bizyengine 1.2.49__py3-none-any.whl → 1.2.50__py3-none-any.whl

bizyengine/bizyair_extras/nodes_gemini.py

@@ -1,11 +1,15 @@
 import base64
 import io
 import json
+import logging
 import re
 
 import numpy as np
 import torch
-from comfy_api_nodes.apinode_utils import tensor_to_base64_string
+from comfy_api_nodes.apinode_utils import (
+    bytesio_to_image_tensor,
+    tensor_to_base64_string,
+)
 from PIL import Image, ImageOps
 
 from bizyengine.core import BizyAirBaseNode, pop_api_key_and_prompt_id
@@ -58,6 +62,94 @@ def base64_to_image(base64_string):
     return image
 
 
+def get_parts_from_response(
+    response: dict,
+):
+    return response["candidates"][0]["content"]["parts"]
+
+
+def get_parts_by_type(response: dict, part_type: str):
+    parts = []
+    for part in get_parts_from_response(response):
+        if part_type == "text" and part.get("text", None):
+            parts.append(part)
+        elif (
+            part.get("inlineData", None) and part["inlineData"]["mimeType"] == part_type
+        ):
+            parts.append(part)
+        # Skip parts that don't match the requested type
+    return parts
+
+
+def get_text_from_response(response: dict) -> str:
+    parts = get_parts_by_type(response, "text")
+    logging.debug(f"Text parts: {parts}")
+    return "\n".join([part["text"] for part in parts])
+
+
+def get_image_from_response(response: dict) -> torch.Tensor:
+    image_tensors: list[torch.Tensor] = []
+    parts = get_parts_by_type(response, "image/png")
+    for part in parts:
+        b64_data = part["inlineData"]["data"]
+        if b64_data:
+            image_data = base64.b64decode(b64_data)
+            returned_image = bytesio_to_image_tensor(io.BytesIO(image_data))
+            image_tensors.append(returned_image)
+    if len(image_tensors) == 0:
+        return torch.zeros((1, 1024, 1024, 4))
+    return torch.cat(image_tensors, dim=0)
+
+
+# FROM: https://github.com/ShmuelRonen/ComfyUI-NanoBanano/blob/9eeb8f2411fd0ff08791bdf5e24eec347456c8b8/nano_banano.py#L191
+def build_prompt_for_operation(
+    prompt,
+    operation,
+    has_references=False,
+    aspect_ratio="1:1",
+    character_consistency=True,
+):
+    """Build optimized prompt based on operation type"""
+
+    aspect_instructions = {
+        "1:1": "square format",
+        "16:9": "widescreen landscape format",
+        "9:16": "portrait format",
+        "4:3": "standard landscape format",
+        "3:4": "standard portrait format",
+    }
+
+    base_quality = "Generate a high-quality, photorealistic image"
+    format_instruction = f"in {aspect_instructions.get(aspect_ratio, 'square format')}"
+
+    if operation == "generate":
+        if has_references:
+            final_prompt = f"{base_quality} inspired by the style and elements of the reference images. {prompt}. {format_instruction}."
+        else:
+            final_prompt = f"{base_quality} of: {prompt}. {format_instruction}."
+
+    elif operation == "edit":
+        if not has_references:
+            return "Error: Edit operation requires reference images"
+        # No aspect ratio for edit - preserve original image dimensions
+        final_prompt = f"Edit the provided reference image(s). {prompt}. Maintain the original composition and quality while making the requested changes."
+
+    elif operation == "style_transfer":
+        if not has_references:
+            return "Error: Style transfer requires reference images"
+        final_prompt = f"Apply the style from the reference images to create: {prompt}. Blend the stylistic elements naturally. {format_instruction}."
+
+    elif operation == "object_insertion":
+        if not has_references:
+            return "Error: Object insertion requires reference images"
+        final_prompt = f"Insert or blend the following into the reference image(s): {prompt}. Ensure natural lighting, shadows, and perspective. {format_instruction}."
+
+    if character_consistency and has_references:
+        final_prompt += " Maintain character consistency and visual identity from the reference images."
+
+    return final_prompt
+
+
 class NanoBanana(BizyAirBaseNode):
     def __init__(self):
         pass
@@ -73,6 +165,13 @@ class NanoBanana(BizyAirBaseNode):
                         "default": "",
                     },
                 ),
+                "operation": (
+                    ["generate", "edit", "style_transfer", "object_insertion"],
+                    {
+                        "default": "generate",
+                        "tooltip": "Choose the type of image operation",
+                    },
+                ),
                 "temperature": (
                     "FLOAT",
                     {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.05},
@@ -90,73 +189,123 @@ class NanoBanana(BizyAirBaseNode):
                 "image3": ("IMAGE",),
                 "image4": ("IMAGE",),
                 "image5": ("IMAGE",),
+                "quality": (
+                    ["standard", "high"],
+                    {"default": "high", "tooltip": "Image generation quality"},
+                ),
+                "aspect_ratio": (
+                    ["1:1", "16:9", "9:16", "4:3", "3:4"],
+                    {"default": "1:1", "tooltip": "Output image aspect ratio"},
+                ),
+                "character_consistency": (
+                    "BOOLEAN",
+                    {
+                        "default": True,
+                        "tooltip": "Maintain character consistency across edits",
+                    },
+                ),
             },
         }
 
