Sign In

Making a Advertisement Video with Stable Diffusion and chatGPT

MisterMR
Bronze Writer Badge
Mar 3, 2024
generation guide
videosstable diffusion video
Making a Advertisement Video with Stable Diffusion and chatGPT

Objective

Making a script and Video using AI technology

Preface

I never made and advertisement in my life, this is a test to have and objective to learn the tools

Tools

Checkpoint

Writing the script

My idea is to write and easy AD for a burger company, I want to show the burger with chips.

I will use ChatGPT to write my script, here is the question:

I want to make a video advertisement of my burger with chips, give me some script and ideas

ChatGPT answer

Title: "Savor the Flavor: [Your Burger Joint's Name] Signature Burger & Chips"

[Opening shot of a sizzling grill, with the sound of meat cooking. The camera slowly zooms in as the narrator begins speaking.]
Narrator: "Welcome to [Your Burger Joint's Name], where every bite tells a story of flavor and satisfaction."
[Cut to a close-up shot of your signature burger being assembled, showcasing fresh ingredients and juicy patties.]
Narrator: "Introducing our mouthwatering signature burger, crafted with the finest ingredients, seasoned to perfection, and made with love."
[Showcase shots of each component of the burger being layered together: the toasted brioche bun, the juicy beef patty, melted cheese, crisp lettuce, ripe tomatoes, and tangy sauce.]
Narrator: "Each bite is a symphony of flavors, a harmony of textures that will leave your taste buds dancing with delight."
[Transition to a shot of golden, crispy chips being lifted out of a fryer, steam rising.]
Narrator: "And what's a burger without a sidekick? Our perfectly crispy chips, hand-cut and seasoned just right, are the ideal companion to our delicious burgers."
[Showcase shots of the chips being sprinkled with seasoning, served in a paper cone, and paired with a dipping sauce.]
Narrator: "Whether you're craving a classic cheeseburger or our specialty creation, we've got something for every appetite."
[Cut to a shot of a group of friends or family enjoying their burgers and chips together, laughter and smiles filling the air.]
Narrator: "So come on down to [Your Burger Joint's Name], where flavor meets fun and every bite is an experience worth savoring."
[Closing shot of your logo with the tagline: "Indulge in Flavor, Taste the Difference."]
Narrator: "[Your Burger Joint's Name] – Where Every Bite Tells a Story."
[End with contact information and location details.]
Ideas:
1. Consider showcasing the ingredients being sourced locally or the cooking process in your kitchen to emphasize freshness and quality.
2. Highlight any unique or specialty burgers on your menu to intrigue viewers and encourage them to visit your restaurant.
3. Incorporate shots of satisfied customers enjoying their meals to create a sense of community and satisfaction.
4. Use upbeat background music that matches the energetic and inviting tone of the advertisement.
5. Incorporate vibrant visuals and close-up shots of the food to make viewers' mouths water and entice them to visit your restaurant.

Generate videos

I will generate many videos to have several to assemble

I will use a simple resolution 512x512 to keep it fast .

Sampler: euler a

Sampling steps: 20

CFG scale : 7

I set the batch count to 5: to generate at least 5 videos for each prompt.

I will keep the seed to -1 to be random.

Animatediff settings

I will use the mm_sd_v15_v2.ckpt model, with mp4 as output (for the article purpose also webp)

16 frames for the video with 32fps because I will use the "Film" option with 8 interpolation frames between each generated frames. This will make the video more fluid. (interpolation uses the deforum component, you need to install it)

Generating the videos

I will use directly the ChatGPT suggestions

1° prompt: Opening shot of a sizzling grill, with the sound of meat cooking. The camera slowly zooms in as the narrator begins speaking.

2° prompt : Cut to a close-up shot of your signature burger being assembled, showcasing fresh ingredients and juicy patties.

3° prompt: Showcase shots of each component of the burger being layered together: the toasted brioche bun, the juicy beef patty, melted cheese, crisp lettuce, ripe tomatoes, and tangy sauce.

