Stable Video 3d(SV3D) has been released on March 18th, 2024 by StabilityAI. This model is based on the Stable Video Diffusion gives more grip in the field of 3D generation with an improved performance in terms of consistency and quality. It includes SV3D_u (creates orbital video without camera angle shot) and SV3D_p (creates 3d video merging images and orbital angle).
If you want to go into the details of the model then research paper are already prepared for you. The model can be used for Commercial purpose by joining the Membership plan and for Non Commercial purpose it can be downloaded from the official Hugging face Repository.
Steps to install in ComfyUI:
1. First navigate to the official Hugging Face page provided by StabilityAI.
2. Now, sign into hugging face account and accept the agreement to use the Stable Video 3d model for non commercial purpose. But, if you don't want to get around those marketing email then you can unsubscribe from your email bottom section.
After accepting you will get a message "You have been granted access to this model".
3. Click the "Files and Version" section and move to the files section.
And download the "sv3d_u.safetensors" file which helps to make a 3D orbital video. After downloading, you need to save into the ComfyUI's models folder as usual.
4. Now you need to download "ComfyUI-KJNodes" and "ComfyUI-Frame-Interpolation" from the ComfyUI Manager. Download and install these two nodes by searching from the search box.
Or alternatively you can also download ComfyUI-KJNodes and ComfyUI-Frame-Interpolation from the its GitHub repository.
5. After downloading, just update and restart your ComfyUI with closing the command window to take effect.
If you are getting any custom nodes missing errors then, simply from ComfyUI manager install the missing nodes by clicking on the "Install custom nodes".
6. Next is move to the ComfyUI workflow which we have provided into our Hugging Face platform to get the best out of it. Download the file "ComfyUi-Nodes-Workflow.json" file and add into ComfyUI to take effect.
https://huggingface.co/datasets/stablediffusiontutorials/Stable-Video-3d-ComfyUI-nodes/tree/main
Tips to get Best output:
1. First upload your image by click on the "Choose the file to Upload" button.
2. Select the checkpoints to "sv3d_u.safetensors" from the list. Leave the rest of the settings as it is because the ComfyUI workflow already set it up.
3. If you want you can play with the K-Sampler settings for perfection.
4. Change the frame per second(fps) to around 12 as value to get the perfect smooth 3d orbital effect.
5. This is the result used with Stable video 3d:
Conclusion:
Stable Video 3d model has been inspired by Stable Video Diffusion is more improved when compared with Stable Zero-123. It can be used in multiple use cases like product showcase in ecommerce platform, social media marketing, Game development.
You can use it for non commercial purpose from the Hugging face repository but commercially it can be accessible by joining their membership plan.
