Trained using RemadeAI's Wan LoRA trainer - give it a 'shot' ;) !
Join us: https://discord.gg/7tsKMCbNFC
SHOOTIT is a LoRA for shootin' firearms at things.
Trained off of high-quality clips from Grand Theft Auto V in 'first-person mode', it makes some pretty cool effects!
Due to Civitai ToS, I have chosen to pick out just some safer/simple sample prompts using T2V but it also works with I2V. You definitely should experiment with prompts on your own. Remember, Food Mage is a family-friendly page. ;)
Unload your whole mag into subjects, with ragdoll effects included.
Be the hero! SHOOT IT!
Trigger: Sh00tit shoot
EXAMPLE PROMPT CONCEPT
The viewer is holding a machine pistol, the viewer aims the gun at (subject) and shoots the pistol. The muzzle flashes and the (subject) falls over, the pistol continues to sh00tit shoot the (subject) while it lays on the ground
Try shootin' at monsters, vehicles, objects, and more!
Works with Image2Video, so aim at what you please! Be weary of Civitai ToS before posting.
Features:
Falling on the ground/ragdoll
Rifles, Shotguns, Handguns trained
Only trained on video game footage
Specify continuous shooting, single shots, and where the shot is fired
Includes 'muzzle flash' for weapons firing
Colorful splats when specified
Have fun!
~Mage !
Description
FAQ
Comments (27)
works well. if you use a lower strength and i2v you can get other views too.
Hmm interesting. Care to share?
good stuff
Thank ya AndroidXL! Let's do more!
Remember... no Russian...
I laughed when I read this
When I shoot at a human with i2v, a blood-stained effect occurs. How can I get rid of this? This applies even if it's not a weapon.
I don't know, have you tried blood in negative or something similar?
@thefoodmage
I didn't use any prompts related to blood anywhere, and putting it in negative didn't change anything. I thought it was something that would always happen, but it doesn't seem to be, so I'll try it some more.Thank you！
imagine the videos trained used is from call of duty or battlefield ! dang! that would've been great! anyway its fun playing around this lora :D
Send me clips and I can make a V3..!
@thefoodmage i dont record clips sryyy! but youtube had it all , not demanding you to do it, maybe when you feel like it ill be following this post just in case haha ! additional idea (body cam fps game) might work also XD
@minanimator28 how dare you not demand of me :o lol
@thefoodmage i forgot to ask, does the arsenal include rpg? so i can shoot and explode stuff
@minanimator28 Sadly no but if someone gives me footage I can easily train it :)
You can TRY though
could we get a 2.2 version of this please
Yes tomorrow I will download Wan2.1 architecture and see about re-training it :)
I meant Wan 2.2.
When I recharge my cloud GPU account I can train it again.
@thefoodmage Hello Foodmage,
Are you still planning to train this Lora for WAN 2.2? Or have you decided against it?
In any case, thank you very much for your work.
@crhpvsmb II could do it but I am lazy. Do you want the training data to train it?
I can't train video locally but I can train images.
@thefoodmage Thank you. I am in the 12GB VRAM camp. Training characters with images works, videos unfortunately not. No need to use the cloud and waste money unnecessarily.
@crhpvsmb Hey sorry I took so long to get back to you. I'd be happy to train it but I'd need like $10 in credits just to train this 1 model! I know how to do it though.
A LTX-2 Version would be so awesome!
At first I ignored this thinking it was a weird comment asking me to go back to very old models but now I see. I will have to download LTX-2 some time and also train on it. I think it looks decently impressive, at least for memes!
If you can train it yourself, I can just give you the training data too.
@thefoodmage I would happily do it, but I lack the hardware, at the moment... I can barely run the LTX2 model, let alone train it :( But if you could share the training data for the original lora, it would be awesome. Who knows, maybe I'll find a way to attempt training models at some point.
@paleopterix You can train them on cloud GPUs for 1-2 USD per hour. This one might take 6 hour to train since it's a video dataset. Do you think you may be able to put in around $10 for training? If so, I can give you the dataset. Would be nice if you shared in the end but it's your money ;)
LTX 2.3 can do it natively
Details
Files
Available On (3 platforms)
Same model published on other platforms. May have additional downloads or version variants.