Thanks again for such an amazing video on faster Flux Lora training! Would you be able to create a tutorial on how to combine different trained Lora's, for example a product and then combine it with the human model? I loved your example of product placement in this video and would love to get some more insight on how to utilize multiple trained Loras.
Hey! I see you’re always looking for feedback on your community, which is awesome. Honestly, when it comes to AI, I’ve got to say that I’m more of a budget-conscious guy, so I usually lean towards open-source options. They just make more sense for someone like me who’s trying to get the most out of what’s available without breaking the bank. Keep up the great work, though! I love seeing how passionate you are about making this space better for everyone ♥.
Yes same here as well. That's honestly why this also made sense because you spend $2 and fine-tune your model and then re use the model with a local solution. Also this is more for people who are building SaaS where they can almost fine-tune on the fly. Hopefully we'll have something faster open solution
Guess what, when they first launch the LoRA trainer it was 5$ explicitly mentioned but then I tweet compare it with replicate 2$ training cost with 1000 step and they reduce the price within 5 hours😂😂
I would be incredibly interested in a video about training a single model (or multiple models if necessary) on a range of products. Ideally having a local system (that uses an api or not), where I can have multiple folders, each representing a product and having 10 images each, and training a flux model on all of them, so that I can call the model to generate images from either of them or them together at any point. This would be groundbreaking. But until then, great job, really like your videos!
Oh, i had $1 when I registered using your link in the description and had $1. I just clicked the link again and now I have $8.50 I guess there was some error applying the coupon initially when i registered.
Any idea why for Replicate we uses minimum 1000 step but here 100 step? After training multiple LoRA how do I access it from single prompt coz in replicate there are option in prompt?
@@1littlecoder what does it means? How do i use multiple LoRA to generate just 1 image? Sorry i am new to the world of ai 😅 but I've watched hundreds of videos, and this one of you has been definitely the most effective method yet!!! I really have to thank you very much
so you dont need to add a text prompt along with the picture? it seems that would guide it better. is there a way to do that with this new provider you found?
Is it working for anybody using redeemed credits?? Whenever I click start button to train lora it loads for several seconds but nothing comes in training request or history section.
Raja I am your loyal subscriber I won't miss even a single video of yours. But I think my bad luck and I didn't got the free credits because I am bit late. But hope it for next time. And thank you for Awesome content.
In Fal its 100% exact same character when using the Lora, but in Comfyui when using it localy on ur computer is only 50% resmble. why is that happening. i downloaded the Lora and put it inside Lora folder in comfyui, what esle do i need to put in folder ? what abt the Jason file, where do i need to put that in whic folder ?
Hi, if I may ask you a question, is Flux.1 Dev fined tunes lora can br used for commercial use or not? I mean in the official license said it can't be used for commercial use but some APIs providers such as Replicate and Fal said it can, it's confusing
I'm very sorry. But I guess generally if you sign up for them you might get some initial credit. I'm not exactly sure I signed up long back so currently I'm not sure what is the state
@@1littlecoder No worries, not your fault. Great vid, I still may try it. I would like to see you use the DHANUSH lora in comfy locally so we can compare both results. That's the true test...
@@1littlecoder will be good a video on how to fine tune an existing flux Lora, to add some new details and textures, instead or making a new lora. thanks
@@1littlecoder i downloaded the lora and using it localy on comfyui and it isnt any close to the model i trained it looks so differnt . do u know the setting of lora . my cfg setting is 1.0 , Flux Guidance is 2.5 , Lora Strength model is 5.0 , strength clip is 2.0 . its resmbles like only 60%
If you have seen anything faster than this, let me know!
Probably not faster, but could be cheaper if you skip the middleman.
Thanks again for such an amazing video on faster Flux Lora training! Would you be able to create a tutorial on how to combine different trained Lora's, for example a product and then combine it with the human model? I loved your example of product placement in this video and would love to get some more insight on how to utilize multiple trained Loras.
Yeah Exactly my requirement too... in fal there is only one lora compatability.
Hey! I see you’re always looking for feedback on your community, which is awesome. Honestly, when it comes to AI, I’ve got to say that I’m more of a budget-conscious guy, so I usually lean towards open-source options. They just make more sense for someone like me who’s trying to get the most out of what’s available without breaking the bank. Keep up the great work, though! I love seeing how passionate you are about making this space better for everyone ♥.
Yes same here as well. That's honestly why this also made sense because you spend $2 and fine-tune your model and then re use the model with a local solution. Also this is more for people who are building SaaS where they can almost fine-tune on the fly. Hopefully we'll have something faster open solution
youre a life save buddy, keep up the good work ♥
Guess what, when they first launch the LoRA trainer it was 5$ explicitly mentioned but then I tweet compare it with replicate 2$ training cost with 1000 step and they reduce the price within 5 hours😂😂
Thank you -- great channel, learning a lot!
Great to hear!
I would be incredibly interested in a video about training a single model (or multiple models if necessary) on a range of products.
Ideally having a local system (that uses an api or not), where I can have multiple folders, each representing a product and having 10 images each, and training a flux model on all of them, so that I can call the model to generate images from either of them or them together at any point.
This would be groundbreaking. But until then, great job, really like your videos!
Thank you, i enjoyed this today
Glad you liked it
Wow, thanks for the extra free credits! I guess I'm one of the first hundred people who have signed up using the link.
Oh great. How much credit did you get ?
I signed up 3 hours back and got just $1
Oh, i had $1 when I registered using your link in the description and had $1. I just clicked the link again and now I have $8.50
I guess there was some error applying the coupon initially when i registered.
