Creating a quality video with AI – Creating a character

in #tribes6 days ago

$1

Our project is to create a video for the Liotes project. We are now working on the second scene of the project. According to our story board for the second scene of this project we have the following content:

In the second scene we will have Tania Lemaire talking to the camera. She will explain the situation on the planet and how things are evolving.

Creating the character of Tania Lemaire

The problem here is that Tania Lemaire is a totally invented person and we have no image material for her. This means that we will have to 'invent her'. To do that, let's write down what we know about her:

She is the expedition leader on the surface of Liotes. The sound of her name makes her French in my opinion. Since she has a position with responsibility, she is probably of mature age. We are going to feed this information to AI so that it can create an image for us. This is the prompt that I will use:

A portrait of a trustworthy, middle aged woman in a leader position. She is an astronaut of French origin. She wears futuristic cloths. Photorealistic. White background.

For the model, I will use Flux pro which gives quite high quality of images in photorealistic style. I will chose a ratio 1:1 because this is not the end of the process for the creation of the character.

The AI provided the 4 following images for me with this prompt:

$1

The choice is now mine. I think that image number 3 is our Tania Lemaire.

So we have now met Tania Lemaire. I hope you like her. The next step is to set her in a background that we like. I believe the easiest is to put her in a room that might be in a dome in Liotes.

That is the prompt that I have run for the background:

An office in a dome on a different planet. Milky white background.

I used again the Flux pro model and I got the following choice:

$1

I quite like image 1 and image 4. I think I will chose image 1.

Combining the images

In openart, it's possible to merge images together thanks to a model called flux kontext. I put the image of Tanja and the dome as omni-reference and gave the following prompt:

The woman sits at the table in the dome. Close up on her face.

The result is the following:

$1

This is the result of our work and it's the image that I will use for the scene 2:

$1

In the next post, we will make Tanja talk!

Check the previous posts of this serie:


With @ph1102, I'm running the @liotes project.

Please consider supporting our Witness nodes:

Sort:  

Been messing around with video myself, I got started with Leonardo.ai but I should
Try other models, though Leonardo has the API for most other models at your fingertips.

The imagine to video can be tricky with the voice because in veo 3 it’s just generic voices that everyone uses in their videos as you know, so now I gotta use Elevenlabs to get unique voices.

Great idea Turing the Ai Creation process in to content.its surprising how many people don’t want to touch this stuff or think it’s too hard.

Veo 3 is a great model and what I like about it is that it adds all the sound and text directly into the finished product. However as you said it's not possible to use the voice of your choice. I use other platforms for voice than elevenlabs that are completely free to use. The quality is slightly less good.

Great idea Turing the Ai Creation process in to content.its surprising how many people don’t want to touch this stuff or think it’s too hard.

I think that a lot of people don't really grasp what AI can do. That's why I thought it might be a good idea to show the creation process of a video here.

Looking forward to seeing the free audio apps in your posts soon.

I think it's great that you were able to generate a picture that you look. The process is working out well, and it will be interesting to see how you make her talk.

The talking aspect is the most complicated one since AI often doesn't manage to do it that well.

It looks like Angelina Jolie to me. Eyes and lips. 😎

There is definitely some resemblance there :-)

Oh wow, having to combine two images into one is new to me. Interesting and cool image outcome, the process is unfolding nicely :)

It's something I have discovered recently and it opens tons of possibilities.

It's fascinating the things that can be done with this new technology.

I believe so too and every week, it's possible to do some more things.

This next animation ought to be interesting. Will we see facial movements? Or just some background movement?

The idea is to have the character talking on the camera with facial movements and lip synchronization. This is however a tricky part still for AI.