IMO I agree! 😁

As I said on your other comment, these nodes used in the video are regularly maintained and up to date. Please double check that you are searching for the correct ones.

Are you sure about that? I was just using them yesterday. Make sure you search for the correct node package, you want "RGThree"

This training is only based on SD1.5 and is quite dated. You can't even train SDXL models. So you are limited to these specifications. Best thing to do is create the image in a particular size (640x832 as an example) and upscale the image. You can also outpaint to get a bit more but bigger sizes like 16:9 may not turn out well since native resolution is 512x512.

@ okay thanks mate

This is an older video my friend, if you look at the pinned comment you will see that this version is no longer supported. It should still work fine however, there will be no updates or improvements made. Best to use the original Fooocus or another fork of fooocus. Even the original Fooocus is no longer going to be developed further, just maintained.

@@MonzonMedia Thanks for your advice. I will stick with the vanilla version then. It's a pitty however that the original Fooocus is no longer going to be developed further, just maintained. It's a beautiful piece of software, Powerfull but yet simple and beginer friendly. Nothing against Forge, or Comfy UI but the learning curve is high and for many content developers, Fooocus is a breeze. Thanks for your reply friend.

😆☺️

Appreciate that, glad you found it helpful! Just note this workflow was when Flux was fairly new, it's much more easier to create a Flux workflow now. I have a series of videos (6 so far) that go over creating and SDXL workflow then converting them to Flux rather simply. Just check my Comfyui tutorials playlist.

🙏Thank you!

Thanks, I had a laugh making it too! 😂

@ 👍👍

@@ennanoff honestly I’d say this model feels like minimax hailuo. Still playing with it. Easy to prompt though.

😁Thank you! Was so fun to make! 😊

❤ all the elements I love😂

And it's fun to play with!

Hahaha! If only bud. Would be fun though! 😬

Pika 2.1 as of now can do 5 sec clips. Eventually I'm sure they'll make it where you can extend it further. It takes a lot of GPU power to generate these at a reasonable time. The open source models can take several minutes depending on the GPU.

Thanks bud! 👋🏽👊🏼

More like a demo. I’ll publish a video soon and go over some things. 👍🏼

@ ok

He’s evolved over time. 😊 still with me 😬

Not there yet but perhaps sooner than later.

🙏🏽🙌🏼❤️

Still playing around but definitely an improvement from the last 1.5 model. Kling is still king at the moment but great to have competition

The way last year ended I knew 2025 would be the year of ai video. It's definitely going to get interesting this year!

Yes it is : /

Super fun to try it out! 😁

Thanks! I'm glad you found it useful. 🙌🙏

Love to hear that! Make sure to check out the rest of the vids on the series. Working on more to publish soon!

They have the main video platforms available now, Kling, Hailuo, Luma and Runway. But there is an allotted amount per tier, check their pricing page. I believe all models are available for any tier. Some of them are based on SDXL, not sure if their custom models are based off SDXL or Flux. I'll see if I can find out. They do have Flux dev available for image generation.

@@MonzonMedia Thanks. I've been watching your videos for awhile now from a different account. Thanks for doing these for us!

I'm honored and thank you! 🙏 I will be posting more about Dzine and I'm working on something I just got news about from another video platform! Might publish tomorrow.

@@MonzonMedia any hints on what it is? I use alot of kling, but it gets so expensive. I got into the creator program for Nim, but it doesnt look anywhere as good as Dzine yet. I think I might jump in and play around with it because of the uncensored videos someone kept mentioning.

Pika 2.1 ssshhhhh hahaha! Still playing with it. Regarding uncensored stuff, he was talking about images. Videos might be a different story.

you only need to input the token if invoke ai prompts you to. typically there would be another field here shown in the video....but in this example it wasn't required. kzbin.info/www/bejne/joGlapqCed6dndksi=QnpoNoixMoS9PWij&t=191 That being said, it's probably easier to just download the model locally and link to the folder manually in the model manager. Much easier. I find the hugging face token thing can be unstable and sometimes fail.

Thank you for watching! Glad you found it helpful. 👍

To be honest I've yet to try as I've been busy with other content but I've been meaning to try just this. Will definitely post something about it soon. Thanks for the reminder. Perhaps you can share what you want to do?

Have you tired this method? kzbin.info/www/bejne/nJTZdoB4dpKZpqMsi=_XAWRq0WtFDhDXTk

Thank you! Glad you found it useful. 👍