4° prompt: Cut to a shot of a group of friends or family enjoying their burgers and chips together, laughter and smiles filling the air.

5° prompt: Closing shot of your logo with the tagline: "Indulge in Flavor, Taste the Difference.

1° prompt

Opening shot of a sizzling grill, with the sound of meat cooking. The camera slowly zooms in as the narrator begins speaking.

2° prompt

Cut to a close-up shot of your signature burger being assembled, showcasing fresh ingredients and juicy patties.

3° prompt

Showcase shots of each component of the burger being layered together: the toasted brioche bun, the juicy beef patty, melted cheese, crisp lettuce, ripe tomatoes, and tangy sauce.

4° prompt

Cut to a shot of a group of friends or family enjoying their burgers and chips together, laughter and smiles filling the air.

I did run several times because faces were not good. I use I Adetailer to improve the faces (check my article about this : https://civitai.com/articles/3887/stable-diffusion-testing-adetailer-on-videos-with-different-options)

I have generated many videos but I kept 9 that were quite good.

5° prompt

Closing shot of your logo with the tagline: "Indulge in Flavor, Taste the Difference.

Making the logo

Let's ask to ChatGPT how I can call my company

give me 5 titles for my burger company

ChatGPT Answer are:

1. Bite Bliss Burgers

2. FlavorFusion Burgers

3. CraveCraft Burgers

4. PrimePatty Burgers

5. SavoryStack Burgers

I like Bite Bliss Burgers

I select DynaVision XL checkpoint (https://civitai.com/models/122606/dynavision-xl-all-in-one-stylized-3d-sfw-and-nsfw-output-no-refiner-needed?modelVersionId=297740)

because XL it is better with text and generate 9 images with:

Prompt:

Bite Bliss Burgers Logo icon with text "Bite Bliss Burgers "

Negative:

blurry, blurry_image, Lowres, Low_resolution, Low_picture_quality, Low_picture_anime, extra_anatomy, extra_body, extra_navel, extra_face, extra_eyes, extra_chest, extra_nipples, extra_hips, extra_arms, extra_hands, extra_fingers, extra_legs, extra_feet, extra_toe, missing_anatomy, missing_body, missing_navel, missing_face, missing_eyes, missing_chest, missing_nipples, missing_hips, missing_arms, missing_hands, missing_fingers, missing_legs, missing_feet, missing_toe, sketch, sketching, (worst quality: 2), (low quality:2), ((monochrome)), ((grayscale)), inpainting, HDR, bad_prompt_version2, jpeg artifacts, signature, watermark, text, error, missing fingers

I get my results

My Choise is this one:

I use Krita to remove the background , selecting the white with "contiguos selection tool" .

Save the file in the logo.png in my project folder.

Rename the files

To avoid confusion I rename the files with p1_x.mp4,p2_x.mp4,etc (where x is progressi)

Example for the first prompt p1_1.mp4,etc..

Making the final video

I run KdenLive to assemble the best videos

Drag an drop the video in your project, it will ask to change the dize to 512x512, clic Switch.

If doesn't ask , click on one video and click menu Project >>"Adjust profile to current clip"

I drag and drop all the video that I like

At this point using Kdenlive I assemble the images and the logo

I kdenlive I like to use fadein and fadeout, and transform for changing the size of the images/videos.

This time I won't add music or sounds, my idea was to concentrate with the graphics.

For audio voices normally I use : https://misterm.itch.io/mr-text-to-speech that uses openAI speech to text.

For music I normally use music free from internet, but you can also try with https://www.riffusion.com/ (check my article about a little video with music https://civitai.com/articles/4225/making-a-video-with-lyrics-and-music-with-chatgpt-stable-diffusion-and-riffusion)

To generate the full video go to Project >> Render...

I hope you enjoy my article, it was fun.. I learn many software and settings to produce the final video.

Check the output video here : https://civitai.com/posts/1601865 (512x512 mp4 version 32fps)

Here is a 320x320 8fps in wepb format to be less than 5 MB.

21

Comments