app.py

import gradio as gr
import numpy as np
# from edict_functions import EDICT_editing
from PIL import Image
from utils import Endpoint, get_token
from io import BytesIO
import requests


endpoint = Endpoint()

def local_edict(x, source_text, edit_text,
         edit_strength, guidance_scale,
          steps=50, mix_weight=0.93, ):
    x = Image.fromarray(x)
    return_im =  EDICT_editing(x,
                         source_text,
                         edit_text,
                  steps=steps,
                  mix_weight=mix_weight,
                  init_image_strength=edit_strength,
                  guidance_scale=guidance_scale
                              )[0]
    return np.array(return_im)

def encode_image(image):
    buffered = BytesIO()
    image.save(buffered, format="JPEG", quality=95)
    buffered.seek(0)

    return buffered



def decode_image(img_obj):
    img = Image.open(img_obj).convert("RGB")
    return img

def edict(x, source_text, edit_text,
         edit_strength, guidance_scale,
          steps=50, mix_weight=0.93, ):

    url = endpoint.url
    url = url + "/api/edit"
    headers = {### Misc.

        "User-Agent": "EDICT HuggingFace Space",
        "Auth-Token": get_token(),
    }

    data = {
        "source_text": source_text,
        "edit_text": edit_text,
        "edit_strength": edit_strength,
        "guidance_scale": guidance_scale,
    }

    image = encode_image(Image.fromarray(x))
    files = {"image": image}  

    response = requests.post(url, data=data, files=files, headers=headers)
    
    if response.status_code == 200:
        return np.array(decode_image(BytesIO(response.content)))
    else:
        return "Error: " + response.text   
    # x = decode_image(response)
    # return np.array(x)

examples = [
        ['square_ims/american_gothic.jpg', 'A painting of two people frowning', 'A painting of two people smiling', 0.5, 3],
        ['square_ims/colloseum.jpg', 'An old ruined building', 'A new modern office building', 0.8, 3],
    ]


examples.append(['square_ims/scream.jpg', 'A painting of someone screaming', 'A painting of an alien', 0.5, 3])
examples.append(['square_ims/yosemite.jpg', 'Granite forest valley', 'Granite desert valley', 0.8, 3])
examples.append(['square_ims/einstein.jpg', 'Mouth open', 'Mouth closed', 0.8, 3])
examples.append(['square_ims/einstein.jpg', 'A man', 'A man in K.I.S.S. facepaint', 0.8, 3])
"""
examples.extend([
        ['square_ims/imagenet_cake_2.jpg', 'A cupcake', 'A Chinese New Year cupcake', 0.8, 3],
        ['square_ims/imagenet_cake_2.jpg', 'A cupcake', 'A Union Jack cupcake', 0.8, 3],
        ['square_ims/imagenet_cake_2.jpg', 'A cupcake', 'A Nigerian flag cupcake', 0.8, 3],
        ['square_ims/imagenet_cake_2.jpg', 'A cupcake', 'A Santa Claus cupcake', 0.8, 3],
        ['square_ims/imagenet_cake_2.jpg', 'A cupcake', 'An Easter cupcake', 0.8, 3],
        ['square_ims/imagenet_cake_2.jpg', 'A cupcake', 'A hedgehog cupcake', 0.8, 3],
        ['square_ims/imagenet_cake_2.jpg', 'A cupcake', 'A rose cupcake', 0.8, 3],
    ])
"""

for dog_i in [1, 2]:
    for breed in ['Golden Retriever', 'Chihuahua', 'Dalmatian']:
        examples.append([f'square_ims/imagenet_dog_{dog_i}.jpg', 'A dog', f'A {breed}', 0.8, 3])


description = '**For safety and ethical considerations, we have disabled image uploading from March 21. 2023.\nPlease try examples provided below.**\nA gradio demo for [EDICT](https://arxiv.org/abs/2211.12446) (CVPR23)'
# description = gr.Markdown(description)

article = """

### Prompting Style

As with many text-to-image methods, the prompting style of EDICT can make a big difference. When in doubt, experiment! Some guidance:
* Parallel *Original Description* and *Edit Description* construction as much as possible. Inserting/editing single words often is enough to affect a change while maintaining a lot of the original structure
* Words that will affect the entire setting (e.g. "A photo of " vs. "A painting of") can make a big difference. Playing around with them can help a lot

### Parameters
Both `edit_strength` and `guidance_scale` have similar properties qualitatively: the higher the value the more the image will change. We suggest
* Increasing/decreasing `edit_strength` first, particularly to alter/preserve more of the original structure/content
* Then changing `guidance_scale` to make the change in the edited region more or less pronounced.

Usually we find changing `edit_strength` to be enough, but feel free to play around (and report any interesting results)!

### Misc.

Having difficulty coming up with a caption? Try [BLIP](https://huggingface.co/spaces/Salesforce/BLIP2) to automatically generate one!

As with most StableDiffusion approaches, faces/text are often problematic to render, especially if they're small. Having these in the foreground will help keep them cleaner.

A returned black image means that the [Safety Checker](https://huggingface.co/CompVis/stable-diffusion-safety-checker) triggered on the photo. This happens in odd cases sometimes (it often rejects
the huggingface logo or variations), but we need to keep it in for obvious reasons.
"""
# article = gr.Markdown(description)

iface = gr.Interface(fn=edict, inputs=[gr.Image(interactive=False),
                                       gr.Textbox(label="Original Description"),
                                       gr.Textbox(label="Edit Description"),
                                       # 50, # gr.Slider(5, 50, value=20, step=1),
                                       # 0.93, # gr.Slider(0.5, 1, value=0.7, step=0.05),
                                       gr.Slider(0.0, 1, value=0.8, step=0.05),
                                       gr.Slider(0, 10, value=3, step=0.5),
                                      ],
                     examples = examples,
                     outputs="image",
                     description=description,
                     article=article,
                     cache_examples=True)
iface.launch()