genai first commit

README.md

@@ -1,8 +1,8 @@
 ---
-title: Generative Ai
-emoji: 🏢
-colorFrom: green
-colorTo: blue
+title: Text To Video
+emoji: 🐠
+colorFrom: purple
+colorTo: purple
 sdk: streamlit
 sdk_version: 1.30.0
 app_file: app.py

app.py

@@ -0,0 +1,36 @@
+import streamlit as st
+from src.page1 import page1
+from src.page2 import page2
+from src.page3 import page3
+from src.page4 import page4
+from src.page5 import page5
+from src.page6 import page6
+from src.page7 import page7
+from src.page8 import page8
+# import openai
+import os
+from dotenv import load_dotenv
+load_dotenv()
+st.set_page_config(layout="wide")
+
+
+# openai.api_key = os.getenv("OPENAI_API_KEY")
+
+
+pages = {
+    # "Entry point": page1,
+    # "Text to image": page2,
+    # "Image variation": page3,
+    # "Image edit": page4,
+    "Text to Image":page5,
+    "Image to Text": page6,
+    "Image to Video":page7,
+    "Text to Video": page8
+    
+}
+
+# Create the selectbox in the sidebar
+page = st.sidebar.selectbox("Select a page", list(pages.keys()))
+
+# Display the selected page
+pages[page]()

src/.env

@@ -0,0 +1,2 @@
+OPENAI_API_KEY="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"
+REPLICATE_API_TOKEN='r8_4fktoXrDGkgHY8uw1XlVtQJKQlAILKv0iBmPI'

src/page1.py

@@ -0,0 +1,46 @@
+import streamlit as st
+
+introduction = """
+Brief details about OpenAI DALL-E:
+
+### What is OpenAI DALL-E?
+
+DALL-E is a 12-billion parameter version of GPT-3 trained to generate images from text 
+descriptions, using a dataset of text–image pairs². It can create original, realistic 
+images and art from a text description¹. 
+
+### What can OpenAI DALL-E do?
+
+DALL-E has a diverse set of capabilities, including creating anthropomorphized versions 
+of animals and objects, combining unrelated concepts in plausible ways, rendering text, 
+and applying transformations to existing images². DALL-E 2 can make realistic edits to 
+existing images from a natural language caption³.
+
+### How does OpenAI DALL-E work?
+DALL-E works by using a dataset of text–image pairs to learn how to generate images 
+from textual descriptions². It uses a 12-billion parameter version of GPT-3².
+
+### How can I use OpenAI DALL-E?
+DALL-E is available in beta and can be used by signing up on their website⁵. 
+
+```
+Example usage:
+Input: A cat made of sushi
+Output: An image of a cat made out of sushi
+```
+
+Source: Conversation with Bing, 4/29/2023
+
+- (1) DALL·E: Creating images from text - OpenAI. https://openai.com/research/dall-e.
+- (2) DALL·E 2 - OpenAI. https://openai.com/product/dall-e-2.
+- (3) DALL·E - OpenAI. https://labs.openai.com/.
+- (4) DALL·E now available in beta - OpenAI. https://openai.com/blog/dall-e-now-available-in-beta/.
+- (5) DALL·E 2 - openai.com. https://openai.com/product/dall-e-2?ref=promptsreport.com.
+- (6) OpenAI API. https://platform.openai.com/docs/models/dall-e.
+
+"""
+
+# Define the pages
+def page1():
+    st.title("OpenAI DALL·E")
+    st.markdown(introduction)

src/page2.py

@@ -0,0 +1,37 @@
+import streamlit as st
+# import openai
+import replicate
+
+def page2():
+    st.title("OpenAI DALL·E Image Generation")
+    st.info("""#### NOTE: you can download image by \
+    right clicking on the image and select save image as option""")
+
+    with st.form(key='form'):
+        prompt = st.text_input(label='Enter text prompt for image generation')
+        size = st.selectbox('Select size of the images', 
+                            ('256x256', '512x512', '1024x1024'))
+        num_images = st.selectbox('Enter number of images to be generated', (1,2,3,4))
+        submit_button = st.form_submit_button(label='Submit')
+
+    if submit_button:
+        if prompt:
+            output = replicate.run(
+                "stability-ai/stable-video-diffusion:3f0457e4619daac51203dedb472816fd4af51f3149fa7a9e0b5ffcf1b8172438",
+                input={
+                    "input_image": "https://example.com/path/to/file/input_image"
+                }
+            )
+            print(output)
+            
+            # response = openai.Image.create(
+            #         prompt = prompt,
+            #         n = num_images,
+            #         size=size,
+            #     )
+            
+            # for idx in range(num_images):
+            #     image_url = response["data"][idx]["url"]
+
+            #     st.image(image_url, caption=f"Generated image: {idx+1}",
+            #              use_column_width=True)

