Do you have a good GPU to test this locally? What are your understanding about the licene? Cause, besides the Schnell version, the other two variants are explicit NON-COMMERCIAL use. Just like the first realease of SD3. I wonder if the schanell could run on typical domestic GPU, and if so, if it coild ve trained. Cause if it could, I cant wait to see a “Juggernault” Flux version, or a “Ponny” variant. What do you think?
Yes I've been holding off doing it locally because I use forge as my UI primarily, and they are in the process of implementing updates to allow for the use of flux. But pinokio just released a extremely simple gradio interface that's just a one click install to allow you to run flux locally. It does all the install for you and downloads both the Dev model and the Schnell model. I got that installed today and it actually works really well. I can render four images in about 60 seconds. I have an RTX 3080 TI with 12 gigs of VRAM. My understanding is that people have already started doing training and customizing. There are already a couple variants out. Keep your eye on Civitai
@@AIchemywithXerophayze-jt1gg thank you brother for your answear and insights. The only thing confusing to me is about the license. Cause, if the license will follow a similar path as the SD3 did the community will abandon the model just like they did at SD3.
No the license for the Schnell version allows you to use it for commercial purposes and many other things. The Dev version has a different license though, images generated with that one I don't believe can be used for commercial purposes.
@@AIchemywithXerophayze-jt1gg I have V2 of Flux and tried using a lora from CivitAi, I got an error of not having enough vram. Im using 3060 12gb. Maybe there isnt support but there was an update for support. Not sure what Im doing wrong.
just imagine the finetunes! how could anything else compete
Agreed.
Do you have a good GPU to test this locally?
What are your understanding about the licene? Cause, besides the Schnell version, the other two variants are explicit NON-COMMERCIAL use. Just like the first realease of SD3. I wonder if the schanell could run on typical domestic GPU, and if so, if it coild ve trained. Cause if it could, I cant wait to see a “Juggernault” Flux version, or a “Ponny” variant. What do you think?
Yes I've been holding off doing it locally because I use forge as my UI primarily, and they are in the process of implementing updates to allow for the use of flux. But pinokio just released a extremely simple gradio interface that's just a one click install to allow you to run flux locally. It does all the install for you and downloads both the Dev model and the Schnell model. I got that installed today and it actually works really well. I can render four images in about 60 seconds. I have an RTX 3080 TI with 12 gigs of VRAM. My understanding is that people have already started doing training and customizing. There are already a couple variants out. Keep your eye on Civitai
@@AIchemywithXerophayze-jt1gg thank you brother for your answear and insights. The only thing confusing to me is about the license. Cause, if the license will follow a similar path as the SD3 did the community will abandon the model just like they did at SD3.
No the license for the Schnell version allows you to use it for commercial purposes and many other things. The Dev version has a different license though, images generated with that one I don't believe can be used for commercial purposes.
I have a geforce 6mb rtx 2060. My computer ram is 64gb. Will i be able to run Flux?
Yes I believe so if you use the latest version of forge WebUI
If this thing is finetune friendly, its over.
I believe it is and they are already rolling out some. Keep you eye on civitai
@@AIchemywithXerophayze-jt1gg I have V2 of Flux and tried using a lora from CivitAi, I got an error of not having enough vram. Im using 3060 12gb. Maybe there isnt support but there was an update for support. Not sure what Im doing wrong.