=False)
|
||||
|
||||
if not images_output:
|
||||
return (None, f"Unexpected or empty image response:\n{pretty}")
|
||||
|
||||
try:
|
||||
batch = torch.cat(images_output, dim=0)
|
||||
except Exception:
|
||||
batch = images_output[0]
|
||||
|
||||
return (batch, pretty)
|
||||
|
||||
+15
-10
@@ -94,7 +94,7 @@ class BillBum_Modified_StreamResponse_LLM_API:
|
||||
encoded_images.append(encoded)
|
||||
return encoded_images
|
||||
|
||||
@tenacity.retry(wait=tenacity.wait_exponential(multiplier=1.25, min=5, max=30), stop=tenacity.stop_after_attempt(3))
|
||||
@tenacity.retry(wait=tenacity.wait_exponential(multiplier=1.25, min=5, max=30), stop=tenacity.stop_after_attempt(3), reraise=True)
|
||||
def get_llm_stream_response(
|
||||
self,
|
||||
prompt,
|
||||
@@ -145,16 +145,21 @@ class BillBum_Modified_StreamResponse_LLM_API:
|
||||
if extra_body:
|
||||
request_kwargs["extra_body"] = extra_body
|
||||
|
||||
completion = client.chat.completions.create(**request_kwargs)
|
||||
try:
|
||||
completion = client.chat.completions.create(**request_kwargs)
|
||||
|
||||
full_content = ""
|
||||
for chunk in completion:
|
||||
if chunk.choices and chunk.choices[0].delta.content is not None:
|
||||
delta = chunk.choices[0].delta.content
|
||||
full_content += delta
|
||||
print(delta, end="")
|
||||
|
||||
return (full_content,)
|
||||
full_content = ""
|
||||
for chunk in completion:
|
||||
if chunk.choices and chunk.choices[0].delta.content is not None:
|
||||
delta = chunk.choices[0].delta.content
|
||||
full_content += delta
|
||||
# print(delta, end="") # For debugging stream output
|
||||
return (full_content,)
|
||||
|
||||
except Exception as e:
|
||||
print(f"LLM API Error: {type(e).__name__} - {e}")
|
||||
# Re-raise the exception to allow tenacity to handle retries
|
||||
raise
|
||||
|
||||
|
||||
class Url2Image:
|
||||
|
||||
+1
-1
@@ -1,7 +1,7 @@
|
||||
[project]
|
||||
name = "comfyui_billbum_api_nodes"
|
||||
description = "API call node for Third-party platforms both official and local. Support VLMs LLMs Dalle3 Flux-Pro(Support kontext, banana etc... now!! and new Support gpt-image-1!!). And some little tools: img to b64 url, b64 url to img, b64 url to b64 data, reg text to word and ',' only, etc."
|
||||
version = "1.1.8"
|
||||
version = "1.1.9"
|
||||
license = {file = "LICENSE"}
|
||||
dependencies = ["tenacity", "openai", "pillow", "requests", "numpy", "tiktoken", "urlextract"]
|
||||
|
||||
|
||||
Reference in New Issue
Block a user