FilmBot v1

Hi everyone, this is the evolution of my previous workflow, now with SDV, ControlNets, IpAdapter and the ideia to Split the latent in the middle of the process to add more motion maintaining coherence trough the shot adding the loop effect. There is a simple Upscaler for better results at the finals passes.

I now will work in enhance the quality and fix the Latent Colorful glitch (You will notice the secod pass are more "shining colorful" because, I think, of the latent split.)

How this work:

Prompt > Image > img2video SDV > ControlNet > AnimateDiff > Latent Split (You jump this if you want) > AnimateDiff "Glue" > AnimateDiff Refinement > Extend Frames with RIFE VFI interpolation > Upscale Latent > Refine > ControlNet Checkpoint > Final Refiner

It's slowwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww. I'm trying to figure out how I can make it more faster with less steps and better quality.

I forgot to connect the Latent Split, just do it:

Checkpoint:

AnalogMadness:  https://civitai.com/models/8030/analog-madness-realistic-model

AnimateDiff Lightning:  https://huggingface.co/ByteDance/AnimateDiff-Lightning

Lora:

epi_noiseoffset (Optional): https://civitai.com/models/13941/epinoiseoffset

I can't find anymore the other two LoRAS links, so in the next update I will erase them.

There are 3 Lora Slots, bypass or use anything you want there.

Work well with:

AnalogMadness

DreamShaper

Try with others chekpoints and share results to make this list bigger.

One more thing, I make this name FilmBot just for fun, it's is just a workflow, okay?

Next Steps for next version:

- Make Vid2Vid Done

- ClipSeg Masking Done

- Face Detailer (too slow)

- Less Steps Done

- 10 seconds (240 frames) Unlimited

- Better Ipadapter settings Done

- Motion Lora working Done

- Wave2Lip

- WD14-Tagger

- Better FreeU_v2 Settings

- Better AnimateDiff Lightning Settings Changed to LCM

- Better Latent controlers with Lerp or Slerp Aborted

- Less Slow Motion Soon with natural motion capture you will able to make commercial ready scenes

- Make Prompt Travel Batch work trough all steps

- Find better LoRAS Done

- Way to transform the 512x512 to 1024x576 without distortions Done