src/page3.py

@@ -0,0 +1,38 @@
+import streamlit as st
+# import openai
+from PIL import Image
+from src.utils import get_width_height, resize_image
+
+def page3():
+    st.title("OpenAI DALL·E Image Variation")
+    st.info("""#### NOTE: you can download image by \
+    right clicking on the image and select save image as option""")
+
+    with st.form(key='form'):
+        uploaded_file = st.file_uploader("Choose an image file", type=['png', 'jpg'])
+        size = st.selectbox('Select size of the images', 
+                            ('256x256', '512x512', '1024x1024'))
+        num_images = st.selectbox('Enter number of images to be generated', (1,2,3,4))
+        submit_button = st.form_submit_button(label='Submit')
+
+    if submit_button:
+        if uploaded_file is not None:
+
+            image = Image.open(uploaded_file)
+            
+            st.image(image, caption="Uploaded image", use_column_width=True)
+            
+            width, height = get_width_height(size)
+            image = resize_image(image, width, height)
+            
+            # response = openai.Image.create_variation(
+            #         image=image,
+            #         n = num_images,
+            #         size=size,
+            #     )
+            
+            # for idx in range(num_images):
+            #     image_url = response["data"][idx]["url"]
+
+            #     st.image(image_url, caption=f"Generated image: {idx+1}",
+            #              use_column_width=True)

src/page4.py

@@ -0,0 +1,51 @@
+import streamlit as st
+# import openai
+from PIL import Image
+# from src.utils import get_width_height, resize_image
+# from rembg import remove
+
+def page4():
+    st.title("OpenAI DALL·E Image Editing")
+    st.info("""#### NOTE: you can download image by \
+    right clicking on the image and select save image as option""")
+
+    with st.form(key='form'):
+        uploaded_file = st.file_uploader("Choose an image file", type=['png', 'jpg'])
+        mask_file = st.file_uploader("Choose an mask file", type=['png', 'jpg'])
+        prompt = st.text_input("Enter a text prompt")
+        size = st.selectbox('Select size of the images', ('256x256', '512x512', '1024x1024'))
+        num_images = st.selectbox('Enter number of images to be generated', (1,2,3,4))
+        submit_button = st.form_submit_button(label='Submit')
+
+    if submit_button:
+        if (uploaded_file is not None) and (mask_file is not None) and prompt:
+            our_image = Image.open(uploaded_file)
+            mask_image = Image.open(mask_file)
+
+            width, height = get_width_height(size)
+
+            our_image = resize_image(our_image, width, height)
+            mask_image = resize_image(mask_image, width, height)
+
+            st.image(our_image, caption="Uploaded image", use_column_width=True)
+            st.image(mask_image, caption="Uploaded mask", use_column_width=True)
+
+            backround_removed_mask = remove(mask_image)
+
+            st.image(backround_removed_mask, caption="backround_removed_mask", 
+                     use_column_width=True)
+            
+            # response = openai.Image.create_edit(
+            #     image=our_image,
+            #     mask=backround_removed_mask,
+            #     prompt=prompt,
+            #     n=num_images,
+            #     size=size
+            # )
+
+            # for idx in range(num_images):
+            #     image_url = response["data"][idx]["url"]
+
+            #     st.image(image_url, caption=f"Generated image: {idx+1}",
+            #              use_column_width=True)
+