-    RETURN_TYPES = ("IMAGE",)
+    RETURN_TYPES = ("IMAGE", "STRING")
     FUNCTION = "execute"
     OUTPUT_NODE = False
     CATEGORY = "☁️BizyAir/External APIs/Gemini"
 
-    def execute(self, prompt, temperature, top_p, seed, max_tokens, **kwargs):
-        url = f"{BIZYAIR_SERVER_ADDRESS}/proxy_inference/VertexAI/gemini-2.5-flash-image-preview"
-        extra_data = pop_api_key_and_prompt_id(kwargs)
-        parts = [{"text": prompt}]
-        for _, img in enumerate(
-            [
-                kwargs.get("image", None),
-                kwargs.get("image2", None),
-                kwargs.get("image3", None),
-                kwargs.get("image4", None),
-                kwargs.get("image5", None),
-            ],
-            1,
-        ):
-            if img is not None:
-                parts.append(
-                    {
-                        "inline_data": {
-                            "mime_type": "image/png",
-                            "data": tensor_to_base64_string(img),
+    def execute(
+        self,
+        prompt,
+        operation,
+        temperature,
+        top_p,
+        seed,
+        max_tokens,
+        quality=None,
+        aspect_ratio=None,
+        character_consistency=None,
+        **kwargs,
+    ):
+        try:
+            url = f"{BIZYAIR_SERVER_ADDRESS}/proxy_inference/VertexAI/gemini-2.5-flash-image-preview"
+            extra_data = pop_api_key_and_prompt_id(kwargs)
+
+            parts = []
+            for _, img in enumerate(
+                [
+                    kwargs.get("image", None),
+                    kwargs.get("image2", None),
+                    kwargs.get("image3", None),
+                    kwargs.get("image4", None),
+                    kwargs.get("image5", None),
+                ],
+                1,
+            ):
+                if img is not None:
+                    parts.append(
+                        {
+                            "inline_data": {
+                                "mime_type": "image/png",
+                                "data": tensor_to_base64_string(img),
+                            }
                         }
-                    }
+                    )
+
+            prompt = build_prompt_for_operation(
+                prompt,
+                operation,
+                has_references=len(parts) > 0,
+                aspect_ratio=aspect_ratio,
+                character_consistency=character_consistency,
+            )
+            if quality == "high":
+                prompt += " Use the highest quality settings available."
+            parts.append({"text": prompt})
+
+            data = {
+                "contents": {
+                    "parts": parts,
+                    "role": "user",
+                },
+                "generationConfig": {
+                    "seed": seed,
+                    "responseModalities": ["TEXT", "IMAGE"],
+                    "temperature": temperature,
+                    "topP": top_p,
+                    "maxOutputTokens": max_tokens,
+                },
+            }
+            json_payload = json.dumps(data).encode("utf-8")
