File size: 2,212 Bytes
7dbdbc9
 
3c71e14
b6b2b75
3c71e14
 
 
7dbdbc9
 
 
63f6eb9
 
7dbdbc9
3c71e14
7dbdbc9
 
 
 
 
 
3c71e14
 
 
 
 
e51059c
3c71e14
 
 
 
 
 
 
 
e51059c
3c71e14
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e51059c
 
082ff75
3c71e14
e51059c
 
3c71e14
b54e214
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
import base64
from io import BytesIO
from typing import Any, Dict

import torch
from diffusers import FluxFillPipeline
from PIL import Image


def decode_image(b64_string):
    image_data = base64.b64decode(b64_string)
    return Image.open(BytesIO(image_data)).convert("RGB")


def encode_image(image):
    buffer = BytesIO()
    image.save(buffer, format="PNG")
    return base64.b64encode(buffer.getvalue()).decode("utf-8")


class EndpointHandler:
    def __init__(self, path="shangguanyanyan/flux1-fill-dev-custom"):
        self.pipe = FluxFillPipeline.from_pretrained(
            path, torch_dtype=torch.bfloat16
        ).to("cuda" if torch.cuda.is_available() else "cpu")

        self.parameters = {
            "height": 1632,
            "width": 1232,
            "guidance_scale": 30,
            "num_inference_steps": 50,
            "max_sequence_length": 512,
            "generator": torch.Generator("cpu").manual_seed(0),
        }

    def __call__(self, data: Any) -> Dict[str, Any]:
        """
        data: {
            "inputs": {
                "image": base64_image,
                "mask": base64_mask,
                "prompt": prompt
            },
            "parameters": {
                "height": 1632,
                "width": 1232,
                "guidance_scale": 30,
                "num_inference_steps": 50,
                "max_sequence_length": 512,
            }
        }
        """
        inputs = data.pop("inputs", data)
        parameters = data.pop("parameters", {})

        parameters.update(self.parameters)
        base64_image = inputs.pop("image", "")
        base64_mask = inputs.pop("mask", "")
        prompt = inputs.pop("prompt", "")

        if not base64_image or not base64_mask or not prompt:
            return {
                "error": "Please provide image, mask and prompt",
                "status": "failed",
            }

        image = decode_image(base64_image)
        mask = decode_image(base64_mask)

        image = self.pipe(
            prompt=prompt,
            image=image,
            mask_image=mask,
            **parameters,
        ).images[0]

        return {"image": encode_image(image), "status": "success"}