src/page5.py

@@ -0,0 +1,115 @@
+import streamlit as st
+# import openai
+from streamlit_extras.stylable_container import stylable_container
+
+import replicate
+
+streamlit_style = """
+            <style>
+            #MainMenu {visibility: hidden;}
+            footer {visibility: hidden;}
+            video{width:200px;}
+            .css-1wbqy5l {visibility: hidden;}
+            .css-15zrgzn {visibility: hidden;}
+            .css-klqnuk {visibility: hidden;}
+            .en6cib64 {visibility: hidden;}
+            .css-1u4fkce {visibility: hidden;}
+            .en6cib62 {visibility: hidden;}
+            .css-19rxjzo, .ef3psqc11 {
+            background-color: purple;
+            text-color: white;
+            }
+            div.stButton > button:first-child {
+            background-color: red;
+            text-weight: bold;
+            }
+            </style>
+            """
+def page5():
+    # st.title("Text to Image")
+    with stylable_container(
+         key="title",
+         css_styles=[
+             """ span {
+                text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }"""
+                ,
+                """
+                st-emotion-cache-0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }""",
+                
+                """
+                .e1f1d6gn0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }
+        """,
+        ],
+        
+    ):
+        st.markdown("<h3>Text to Image<h3>", unsafe_allow_html=True)
+        st.markdown(streamlit_style, unsafe_allow_html=True) 
+
+    # st.info("""#### NOTE: you can download image by \
+    # right clicking on the image and select save image as option""")
+    with st.form(key='form'):
+        prompt = st.text_input(label='Enter text prompt for image generation')
+        placeholder=st.empty()
+        col1,col2=placeholder.columns(2)
+        number_of_image = col1.number_input("Number of Images to Generate", step=1, min_value=1, max_value=8, value=1, placeholder="Type a number...")
+        # size = st.selectbox('Select size of the images', 
+        #                     ('256x256', '512x512', '1024x1024'))
+        # num_images = st.selectbox('Enter number of images to be generated', (1,2,3,4))
+        
+        submit_button = st.form_submit_button(label='Generate Image')
+        
+
+    if submit_button:
+        if prompt and number_of_image:
+            with st.spinner("Generating Image"):
+                outputs = replicate.run(
+                "stability-ai/stable-diffusion:ac732df83cea7fff18b8472768c88ad041fa750ff7682a21affe81863cbe77e4",
+                input={
+                "width": 768,
+                "height": 448,
+                "prompt": prompt,
+                "scheduler": "K_EULER",
+                "num_outputs": number_of_image,
+                "guidance_scale": 7.5,
+                "num_inference_steps": 50
+                }
+                    )
+                
+                # for output in outputs:
+                #     st.image(output,caption=prompt)
+                placeholder=st.empty()
+                col1,col2=placeholder.columns(2)
+                for index, output in enumerate(outputs):
+                    if index%2==0:
+                        col1.image(output)
+                    else:
+                        col2.image(output)
+            
+            # response = openai.Image.create(
+            #         prompt = prompt,
+            #         n = num_images,
+            #         size=size,
+            #     )
+            
+            # for idx in range(num_images):
+            #     image_url = response["data"][idx]["url"]
+
+            #     st.image(image_url, caption=f"Generated image: {idx+1}",
+            #              use_column_width=True)

src/page6.py

@@ -0,0 +1,102 @@
+import streamlit as st
+# import openai
+import replicate
+import os
+from dotenv import load_dotenv
+from streamlit_extras.stylable_container import stylable_container
+import streamlit_extras
+# load_dotenv()
+# REPLICATE_API_TOKEN = os.environ.get("REPLICATE_API_TOKEN")
+os.environ['REPLICATE_API_TOKEN'] = 'r8_4fktoXrDGkgHY8uw1XlVtQJKQlAILKv0iBmPI'
+replicate = replicate.Client(api_token='r8_4fktoXrDGkgHY8uw1XlVtQJKQlAILKv0iBmPI')
+
+streamlit_style = """
+            <style>
+            #MainMenu {visibility: hidden;}
+            footer {visibility: hidden;}
+            video{width:200px;}
+            .css-1wbqy5l {visibility: hidden;}
+            .css-15zrgzn {visibility: hidden;}
+            .css-klqnuk {visibility: hidden;}
+            .en6cib64 {visibility: hidden;}
+            .css-1u4fkce {visibility: hidden;}
+            .en6cib62 {visibility: hidden;}
+            .css-19rxjzo, .ef3psqc11 {
+            background-color: purple;
+            text-color: white;
+            }
+            div.stButton > button:first-child {
+            background-color: darkgreen;
+            text-weight: bold;
+            }
+            
+            </style>
+            """
+def page6():
+    with stylable_container(
+         key="title",
+         css_styles=[
+             """ span {
+                text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }"""
+                ,
+                """
+                st-emotion-cache-0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }""",
+                
+                """
+                .e1f1d6gn0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }
+                
+               
+        """,
+        ],
+        
+    ):
+        st.markdown("<h3>Image to Text</h3>", unsafe_allow_html=True) #This is under a css style 
+    st.markdown(streamlit_style, unsafe_allow_html=True) 
+    
+    
+    image_file=st.file_uploader("Select Image", type=['jpeg','jpg','png'])
+    
+    if image_file is not None:
+        placeholder=st.empty()
+        col1,col2=placeholder.columns(2)
+        col1.text("Uploaded Image")
+        col1.image(image_file)
+        prompt = st.text_input(label='Ask question related to image')          
+
+    
+    submit_button = st.button(label='Requestion Answer')
+    
+    
+    
+    if submit_button:
+        if prompt and (image_file is not None):
+            with st.spinner("Recognizing Image...."):
+                output = replicate.run(
+                    "nateraw/video-llava:a494250c04691c458f57f2f8ef5785f25bc851e0c91fd349995081d4362322dd",                    input={
+                        "image_path": image_file,
+                        "text_prompt": prompt
+                    }
+                )
+                print(output)
+                
+                col2.text("Response")
+                col2.markdown(output)
+    
+                