@@1littlecoder i got 7 , lucky me
7 is good. Great to know!
hey sir plz make workflow for comfyui for using this lora localy
thanks bro, i will try multi cooking loras for product placement with a model :)
Awesome. That can very well turn into a decent SaaS
Any idea why for Replicate we uses minimum 1000 step but here 100 step? After training multiple LoRA how do I access it from single prompt coz in replicate there are option in prompt?
wow! super easy method!!! how to combine multiple LoRAs??
Replicate has a multi lora explorer!
@@1littlecoder what does it means? How do i use multiple LoRA to generate just 1 image? Sorry i am new to the world of ai 😅 but I've watched hundreds of videos, and this one of you has been definitely the most effective method yet!!! I really have to thank you very much
so you dont need to add a text prompt along with the picture? it seems that would guide it better. is there a way to do that with this new provider you found?
model accuracy is not good as compare to replicate one. how to overcome this problem ???
Thanks for this! Is there a way to use multiple LoRAs on FAL? Or is this something you have to set up in comfyUI?
here - fal.ai/models/fal-ai/flux-lora/playground
Is it working for anybody using redeemed credits?? Whenever I click start button to train lora it loads for several seconds but nothing comes in training request or history section.
Raja I am your loyal subscriber I won't miss even a single video of yours. But I think my bad luck and I didn't got the free credits because I am bit late. But hope it for next time. And thank you for Awesome content.
Can you please email me once? I will check with them if there is anything could be done or I can train the Lora for you
Same here I couldn't get the free credit
Can i use this on Google colab?
And is it free?
Can i clone my self for making my ai pictures?
can I use this on mobile browser
Thanks. A doub, the images can be of different resolutions? No .txt file description?
their processing actually segments the image and file description is not mandatory.
also check out the advanced tab :)
@@1littlecoder ok thanks bro. Already registered!
@@surflaweb good luck!
Bro, how to run this in local that is in a runpod instance ?
Im really confused on the next step after this. I have no idea the best method to run this new lora. Id like it to be private since its my real face
Download the lora files and then follow this to run locally ruclips.net/video/3uuxp0v3FSQ/видео.html
@@1littlecoder That was fast, thank you for the help!
can u guide me with product placement lora? u trained two loras and combined them? what did u use? and how can i try it?
Yes .. pls show the steps to use 2 loras combined
In Fal its 100% exact same character when using the Lora, but in Comfyui when using it localy on ur computer is only 50% resmble. why is that happening. i downloaded the Lora and put it inside Lora folder in comfyui, what esle do i need to put in folder ? what abt the Jason file, where do i need to put that in whic folder ?
@@user-nb6kx3qn3p I guess you might have to increase the lora strength and try on comfy UI
Hi, if I may ask you a question, is Flux.1 Dev fined tunes lora can br used for commercial use or not? I mean in the official license said it can't be used for commercial use but some APIs providers such as Replicate and Fal said it can, it's confusing
I have the same confusion. I don't have an answer. Not sure if they made any deal with the Flux company
@@1littlecoder Thank you for your response. I already subscribed and I'll wait the video about this. This is important.
Hey great video ! Btw are there free alternatives with similar performance
@@musiclovrrap862 from what I know, there aren't even paid ones with 5 mins fine tuning
after training can i generate unlimited images with no cost?
You can do it locally for free ruclips.net/video/3uuxp0v3FSQ/видео.html
Looks like you got your first 100 already because I didn't get anything, but thank you for sharing...
I'm very sorry. But I guess generally if you sign up for them you might get some initial credit. I'm not exactly sure I signed up long back so currently I'm not sure what is the state
@@1littlecoder No worries, not your fault. Great vid, I still may try it. I would like to see you use the DHANUSH lora in comfy locally so we can compare both results. That's the true test...
can you tell me how can i finetune lora with multiple characters. is it even possible
you have to fine tune them separately as separate loras
@@1littlecoder if so how can i create image with multiple custom characters (lora) in one image?
yes you can
I couldn't get free credit can you do something for me
But...this is not a finetuning, this is a simple lora...no? explain me if im wrong please. thanks
LoRA is a fine tuning method but you're right this is not full fine-tuning.
@@1littlecoder will be good a video on how to fine tune an existing flux Lora, to add some new details and textures, instead or making a new lora. thanks
@giuseppedaizzole7025 you can use multi lora without training a different one (if someone has already trained)
@@1littlecoder I know, but i'm talking about fine tune of an existing lora. thanks
Can u start do videos with white background....not smoth to look all black...otherways nice and gj...like it
@@robertaopd2182 what do you mean by white background? Could you please elaborate
How to make it look more realistic. Can we train flux realism with our face
You can apply flux realism as a lower adapter on top of what you have already trained on a multi-loader
@@1littlecoder is there any API that allows to do this. Can I refer to any RUclips video
How can I train same modal with multiple person?
so far what i know is that you need to train multiple lora. I've not seen anything where a single lora representing more than one person
So can i do that using this platform? Because once I trained a modal with one person then how to train the same model with second one.
@@ankitkhare4831 you train two different Lora's and there are some multi Lora explorers where you can import both the loras
The guy has 6 finger ?
Dhanush with 6 fingers hahahaa
I want something more open source $0
Please use your face as input
😂😂 lighting fix karo
i trained now and it deosnt work. Rubbish
What doesn't work
@@1littlecoder i downloaded the lora and using it localy on comfyui and it isnt any close to the model i trained it looks so differnt . do u know the setting of lora . my cfg setting is 1.0 , Flux Guidance is 2.5 , Lora Strength model is 5.0 , strength clip is 2.0 . its resmbles like only 60%
@@user-nb6kx3qn3p reduce lora strength to 0.7 and try