June 10, Update: stmfnet
Just to save folks a headache the file is available for download as a model now.
June 06, Update: V4.0
Just a minor improvement, using a double interpolation of SFTMVFI results in higher fidelity of dynamic or fast motion scenes. Video is manually loaded to reduce time when generating high quantities of test runs. Now you can select which videos you want to upscale, will copy the name of the original input video.
Included the context framing node from Kijai's wrapper. Usually leave this off will exponentially increase generation times if on; however, can let you generate much longer videos beyond what your vram will allow you.
Interpolation requires higher VRAM if scaling if greater than 1.0. Changed the scaling factor to float for more control.
Updated nodes if there is any issues feel free to let me know.
https://drive.google.com/file/d/1s5JJdt5X69AO2E2uuaes17aPwlWIQagG/view
Place it in the folder: Custom_nodes\ComfyUI-Frame-Interpolation\ckpts\stmfnet
April 14, Update: V3.3
Updated Nodes.
March 27, Update: V3.2
Small changes to the Torch Compile and Tea Cache for more dynamic outputs using my Lora's.
Workflow is suitable for a 24GB card using the 14B models and a 12GB card using the 1.3B model.
Updates:
***Error within the naming of outputs corrected.
Uploaded a very simple and straightforward Wan I2V workflow. The workflow includes an upscaler with frame interpolation to bring your videos to a cool and smooth 60FPS. No BS just a simple addition. Original WanI2V is based on the workflow example of the amazing work from Kijai linked here: https://github.com/kijai/ComfyUI-WanVideoWrapper/tree/main/example_workflows.
Additionally, I've left the annotations within.
Prev Updates:
Compatible with 12GB memory cards. Tested using a 3060 RTX 12GB using this workflow can comfortably go up to 85 frames. 97 Frames is pushing the absolute limits will take about an hour to generate.
Due to demand this workflow is a small improvement to the Super Simple HV2V for those who don't do postprocessing outside of Comfyui.
Designed for use with Hunyuan Video and trained Lora models, it can upscale videos to true 60 FPS. It combines detailed notes on denoising strength, step optimization, aspect ratio adjustments, frame interpolation, and upscaling to create high-quality, smooth videos.
Order matters:
Generate → Upscale → Interpolate. Upscaling first preserves detail for interpolation.
Adjust parameters based on your Lora’s behavior and hardware limits.
Description
This model is an update to Super Simple Hv2v
FAQ
Comments (34)
"for those who don't do postprocessing outside of Comfyui." Got any tips for doing the postprocessing outside of ComfyUI? I like having more options. 😉
Filmora is a cost-effective approach with tons of options. I don't know about now, but I was able to buy a lifetime license of one of their versions. Suitable for making clips for training Lora's, GPU upscaling and interpolating to increase frames, and much more. Although I am pleasantly surprised by this completely free option, the majority of my generations are set and forget and can be used for posting as is with this workflow.
Topaz? Davinci Resolve?
@olivetty never tried either of those
CapCut is a competent video editor with lots of features. I've been using the free version for a long time
Which Video Adapter do you use, where can I get the right one?
hyv casd adapter_model.safetensors
That's the Beautiful Blondes Lora I just Posted
im having a problem with feta_args it shows me red circle
try updating the Hunyuan video wrapper
@SingularUnity i did in custom nodes manager but still the same. I'm not really good at comfy
@pazuzu666 I'm sorry, in not sure. Have you been able to trouble shoot it?
@SingularUnity still confuse, how do you trouble shoot? i'm very new at comfy i mainly use forge more :)
I had the same issue. It turns out Sampler node in the workflow mismatched with latest version of ComfyUI in my case. If I use fix node and reconnect the feta_args connection manually, the issue will be resolved.
@pazuzu666 Sorry I'm not sure how I can assist without fully understanding the issue. It seems @ken522752 has an understanding of what you're experiencing. Hopefully that helps.
Hi, I am quite new to using Tencent Hunyuan. It would be much appreciated if you could answer my question.
1. Does the video in the VideoHelperSuite node matter?
If so, could you kindly provide the resource file “My Video-1.mp4”? I tried using another random video file. Somehow my ComfyUI crash while “sampling 49 frames in 13 latents at 512x720 with 8 inference steps”. Do you have any suggestions on how to resolve this issue? For reference, I’m currently using ComfyUI_windows_portable.
