Important Note:
This LoRA was created as part of a fan project and is intended solely for research purposes, not for commercial use. It is based on the copyrighted movie series "Venom." Users bear the responsibility and risk when using this model. Users are obligated to comply with copyright laws and applicable regulations. This model was developed for non-commercial purposes, and I do not intend to infringe upon any copyrights. I take no responsibility for any damage or legal consequences resulting from the use of this model.
This LoRA model is suitable for Hunyuan text-to-video generation.
Recommended LoRA strength: 0.7
Description
Training with 10 video clips, with lengths of 33 frames and 65 frames respectively.
FAQ
Comments (19)
Absolute cinema
cool idea
Finally an interesting concept that can be used for something more than p*rn! Many thanks.
That's pretty cool
Really nice😍 👍🏻
This is so sick! Shame we cant do image to video otherwise video of myself turning into venom would be so cool
You can train a Lora of yourself using just images. I did that and combined it with this lora, came out pretty good.
@roguenova how do you combine two lora? im using the musubi trainer. the inference code only takes one lora as input.
@RedHibiscus I just combine them directly in my ComfyUI workflow. Using the HunyuanVideoWrapper. Someone wrote a pretty decent article on it, including stacking loras:
https://civitai.com/articles/9584/tips-hunyuan-the-bomb-you-are-sleeping-on-rn
Lora training is still to much of a hassle... It's just to much work and to complicated, with stuff like installing a linux system and such. I could think of using the workflow with Deepflow (here on civit). Maybe use it with any image and then combine it with this lora. This maybe could work ... a bit ...
@KarinKara Bro just use gui for misubi tuner. For captions use florence2. Trainig lora simple as possible
@Haunua Yes, this was indeed easy to use :) But only for pictures. Not locally with videos for some motion :/ This is way to much :/
How to become your fans?
Excellent job. May I ask, could you share your training and dataset config? I have been trying the last 2 days to create a similar in nature dusting sequence from the Thanos snap in Infinity War but am trying to find the sweet spot between frame bucket (probably most important?) and resolution. Any pointers I'd love!
Thank you! I'm a beginner too, and this is my first LoRA training. I set up my dataset by following online guides. I’m happy to share it, but it’s experimental and might not be fully optimized. Hope it helps!
@TuTuPie Thanks a bunch! I will return the favor once my Thanos Snap LORA is complete. Cheers!
@TuTuPie Could you share the guidelines?
This is great! Really able to 'venomify' anything. A nice side effect is that it makes any video cinematic style even if you don't prompt for a transformation. Although more often than not it still triggers some sort of transformation effect at random.
Maybe interesting to make a general Cinematic Universe style lora at some point?
2cool
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.