Creating AI art with neural networks is getting easier and easier!
If you haven’t seen our most recent guide: Make Stunning & Strange Artificial Intelligence (AI) Art [Beginners Guide], you might want to check that out first. In this step-by-step guide, we’ll show you how to make coherent ai video art quickly and easily. This tutorial uses Disco Diffusion Turbo and results in coherent and stunning, surrealistic ai generated video art featuring the content of your choice with neural networks.
Let’s get started!
Step 1: Open the Disco Diffusion Colab in a new tab:
You will see the page below:
Step 2: Confirm the proper notebook is selected. Click the “Connect” drop-down and then select “View resources”
Select GPU and High-RAM, and then click “Save”.
Step 3: Scroll down to section “3. Settings” in the notebook.
- batch_name: Change the batch name to [Something Descriptive for your project]
- width_height: You will also want to change the width_height here. For widescreen, you can do 1280 width by 720 height. Or for Instagram Stories or YouTube Shorts, you will do 720 width by 1280 height. Instagram Stories and YouTube shorts size is 1080px by 1920px but Google Colab will not be able to make videos of this size. We will use an AI Video upscaling later in the tutorial.
- Change cutn_batches to “1” if you are making a video.
Step 4: Input your text prompts at the bottom of Section 3:
Hints for text prompt inputs:
- the # in front of the 2nd two prompts is canceling those out. For a video whose content evolves removed the # from the 3rd prompt.
- You can set the prompt weights with a “:” This is a special character and must be followed by a number.
- Try adding <content> in the style of <your favorite artist> to implement their styles.
Try adding the following bits of text:
- art station
- trending on art station
- Unreal Engine
- beautifully lit
- 8K 3D
- 8k resolution
- matte painting
- highly detailed painting
- at sunset
Understanding Keywords – Extra Credit Resources:
Step 5: Uncheck skip_video_for_run_all.
Step 6: Click Runtime > Run all to run all the cells in the program and make your CGI digital video art.
It will now prompt you to connect with Google Drive and you will want to do this.
It will now take some time to run. After completion, you will have images and a video in your Google Drive folder: /AI/Disco_Diffusion/images_out/Something Descriptive/
The mp4 video will be about 20 seconds long.
Optional Bonus Steps:
We would recommend using an AI video enlarger like Topaz Labs Video Enhance to make the video bigger. Video Enhance will sharpen and grow the size of the picture without making the video more blurry like traditional tools.
If you like how it looks, and you want to make the video longer, you can start another video on the final frame of your last video. To continue the video, proceed to the next step.
Step 7: To continue the video you will need to locate the last frame of the clip from the completed AI output in Google Drive:
Step 8: Back in Section 3. Basic Settings we’re going to update the batch_name:
Step 9: At the bottom of Section 3: Basic Settings paste the path to the last frame.
Step 10: Now you can run the Disco Diffusion AI model again to continue the clip and then stitch together the results.
Upscale Your Video:
Topaz Labs develops popular software for upscaling both photos and videos. Remini is another popular option.
That’s it! Share your awesome videos in the comments. If you’re interested in an overview of easy to use ai software check out this guide: Make Stunning & Strange Artificial Intelligence (AI) Art [Beginners Guide]