src/page7.py

@@ -0,0 +1,110 @@
+import streamlit as st
+# import openai
+import replicate
+import os
+from dotenv import load_dotenv
+from streamlit_extras.stylable_container import stylable_container
+import streamlit_extras
+# load_dotenv()
+# REPLICATE_API_TOKEN = os.environ.get("REPLICATE_API_TOKEN")
+os.environ['REPLICATE_API_TOKEN'] = 'r8_4fktoXrDGkgHY8uw1XlVtQJKQlAILKv0iBmPI'
+replicate = replicate.Client(api_token='r8_4fktoXrDGkgHY8uw1XlVtQJKQlAILKv0iBmPI')
+
+streamlit_style = """
+            <style>
+            #MainMenu {visibility: hidden;}
+            footer {visibility: hidden;}
+            video{width:200px;}
+            .css-1wbqy5l {visibility: hidden;}
+            .css-15zrgzn {visibility: hidden;}
+            .css-klqnuk {visibility: hidden;}
+            .en6cib64 {visibility: hidden;}
+            .css-1u4fkce {visibility: hidden;}
+            .en6cib62 {visibility: hidden;}
+            .css-19rxjzo, .ef3psqc11 {
+            background-color: purple;
+            text-color: white;
+            }
+            div.stButton > button:first-child {
+            background-color: darkgreen;
+            text-weight: bold;
+            }
+            
+            </style>
+            """
+def page7():
+    with stylable_container(
+         key="title",
+         css_styles=[
+             """ span {
+                text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }"""
+                ,
+                """
+                st-emotion-cache-0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }""",
+                
+                """
+                .e1f1d6gn0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }
+                
+               
+        """,
+        ],
+        
+    ):
+        st.markdown("<h3>Image to Video</h3>", unsafe_allow_html=True) #This is under a css style 
+    st.markdown(streamlit_style, unsafe_allow_html=True) 
+    
+    image_file=st.file_uploader("Select Image", type=['jpeg','jpg','png'])
+    if image_file is not None:
+        placeholder=st.empty()
+        col1,col2=placeholder.columns(2)
+        col1.text("Uploaded Image")
+        col1.image(image_file)                      
+    prompt = st.text_input(label='Enter text prompt for Video generation')
+    submit_button = st.button(label='Generate Video')
+    
+    if submit_button:
+        if prompt and (image_file is not None):
+            
+            with st.spinner("Generating Video. It may require few minutes so please wait...."):
+                output = replicate.run(
+                "ali-vilab/i2vgen-xl:5821a338d00033abaaba89080a17eb8783d9a17ed710a6b4246a18e0900ccad4",
+                input={
+                "image": image_file,
+                "prompt": prompt,
+                "max_frames": 25,
+                "guidance_scale": 9,
+                "num_inference_steps": 50
+                    }
+                    )
+                
+                col2.text("Generated Video from Image")
+                col2.video(output)
+                st.markdown(
+                        """
+                        <script>
+                            const video = document.querySelector('video');
+                            video.loop = true;
+                            video.autoplay = true;
+                        </script>
+                        """,
+                        unsafe_allow_html=True,
+                        )
+    
+                

