Deforum Stable Diffusion is a free and open source that allows you to create animated videos. You just need to provide prompts and settings for how the camera moves. In this article, AnonyViet will introduce you to Deforum Stable Diffusion and how you can use it to create unique and impressive artistic videos.
Join the channel Telegram belong to AnonyViet ???? Link ???? |
What is Stable Diffusion?
Stable Diffusion (SD) is a text-to-image tool launched by StabilityAI on August 22, 2022. Today, millions of people use it to create beautiful photos in seconds. Stable Diffusion is built with the goal of helping developers and users create unique and creative images, using just simple text commands.
What is Deforum Stable Diffusion?
Deforum is a tool for Stable Diffusion that allows to create engaging animated transition videos by generating images from stable images with high quality.
Deforum Stable Diffusion offers many features and options for users to create animated videos that flexibly transform according to your preferences.
How to create a hot trending transition video with Deforum Stable Diffusion
Prepare necessary images, Model, Lora
Step 1: Proceed to run Google Colab to be able to access Stable Diffusion’s webUI. If you do not know how to do it, you can follow the instructions through the following article
>> Instructions for using Stable Diffusion are very simple <<
Step 2: In the meantime, download the necessary model and lora tools for making transition videos
You can choose to download any Lora you like, not necessarily the one above. Because here I will follow the style of transforming into dragons, monsters, vampires, .. so I will download Dark Red Vampire.
Step 3: After downloading, go to Google Drive > Open the folder “SD – Data”
- Open the Lora folder > Paste the file “darkRedVampire_v1.0…” downloaded earlier
- Open the Model folder > Paste the file “revAnimated…” into
Step 4: Prepare the image you want to transform. If you use video, take a screenshot of the last scene (the scene about to transform).
To know the size of the image, right-click > Properties > select the “Details” tab, scroll down and you will see how much the width and height of the image is, you should remember this parameter so that you can fill it correctly later. Please.
You can use this image right away to try it out:
Download Deforum Stable Diffusion
After Google Colab outputs 4 links, click on the 1st link to access Stable Diffusion’s WebUI interface.
Step 1: Click the “Extensions” tab > Select “Available” > Select “Load from”
Step 2: You drag your mouse down to see the line “Deforum tab, animation”, click “Install” to start the installation
Step 3: Scroll back up, select the “Installed” tab > “Apply and restart UI”
At this point, you will see an additional tab “Deforum” appear.
Step 4: In the Stable Diffusion checkpoint at the top left of the screen, select the correct model “revAnimated…”. If you do not see that model, you can click the green button next to the system to refresh the page, then select it again.
Fill in the parameters
At the Run tab, enter the parameters as shown:
- Sampler: DPM++ SDE Karras
- Steps: 25
- Width, Height: Width, height of the image. Above I have instructions on how to determine the size, you follow and enter the number here.
- Remember to tick the “Restore Faces” box too
At the Keyframes tab, enter the following parameters:
- Animation mode: 3D
- Border: wrap
- Max frames: 600
- Strength schedule: 0: (0.65), 25: (0.55)
- Translation X 0:(0), 30: (15), 215: (15), 300: (0)
- Translation Y 0:(0)
- Translation Z 0: (0.2), 60: (10), 300: (15)
- Rotation 3D X 0:(0), 60: (0), 90: (0.5), 180: (0.5), 300: (0.5)
- Rotation 3D Y 0: (0), 30: (-3.5), 90: (-2.5), 180: (-2.8), 300: (-2), 420: (0)
- Rotation 3D Z 0:(0), 60: (0.2), 90: (0), 180: (-0.5), 300: (0), 420: (0.5), 500: (0.8)
In the Noise section:
- Noise schedule 0: (-0.06*(cos(3.141*t/15)**100)+0.06)
In the Anti Blur section:
- Amount schedule 0: (0.05)
In Depth Warping & FOV:
At the Prompts tab:
You fill in the following:
{
“0”: “1 girl”,
“130”: “
“290”: “
“480”: “
}
((best quality)), ((masterpiece)), (detailed), (detailed CG unity 8k wallpaper), (realistic)demon girl, Mythical creature with scaly skin, fiery breath, and deadly claws, emanating a sinister malevolence and intimidating aura, inspiring terror in those who encounter its fearsome form
EasyNegative, paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, glans, nsfw, (outdoor:1.6), backlight,(ugly:1.331), (duplicate:1.331), (morbid:1.21), (mutilated:1.21), (tranny:1.331), mutated hands, (poorly drawn hands:1.5), blurry, (bad anatomy:1.21), (bad proportions:1.331), extra limbs, (disfigured:1.331), (more than 2 nipples:1.331), (missing arms:1.331), ( extra legs:1.331), (fused fingers:1.61051), (too many fingers:1.61051), (unclear eyes:1.331), lowers, bad hands, missing fingers, extra digit, (futa:1.1),bad hands, missing fingers
Get image path
Step 1: Back to the Google Colab page, click on the folder icon as shown > drive > MyDrive
Step 2: Click the 3 dots, choose to create a new folder, for example, here I name it “Pictures”
Step 3: Click the 3 dots in the “Images” folder > Select “Upload” > Select an image to upload
Step 4: Click on the 3 dots next to the image file > Select “Copy path” to copy the path
Step 5: Then, you return to the WebUI screen, select the “Init” tab:
- Check the box “Use init”
- Paste the previous path into the “Init image” box.
At the Output tab:
- Southtrack path: Is the audio path of the video. How to upload music files is similar to uploading photos, you also do the same as the instructions above. Here I will leave it as default.
- Because video is a continuous arrangement of hundreds of photos, to avoid heavy machine, you should check the box “Delete imgs” so that it deletes those images after creating the video.
Finally, click “Generate”, wait about 5-10 minutes for Stable Diffusion to output the video.
You just need to access Google Drive > SD – Data > Export > img2img to enjoy the finished product, the video will have a .mp4 extension.
The finished video you can watch here:
Epilogue
So I showed you how to create a transition video with Deforum Stable Diffusion extremely simple. The prompts, parameters, models, etc., you can customize flexibly to your liking to create more unique videos. Good luck!