+            headers = client.headers(api_key=extra_data["api_key"])
+            headers["X-BIZYAIR-PROMPT-ID"] = extra_data[
+                "prompt_id"
+            ]  # 额外参数vertexai会拒绝，所以用请求头传
+            resp = client.send_request(
+                url=url,
+                data=json_payload,
+                headers=headers,
+            )
+            # 解析潜在错误
+            prompt_feedback = resp.get("promptFeedback", None)
+            if prompt_feedback:
+                logging.error(f"Response: {resp}")
+                raise ValueError(f"Prompt blocked: {prompt_feedback}")
+            if len(resp["candidates"]) == 0:
+                logging.error(f"Response: {resp}")
+                raise ValueError("No candidates found in response")
+            if resp["candidates"][0]["finishReason"] != "STOP":
+                logging.error(f"Response: {resp}")
+                raise ValueError(
+                    f"Erroneous finish reason: {resp['candidates'][0]['finishReason']}"
                 )
-        data = {
-            "contents": {
-                "parts": parts,
-                "role": "user",
-            },
-            "generationConfig": {
-                "seed": seed,
-                "responseModalities": ["TEXT", "IMAGE"],
-                "temperature": temperature,
-                "topP": top_p,
-                "maxOutputTokens": max_tokens,
-            },
-        }
-        json_payload = json.dumps(data).encode("utf-8")
-        headers = client.headers(api_key=extra_data["api_key"])
-        headers["X-BIZYAIR-PROMPT-ID"] = extra_data[
-            "prompt_id"
-        ]  # 额外参数vertexai会拒绝，所以用请求头传
-        resp = client.send_request(
-            url=url,
-            data=json_payload,
-            headers=headers,
-        )
-        # 解析base64图片
-        b64_data = None
-        for part in resp["candidates"][0]["content"]["parts"]:
-            if part.get("inlineData", None):
-                b64_data = part["inlineData"]["data"]
-                break
-        if b64_data:
-            i = base64_to_image(b64_data)
-            # 下面代码参考LoadImage
-            i = ImageOps.exif_transpose(i)
-            image = i.convert("RGB")
-            image = np.array(image).astype(np.float32) / 255.0
-            image = torch.from_numpy(image)[None,]
-            return (image,)
-        else:
-            raise ValueError("No image found in response")
+
+            # 解析文本
+            text = get_text_from_response(resp)
+
+            # 解析base64图片
+            image = get_image_from_response(resp)
+
+            return (image, text)
+
+        except Exception as e:
+            logging.error(f"Gemini API error: {e}")
+            raise e

bizyengine/version.txt

@@ -1 +1 @@
-1.2.49
+1.2.50

{bizyengine-1.2.49.dist-info → bizyengine-1.2.50.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bizyengine
-Version: 1.2.49
+Version: 1.2.50
 Summary: [a/BizyAir](https://github.com/siliconflow/BizyAir) Comfy Nodes that can run in any environment.
 Author-email: SiliconFlow <yaochi@siliconflow.cn>
 Project-URL: Repository, https://github.com/siliconflow/BizyAir

