You defintely should make a full ComfyUI-Flux course in English. I would pay for it. Your knowledge is deep. Only problem is, that if someone doesn't know very well ComfyUI, that can't follow a couple of things. But apart from that it is very good, go on! :) And thank you.
Thank you so much for your work and the English translation. I really appreciate it-you dive deep into the details, and your understanding shines through.
After watching your first video moments ago, I wished you would cover and resolve the confusion around the GGUF models as well, and Voila, here we are! Thanks!
Thank you very much.The Flux model is very good,but there is a problem,Flux always generates the same thing :) Even if put up a randomized seed. There can only be a change in generation if you change the promt,CFG or FluxGuidance a bit. In the XL/1.5 models it was necessary to try to achieve a constant result and thanks to IPAdapter/CN. With FLUX, on the contrary, it is necessary to try to make the image generation at least a little bit varied.
I usually take a naked eye look at the requirements file for a new plugin, mainly to see if there are any libraries with a specified version number, and generally there's very little risk of not specifying a version number, otherwise I'd back up the python_embedded folder before installing it, or experiment with --dry-run
You defintely should make a full ComfyUI-Flux course in English. I would pay for it. Your knowledge is deep. Only problem is, that if someone doesn't know very well ComfyUI, that can't follow a couple of things. But apart from that it is very good, go on! :) And thank you.
I’m really happy to hear you found the content helpful, and thank you for the suggestion! 😊 I’ll definitely consider it.
Thank you so much for your work and the English translation. I really appreciate it-you dive deep into the details, and your understanding shines through.
After watching your first video moments ago, I wished you would cover and resolve the confusion around the GGUF models as well, and Voila, here we are! Thanks!
Amazing, thank you this video really deserve it name !
Thank you very much.The Flux model is very good,but there is a problem,Flux always generates the same thing :) Even if put up a randomized seed. There can only be a change in generation if you change the promt,CFG or FluxGuidance a bit.
In the XL/1.5 models it was necessary to try to achieve a constant result and thanks to IPAdapter/CN. With FLUX, on the contrary, it is necessary to try to make the image generation at least a little bit varied.
thanks
Great video!
😊
thanks~
15:00 how to duplicate node like you did? any shortcut?
thx for video, what are ur settings for cuda-system fallback policy?
I usually take a naked eye look at the requirements file for a new plugin, mainly to see if there are any libraries with a specified version number, and generally there's very little risk of not specifying a version number, otherwise I'd back up the python_embedded folder before installing it, or experiment with --dry-run