2. What are the GPU requirements for running this workflow?
How demanding is this workflow in terms of graphics card specifications? Could you also share any recommended settings for a lower-spec GPU? I’m using an RTX 4060 Ti with 16GB VRAM and would like to know if it can handle the workflow decently.
Hello, and I appreciate questions and I'll try to help out as best I can.
1. The video provides the general context and is a pretty big influencer in the video generate. Lora's like mine are meant to assist in generating a consistent character or style. In general, you'll want to pick a video that has the scene you want.
2. People have reported that they've been generating content with 16GB of VRAM Nvidia cards using my model. Occasionally someone has a 24GB card, such as the 4090 or 3090.
The demands are most heavily dependent on the generation resolution, number of frames, and inference steps.
Inference steps are important and a good number for quality content is >20 steps. The notes in my workflow give a good starting explanation; however, I'd encourage you to do your own research.
In terms of sharing prompt videos. DM me and I don't mind sharing.
If you have less than 16 GB of VRAM, you might have a hard time creating content using HunyuanVideo, though it's not impossible; just don't expect high resolutions.
Why didnt you leave the meta in the ones you produced. I have seen too many Hunyuan loras on here that NEVER place the meta data in the vids? WHY? I tried everything and nothing worked. i even place one of you generations into the vid part and it never does anything.... tried MANY other short vids and nothing.
this is the error i always get:
File "Q:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-videohelpersuite\videohelpersuite\load_video_nodes.py", line 370, in load_video
raise RuntimeError("No frames generated")
RuntimeError: No frames generated
I have the most recent comfy, a 4090 and 128gb ddr5. So why is this happening.
...figured it out... the skip first frames was at 200 in your json.... might want to change/update that. Still i wish you left the meta in your great renders!
nevermind now its giving me this error./////
Failed to find C compiler. Please specify via CC environment variable.
@StarW were you able to get it running?
@SingularUnity no unfortunately. i havge everythin ginstalled includinmg triton and it still wont work. cant find MSVC apperently and trying to find solutions. this is really stupid.
@StarW I'm sorry that you're having such a hard time. this workflow was meant to be very easy and in fact I've gotten numerous comments on how simple it is and many people have been able to just deploy it without issue. so I'm sorry that you're happy such difficulties.
@SingularUnity OK!!! i figured it out. You have to add the msvc to PATH, for the updated VS, C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.43.34808\bin\Hostx64\x6
@StarW this means you did not have Triton properly setup.
I think I understand now, you might also need to copy the Triton libs and site packages to your comfyui python folders. but give it a try and see if it works
now when runni git, the outcome is red, very satruated picures. This is jsut running one of your already generated vids for testing. It jsut comes out really bad.... im using your settings, and your vid> what am i doing wrong here?
@SingularUnity it was setup right and triton was already in my comfy folder. It never placed the VS info in the path.
i am not the only person that is having issues installing this on windows. I followed this setup: https://www.reddit.com/r/StableDiffusion/comments/1h7hunp/how_to_run_hunyuanvideo_on_a_single_24gb_vram_card/?rdt=33021 and still there were issues. I have it up and running now, and just with your base settings, and not touching anything, and placing one of your generations into the input, they come out really saturated, blurred, and burnt in. i changed the denoise down to .5 and it did generate but iy came out fuzzy, blurred.
i find it a little bit tedious as i would have to have a specific size and frames for this to work??? Why cant i just use any vid? i would have to spend a lot of time editing vids, and then resizing them, and then make sure to enter in the EXACT settings just to reproduce them? just trying to figure out how to really use this as the result look amazing, but still this is tedious.
@StarW I should've been more explicit, the images and videos I train on are formatted for social media so it's a portrait frame with aspect ratio 3:4 and 4:5. Technically any frame size that meets that aspect ratio will work. I found that for my use and card 3090 with 24GB. I'm able to generate 201 frames long videos with upscale and interpolation at a resolution of 640x896 without getting an OOM error. I'm assuming the 4090 would be caped at this too.
For shorter videos experiment with higher resolutions but remember you won't be able to do 201 frames.
@SingularUnity thanks for the info. i will try it out some and get back to you on it. thanks for all your hard worl!
@StarW Not a problem, let me know I'll try to help out as best I can. If you haven't already drop me a follow for future updates, new Loras, and posts. I'll also be doing up an article on training and generations in the future. Cheers.
Hi, I'm new to AI generating. Would my RTX 3080 with 16gb VRAM be capable of running something like this optimally?
It would definitely be able to run up to 129 frames using the dimensions and steps specified. About a 4 second clip at 30fps.
same as mine and very capable