{bizyengine-1.2.49.dist-info → bizyengine-1.2.50.dist-info}/RECORD

@@ -1,5 +1,5 @@
 bizyengine/__init__.py,sha256=GP9V-JM07fz7uv_qTB43QEA2rKdrVJxi5I7LRnn_3ZQ,914
-bizyengine/version.txt,sha256=AgBNjKfkdaaSpzlSp-niqT0eeAyzzEPm_pDtkPfEXLE,7
+bizyengine/version.txt,sha256=zGrmrSrzHGOEzO23IqIfRL3venxRA-IUVh6N6ZhB4X8,7
 bizyengine/bizy_server/__init__.py,sha256=SP9oSblnPo4KQyh7yOGD26YCskFAcQHAZy04nQBNRIw,200
 bizyengine/bizy_server/api_client.py,sha256=Z7G5IjaEqSJkF6nLLw2R3bpgBAOi5ClQiUbel6NMXmE,43932
 bizyengine/bizy_server/errno.py,sha256=1UiFmE2U7r7hCHgsw4-p_YL0VCmTJc9NyYDEbhkanaY,16336
@@ -22,7 +22,7 @@ bizyengine/bizyair_extras/nodes_custom_sampler.py,sha256=NK-7sdcp8oxJisjTEFfBskk
 bizyengine/bizyair_extras/nodes_dataset.py,sha256=htF0YZb_FHncLhLDEbJfNCVqJ6rvlo1ZLk7iY42Rylc,3440
 bizyengine/bizyair_extras/nodes_differential_diffusion.py,sha256=nSrbD-w0XtrwktwzME5M0Vmi1sI7Z08AqwgymTdThqo,370
 bizyengine/bizyair_extras/nodes_flux.py,sha256=ls94kGBuBNgW5c6uhG36iZLk1TTM2TIoTTcpERgEE5E,2683
-bizyengine/bizyair_extras/nodes_gemini.py,sha256=0Yh8NINFVkU3kh7PuYErhweO5id3SI3IXQORO4J1fNw,5076
+bizyengine/bizyair_extras/nodes_gemini.py,sha256=AdHMxVCYuv7HTBEnxfqSIGLsmICWM5WhDvsLdJWLIqI,10594
 bizyengine/bizyair_extras/nodes_hunyuan3d.py,sha256=dWHLeqX68N7zKnfDMzm9nutmCNtFT6-wwt7P5cPDu7Q,2058
 bizyengine/bizyair_extras/nodes_image_utils.py,sha256=BldF_CKD2M01K8-SnG-QV86u3HZqFz_GP5GrCQ5CFDQ,2875
 bizyengine/bizyair_extras/nodes_ip2p.py,sha256=GSEFJvrs4f2tv0xwYkWqc8uhsXrzAJVPvvwcw0gTjR0,619
@@ -80,7 +80,7 @@ bizyengine/misc/route_sam.py,sha256=-bMIR2QalfnszipGxSxvDAHGJa5gPSrjkYPb5baaRg4,
 bizyengine/misc/segment_anything.py,sha256=wNKYwlYPMszfwj23524geFZJjZaG4eye65SGaUnh77I,8941
 bizyengine/misc/supernode.py,sha256=STN9gaxfTSErH8OiHeZa47d8z-G9S0I7fXuJvHQOBFM,4532
 bizyengine/misc/utils.py,sha256=CKduySGSMNGlJMImHyZmN-giABY5VUaB88f6Kq-HAV0,6831
-bizyengine-1.2.49.dist-info/METADATA,sha256=HgvznMszXmVxmt_bEpKgowgsv5FNUfJLBukriYLIyLk,708
-bizyengine-1.2.49.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-bizyengine-1.2.49.dist-info/top_level.txt,sha256=2zapzqxX-we5cRyJkGf9bd5JinRtXp3-_uDI-xCAnc0,11
-bizyengine-1.2.49.dist-info/RECORD,,
+bizyengine-1.2.50.dist-info/METADATA,sha256=oDy6bxpEMjEUlrLgBDEdsI-Rz1pW7SM_lDl3b5HbNJk,708
+bizyengine-1.2.50.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+bizyengine-1.2.50.dist-info/top_level.txt,sha256=2zapzqxX-we5cRyJkGf9bd5JinRtXp3-_uDI-xCAnc0,11
+bizyengine-1.2.50.dist-info/RECORD,,