This workflow add animate diff refiner pass, if you used SVD for refiner, the results were not good and If you used Normal SD models for refiner, they would be flickering. So AnimateDiff is used Instead.

What this workflow does

Add more Details to the SVD render, It uses SD models like epic realism (or can be any) for the refiner pass.

Also Suitable for 8GB Ram GPUs.

How to use this workflow

All Nodes are organized and Grouped under headings and proper naming format inside the Workflow with How to use instructions.

1) Use SVD as usual. Inside the workflow

• Upload starting image
• Set svd or svd_xt
• Set fps, motion bucket, augmentation
• Set Resolution (it's set automatically but you can also change acc. to your hardware capacity)

2) Set Refiner Upscale Value and Denoise value

• Use a value around 1.3 - 1.7 in the Refiner Upscale to give a little room in the image to add details.
• Use 0.4 - 0.7 in the Denoise for Best results. The control Net Softedge is used to preserve the elements and shape, you can also use Lineart)

3) Setup Animate Diff Refiner

• Select the model you want to use as the refiner (Eg epic realism, mistoon, realisitic vision) which ever you think suits the best for the image ie cartoon, anime or realisitic
• Select Loras if you want to use them, They work as usual , (Some loras are not compatible with AnimateDiff so you can experiment with them
• Type in the Positive and Negative prompts for more accuracy ( General Words are setup by default, Optional if you want to change)

Go Play!

You can Find the Demo with Documented info about this workflow and it's samples here  :  https://www.patreon.com/posts/ai-svd-with-more-93812677

You can also convert this workflow into SDXL for more details, I have on 8GB vram so I can only test with SD.

My PC Specs:

AMD Ryzen 9 6000 s

32 GB Ram

RTX 3070 Ti - 8GB Laptop GPU

Model - Asus Rog flow x16

Custom Node List:

1. https://github.com/WASasquatch/was-node-suite-comfyui
2. https://github.com/Suzie1/ComfyUI_Comfyroll_CustomNodes
3. https://github.com/shiimizu/ComfyUI_smZNodes
4. https://github.com/Kosinkadink/ComfyUI-Advanced-ControlNet
5. https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved
6. https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite
7. https://github.com/giriss/comfy-image-saver
8. https://github.com/kijai/ComfyUI-KJNodes
9. https://github.com/YMC-GitHub/ymc-node-suite-comfyui
10. https://github.com/Stability-AI/stability-ComfyUI-nodes
11. https://github.com/pythongosssss/ComfyUI-Custom-Scripts
12. https://github.com/LucianoCirino/efficiency-nodes-comfyui
13. https://github.com/wolfden/ComfyUi_String_Function_Tree
14. https://github.com/Stability-AI/stability-ComfyUI-nodes
15. https://github.com/AlekPet/ComfyUI_Custom_Nodes_AlekPet
16. https://github.com/FizzleDorf/ComfyUI_FizzNodes
17. https://github.com/mcmonkeyprojects/sd-dynamic-thresholding
18. https://github.com/giriss/comfy-image-saver

--------------------------------------------------------------------

PYTHON SCALER ERROR FIX :

For someone getting the only one element tensors can be converted to Python scalars You just have to delete the Dynamic Thresholding node from the workflow and it will run fine, without any errors. Link for photo reference: imgur.com/a/9bBNPDo

The Recent Update broke the dynamic thresholding node :/

My Discord Server : https://discord.gg/z9rgJyfPWJ