Hyper Realistic Aitana Lopez example
Create Hyper Realistic AI Influencers (Like Aitana Lopez)
Create Hyper Realistic AI Influencers (Like Aitana Lopez)
Video: https://www.youtube.com/watch?v=6pzYxDq5bUE
THERE ARE THE 5 TOOLS THAT YOU WILL USE:
1. Create AI Model with SeaArt: https://www.seaart.ai/
2. FaceSwap of model with SeaArt: https://www.seaart.ai/
3. Reels and Videos with Canva: https://www.canva.com/
4. Create Deep Fake Video with: ROOP.ipnynb Google Collab: https://cutt.ly/wwxhYjmM
(Note: ipynb stands for “Interactive Python Notebook”, the file format of Jupyter Notebook. A superb feature is its ‘cell-based interface’. This allows users to see the output of each individual block of code as it is executing.)
5. Create Deep Fake Audio with: Lalamu Studio for Lip-Sync of AI influencer model https://lalamu.studio/demo/
Transcript of Video: https://www.youtube.com/watch?v=6pzYxDq5bUE
AI influencers are now trending all over the world. Spanish AI influencer Adiana Lopez has taken the Internet by storm. She is earning more than $11,000 per month. She is prompting products of various Brands. Her Instagram account has more than 1 million followers. Many popular magazines featured her in their editorial. This Video is about how you can male an AI model like Adiana Lopez.
1. IMAGE GENERATION PROCESS
First, we need to create a 'digital model' -a model that won't exist in the real world.
Use an AI image generator. You can use any image generato. We will use SeaArt for this example as it's a free AI art generator.
Go to the SeaArt Ai website (https://www.seaart.ai/) and LOGIN. (Create an account at SeaArt.ai (Use your Discord, Google, Facebook or Email account).
After creating an account go to the search bar and then search for images of a model by typing for example; “I want to make a model with blonde hair”.
The search will return a vast number of images with blonde hair. Select the image that you like.
After selecting one of them Click on the ‘Try’ button.
At the bottom of the page you will notice the prompt. This prompt is basically the prompt of the image we selected.
Click on the 'send' button. The Image generation process will start. SeaArt generates two images. We will use just one as our model. You can upscale or make variations of the selected image.
Click on the 'variation' button and you can customize the prompt. We'll do it without any customization. SeaArt generates two variation images of our model.
Once you have the images, 'upscale' one of them. Select the image and Click on the 'upscale' button.
Set the upscale ratio (You can keep it at ‘2’) and Click on the generate button.
Save the image on your device. Select the image right Click and then select; save image as; and give the file a name.
If you want to generate a different style image, copy the Prompt and paste it into the prompt box and then customize the prompt.
After customization, Click on the ‘generate’ button and your images are ready. This way you can customize unlimited images for your AI influencer model.
2. FACE SWAP
Now we must use the 'Face Swap' feature to make the faces the same. Although all of our images may look the same, the faces are a bit different so we will use the Face Swap feature of SeaArt to solve this issue.
Select the Face Swap tool.
Upload the 'Targeted image' first.
Next upload the 'Reference' image (this is the image that you want the 'target image' to look like.
Click on the Generate button. We can generate unlimited images with the same face for our AI model.
Now that the images are ready, we can create Reals or Videos.
If you want to make a professional AI influencer, we have to upload Reels or Videos.
3. MAKE VIDEOS
To upload Reels or Videos:
Open a new tab on your browser and log into Canva at https://www.canva.com/
(If you are new to Canva, create an account. you can enjoy a 30-day free trial of Canva)
Login to Canva and select the Video tab.
If you want a video feed on Facebook or YouTube shorts, then select 'mobile video'.
If you want to make Instagram reals, then select 'Instagram reals'.
On the Canva creator page, go to the ‘Element Section’ and from the search bar, search for a blonde girl. (Because we used a girl for our model who has blonde hair).
Go to the Video tab. You will notice a vast number of blonde girl’s video clips. Find a video with a similar body shape and hair color to our model. Insert the video into the canvas. Adjust the size and positions and then download the video.
Add more videos. When you have added some more videos, download them. Click on the share button and select ‘download’.
Note that when you Click on the 'download Pages', you get all your files in one ‘zip file’. We have to unzip the file to use the Videos.
4. CREATE DEEP FAKE VIDEOS
Now we have to change the faces in the downloaded videos to match our model exactly. We use Deep Fake technology for this.
There are many deep fake generator AI tools available on the internet. You can use any one of them. We will use ROOP on Google Collab. Go to the Link for ROOP.ipnynb Google Collab: https://cutt.ly/wwxhYjmM
Click on the ‘connect’ button to connect your notebook to GPU server(s).
After establishing the connection, Click on the directory icon and then select the upload icon.
Upload the targeted videos and our model's source image.
Before uploading the files make sure your file names do not contain any spaces or special characters otherwise it won't work. (Simply rename the files without any space or special character).
Upload the files into Google Collab.
Go to clone ROOP and install dependencies and Click on the ‘Run’ button to sun the cell (module).
After completing the install dependencies, go to the download model step and Click on the ‘Run’ button to execute to cell (this may take some time as the GPU must do some heavy processing).
After completing the processing. go to the final step. We have to make some changes in this step. Go to the notebook directory and Click on the three dot menu of your 'target video file' then select ‘copy path’ . This will copy the file path for pasting into the 'target' file location in the code panel.
Paste your target file location and then repeat this process for the ‘source image path’ .
After target and source files are all set, Click on the ‘cell run’ button. The process will take some time. Once this process is completed, go to the directory - a new file named ‘swapped’ will be added. (If you are unable to find the file, just refresh the directory).
Click on the three dots menu and select ‘download’ to save the file on your device. Our first Real video is ready. We have an AI generated Instagram model who can do anything.
If you want to create more than one video at the same time, just go back to the Google Collab and select your next Target videos. Copy the target video path and paste it into the code. Then Click on the ‘Run’ button and your videos are ready.
We can create images and videos with the same character for our AI influencer model.
5. CREATE LIP-SYNC
Now we want our model to talk about something. For this we will use ‘Lip-Sync’ technology. There are many tools available for lip-sync.
First, generate a female voice with AI or record your own voice.
After generating the Voice, go to Lalamu Studio website; (https://lalamu.studio/demo/)
Lalamu Studio is basically an app for Android and iOS. The web version is a demo version as it is still under development - but you can try it.
After coming to the Lalamu demo page, you find some sample videos and Sample audio.
To create lip-sync with your model Click on the upload button to upload the Video.
After uploading your video file, go to the audio section and upload your audio file.
Select the uploaded audio file and then Click on the generate Lip-Sync video.
You can do this lip-sync on your Android and iOS as well.
Creating an AI influencer model is now complete. We already made a video on how you can promote products with an AI influencer you can watch it from the AI Lockup channel.
Last updated