src/page8.py

@@ -0,0 +1,113 @@
+import streamlit as st
+# import openai
+import replicate
+import os
+from dotenv import load_dotenv
+from streamlit_extras.stylable_container import stylable_container
+import streamlit_extras
+# load_dotenv()
+# REPLICATE_API_TOKEN = os.environ.get("REPLICATE_API_TOKEN")
+os.environ['REPLICATE_API_TOKEN'] = 'r8_4fktoXrDGkgHY8uw1XlVtQJKQlAILKv0iBmPI'
+replicate = replicate.Client(api_token='r8_4fktoXrDGkgHY8uw1XlVtQJKQlAILKv0iBmPI')
+
+streamlit_style = """
+            <style>
+            #MainMenu {visibility: hidden;}
+            footer {visibility: hidden;}
+            video{width:200px;}
+            .css-1wbqy5l {visibility: hidden;}
+            .css-15zrgzn {visibility: hidden;}
+            .css-klqnuk {visibility: hidden;}
+            .en6cib64 {visibility: hidden;}
+            .css-1u4fkce {visibility: hidden;}
+            .en6cib62 {visibility: hidden;}
+            .css-19rxjzo, .ef3psqc11 {
+            background-color: purple;
+            text-color: white;
+            }
+            div.stButton > button:first-child {
+            background-color: darkgreen;
+            text-weight: bold;
+            }
+            
+            </style>
+            """
+def page8():
+    with stylable_container(
+         key="title",
+         css_styles=[
+             """ span {
+                text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }"""
+                ,
+                """
+                st-emotion-cache-0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }""",
+                
+                """
+                .e1f1d6gn0{
+                     text-align: center;
+                padding-top: 0px;
+                padding-right: 0px;
+                padding-bottom: 0px;
+                padding-left: 0px;
+                }
+                
+               
+        """,
+        ],
+        
+    ):
+        st.markdown("<h3>Text to video</h3>", unsafe_allow_html=True) #This is under a css style 
+    st.markdown(streamlit_style, unsafe_allow_html=True) 
+    
+    with st.form(key='form'):
+        prompt = st.text_input(label='Enter text prompt for Video generation')
+        placeholder=st.empty()
+        col1,col2=placeholder.columns(2)
+        video_length = col1.number_input("Enter video length (seconds)", step=1, min_value=1, max_value=8, value=4, placeholder="Type a number...")
+        submit_button = st.form_submit_button(label='Generate Video')
+    
+    if submit_button:
+        if prompt:
+            placeholder=st.empty()
+            col1,col2=placeholder.columns(2)
+            with st.spinner("Generating Video from Text. Please wait...."):
+                output = replicate.run(
+                    "cjwbw/text2video-zero:e671ffe4e976c0ec813f15a9836ebcfd08857ac2669af6917e3c2549307f9fae",
+                    input={
+                        "fps": 4,
+                        "prompt": prompt,
+                        "model_name": "dreamlike-art/dreamlike-photoreal-2.0",
+                        "timestep_t0": 44,
+                        "timestep_t1": 47,
+                        "video_length": video_length,
+                        "negative_prompt": "",
+                        "motion_field_strength_x": 12,
+                        "motion_field_strength_y": 12
+                    }
+                )
+                
+                col1.text("Generated Video")
+                col1.video(output)
+                st.markdown(
+                        """
+                        <script>
+                            const video = document.querySelector('video');
+                            video.loop = true;
+                            video.autoplay = true;
+                        </script>
+                        """,
+                        unsafe_allow_html=True,
+                        )
+    
+                

src/utils.py

@@ -0,0 +1,37 @@
+from io import BytesIO
+from PIL.PngImagePlugin import PngImageFile
+from PIL.JpegImagePlugin import JpegImageFile
+from typing import Union, List
+
+ImageLike = Union[PngImageFile, JpegImageFile]
+
+def resize_image(image: ImageLike, width: int, height: int) -> bytes:
+    """resize image to the given width and height
+
+    Args:
+        image (ImageLike): input image to resize of type PngImageFile or JpegImageFile
+        width (int): expected width of the image
+        height (int): expected height of the image
+
+    Returns:
+        bytes: a new image with the given width and height in PNG format
+    """
+    image = image.resize((width, height))
+    bytes_stream = BytesIO()
+    image.save(bytes_stream, format='PNG')
+    return bytes_stream.getvalue()
+
+
+def get_width_height(size: str) -> List:
+    """get width and height of the image from the given size as a string, for example - 
+        size = '512x512' 
+
+    Args:
+        size (str): size described as '_width_x_height_' example '512x512'
+
+    Returns:
+        List: returns a list of interger as [width, height] extracted from the 
+        given size
+    """
+    # size = '512x512'
+    return [int(val) for val in size.split("x")] # [512, 512]