Custom Mega Workflow V3
Image has workflow embedded.
Requires the following custom nodes:
Fitsize (Size adjustment)
rgthree (Fast Group Bypasser, switches, Power Lora Loader)
Resolution Master (T2V resolution selector)
Video Helper Suite (video output, t2v frame removal)
mxtoolkit (sliders)
ComfyUI Extended (math)
Note: Fitsize keeps the video at 640 pixels based on the longest side of the image. Adjust if you want it less or more (eg: a 1024x2048 image will result in a video that's 320x640)
Extended will round the input image width and height to the nearest 16 pixels. I had problems with images with weird resolutions.
Edit: I forget to remove Vextra Nodes' Play Sound at Execution node, either just delete it, or install Vextra Nodes and set the location to an audio file you want to play upon completion of your generation.
I personally use a copy of "C:\Windows\Media\Windows Unlock.wav"
V3
Adds Power Lora Loader.
Renames Frame Length to Video Length (in frames).
Colours the Model Loaders.
Simplifies the Video Combine filename prefix to "Wan22-Mega".
V2
Removes EasyUse for built-in primitives.
Adds ComfyUI Extended for rounding math.
Adds VHS Split frame to remove frames at beginning of video (0 for i2v, 3 for t2v).
Adds Video Dimension slider.
Hello and thanks, where I can control the lenght? Or I can't?
Hello and thanks, where I can control the lenght? Or I can't?
i'd start with the "frame length" slider :P
and yes, that is a interesting name. It's a bit too custom for me. The official one seems to work, except there was the video making part missing.
I actually missed the VideoCombine node because I was using "Preview Image", as it wasn't saving all the junk I was testing generations with. Also, it made it easier to fit the example image (VideoCombine is tall). That is why the VideoCombine didn't make it in the provided workflow (yet), ha!
well you can always print the preview images out and make a flip book (or whatever they are called).
Hello and thanks, where I can control the lenght? Or I can't?
i'd start with the "frame length" slider :P
and yes, that is a interesting name. It's a bit too custom for me. The official one seems to work, except there was the video making part missing.
Yeah, I guess the name's a bit weird. I think I was combining 'number of frames' and 'length' in my head. :-P
I probably should have made the Video Combine filename 'simpler' as well.
Thanks for the workflow! I tried using it with wan2.2-rapid-mega-aio-nsfw-v12.2.safetensors, but I'm getting an issue where the character just shakes/jitters in place and doesn't follow the prompt at all. Is this workflow compatible with the latest model, or am I missing something?