EasyMedia - Long Video
All credit goes to the Node Developers
Very easy to use
Multiple Shots with stitching in one go
I2V, F2V, Ref2V with custom audio and video support
Single Pass / Multi Pass
Latent UpScale
Audio preview while viewing the video preview so you can hear the audio generation and can stop the current execution in mid. Same like video preview. Press the speaker icon in the Video Preview panel
MultTrack Editor node let you have multiple segments each with multiple audios and images. You can have different image for each segment. Read the official documentation for huge knowledge and features.
Full Layout
This will break often, couldn't fix it - You have to attach again
Select samplers and scheduler, steps dont work here. Steps are based on sampling_plan if you turn both 1st_pass_only and disable_2nd_noise then ultra_light = 8 steps. Its bit dynamic and fully internal based on what you select.
Video Segmentation
Suppose you have 4 segments
segment_tart_numberis where from the VIDEO RENDERING will startsegment_countis how many segments you want to do. If it is-1it will render from start. If it is 2, it will do 1st and 2nd segment onlyproject_save->overridewill keep only one video.newwill keep all videos and you can view them in theMultiTrack Project Video Combinenode
To Add media and prompt for each
Press + to add segment
Click on segment and middle portion will show Add media
Add media select multiple images
Click on segment again I guess to see the prompt option
Long shot video
Select
matchormaxbased on your requirementsSelect R2V for reference
Extend options will only be visible on segment 2 and onwards. It will not show on first segment.
Add Audio, Video to workflow
Custom Nodes:
Easy Media Node
https://github.com/yolain/ComfyUI-Easy-Media or if you have issues node not working properly check the below one. I dont own any
Different fork (https://github.com/deciia/ComfyUI-Easy-Media) suggested by https://civarchive.com/user/Tar_Tarj the
yolianwas having issues with his comfy version.
Description
FAQ
Comments (24)
This seems promising. Can you please provide some simple instruction how to get this started. For starter, I can't seem to locate any prompt boxes to enter what I want to generate. Not to mention where to add reference pics. Am I missing something?
Yes its complicated nodes, it took me sometime to fully understand this. I have attached another image so you can see where the prompts are
Read the GITHUB repo for full detail on each option. Each node options it self give tooltips and hints.
Hello I've been testing this Workflow and I understood how to put the prompt but I didn't understand how to Make the second video Use the last frame or the last context of the previous video To make it as if it were one long video, and I don't know if it just is possible. Uh. Can you tell me some advices? Or if it is possible? Thanks for the answer ! (Sorry for my English. I'm French, so that's not the best😊)
I understood that you can put different prompts for each 5 sec and different sequences. I've succeeded to make a long video. More than 5 seconds. But after these five seconds. It doesn. Look like the same video from the first video!
@Velmil I have added some instructions on the page. I hope they will help
EDIT EDIT! : this fork of Easy Media will work with the latest Comfyui build https://github.com/deciia/ComfyUI-Easy-Media
EDIT:
Ok, i did the tests. sadly the easy media node has a substantial different ui from the images shown by the node creator (github), you cannot select a custom size.
Audio reference: Minimax has a very strong feature, load an audio e clone that voice/timber for new lines. this node (easy media) would not use that function.
I'll keep the wf on my watchlist, but as for now, is missing key components.
nope. doesn't work. or it does but your workflow it's not properly explained to understand how to make long form clips using a fraction of the previous video generated.
@Tar_Tarj I have added some notes. 2nd video needs context from the end of first video so it can be latent context or shot context. That option only appears on 2nd or 3rd segment. You have to click on a segment to do that. It will not appear on first segment
@PinkPromptPrincess allright, i'll give it a shot in a few minutes! the workflow is very neat, i have not review yet the wf because I needed you to explain your creative mind a little more 🤣
@Tar_Tarj Regarding audio, just load them as separate layers, and tag them as <Audio 1> <Audio 2> and it should work. At least mine did
@PinkPromptPrincess i did ;_; but my sample doesn't get used, in other wf it does,
@Tar_Tarj okay let me try again to confirm how audio is used.
@PinkPromptPrincess i ain't giving up! I WANT TO USE THIS 🤣
@Tar_Tarj I have attached a video you can see (drop in comfy to load workflow) , click on the segment to see the prompt as well. Both Audio are in different tracks and are working
@Tar_Tarj Added the image as well in the description on how to use Audios
@PinkPromptPrincess i found the culprit, is the audio sample is not in the same "time" position of the segment you're working on, it would revert back to the standard model voices lol.
@Tar_Tarj Oh right,,, face_palm that can be true. I was using 15 to 25 sec long audio and didnt noticed that :D - do post your long video when its done and hey congrates. this is very minimilistc workflow indeed
@PinkPromptPrincess still i don't understand why even on a fresh install of the multitrack node, won't show the mpx controls like your screenshot. it stays stuck with resolution-resize method and format.
@Tar_Tarj i was stuck there too, and you have to select the exact option from that dropdown as in screenshot. or else, just download my video and drop in comfy and it will have all same options as in screenshots. Select width x height (megapixel) there
@PinkPromptPrincess nah, i looked deeper, is a problem with the last comfy frontend and this node, i think i found a fork that workaroud this issue
@Tar_Tarj okay great, I had custom patched the comfy and rgthree so that power lora picker and group bypasswer can wrok in Nodes V2, so i am not updating anything unless my PRs got merge.
2ndly .. if you want to say something else post a new comment. this is getting very long lol
@PinkPromptPrincess MOAR COMMENTS! MOAR!! 🤣
I'll work on the final video review now haha
@Tar_Tarj Since apparently you want more comments. What kind of hardware are you running? I have a pretty beefy rig for a home machine and have yet to find one of these H3 director style workflows that can match the video quality I get from a low/high dual pass. Your video looks pretty solid although hard to say about the long video skin plasticizing issue with your characters ;)