ComfyUI Tutorial Series Ep 24: Unlock Flux Redux & Inpainting with LoRA

[@pixaroma]

Free workflows are available on the Pixaroma Discord server in the pixaroma-workflows channel discord.gg/gggpkVgBf3 You can now support the channel and unlock exclusive perks by becoming a member: pixaroma kzbin.info/door/mMbwA-s3GZDKVzGZ-kPwaQjoin Check my other channels: www.youtube.com/@altflux www.youtube.com/@AI2Play

[@LouisGedo]

👋 hi

[@pixaroma]

@@LouisGedo 👋 hi

[@Uday_अK]

Mind blown! 🤯 Inpainting with LoRA in ComfyUI is next-level stuff!

[@pixaroma]

thanks 🙂

[@ThePurpleSnork]

I’ve been working my way through your tutorials, and they’ve taken me from barely understanding any of this, to actually getting reproducible and consistent results. Thank you so much!

[@pixaroma]

Glad I could help ☺️

[@Redemptionz2]

This is the best teaching channel I've ever watched.

[@pixaroma]

Thanks ☺️

[@frankiberlin]

Thank you very much, again very well and detailed explanation. Greetings 😊

[@SebAnt]

Thanks! 🙏As usual another superb tutorial !

[@pixaroma]

Thank you so much Sebant ☺️b

[@reynaldmonnier5498]

Super, it's great to have added audio tracks in different languages. I'll be able to listen to your valuable advice in French and concentrate more easily on what's happening on the screen.

[@pixaroma]

yeah youtube started to do that automatically for new uploaded videos, the old video dont have that only whats new uploaded from now on.

[@AndreyJulpa]

Nice tutorial, especially with LORA, thank you!

[@pixaroma]

thanks 🙂

[@SumoBundle]

Great tutorial as usual

[@ivo_tm]

Thanks for the awesome tutorial !

[@pixaroma]

Thank ivo 🙂

[@comal8249]

U r the best teacher

[@joaocrfonseca]

Thank you for the PT-BR dubbing!! What about the previous videos? May I expect the same to happen???

[@pixaroma]

KZbin does that automatically only for new videos I upload, is not something I activate is just happens for the new uploads, so is not something i can enable in old videos, is something that happens only on new videos I upload from now on.

[@studio_20x30]

my wishlist is a whole video focusing on archviz and improving renders with inpainting in the best optimised way possible with flux or sdxl.

[@baheth3elmy16]

Thank you!

[@dimamakrinic2546]

Hey, great video. Thank you so much! Where would you recommend training the LoRa? Civit, TensorArt, or Replicate?

[@pixaroma]

I didn't try all, so not sure what is best, see what is the cheapest i guess, i used tensorart but not sure is the cheapest but that worked ok for what i need

[@MallikarjunaKasireddy]

Hello I am from 🇮🇳. I'm sticking to ComfyUI for advanced work because, even though it may not be the latest compared to some subscription-based tools, I believe all subscription models will eventually come to ComfyUI. Its level of customization is unmatched, and it’s perfect for my needs. My goal is to create an epic story, which many have attempted using AI, but they often struggle with frame consistency. With ComfyUI, I can ensure that consistency. Currently, I’m working with a 16 GB CPU and NVIDIA GTX 1650, which significantly slows down processing advanced tasks. This has been a challenge, particularly for consistent frame generation in my project. I’m eagerly waiting for the performance of the upcoming NVIDIA Quadro 5090, but I’m also considering the NVIDIA RTX 4090 as an alternative. High-end graphics cards are undeniably expensive, but after watching your workflows, I’ve realized why they are essential for achieving faster and more efficient results. In the meantime, I’ll focus on mastering the basics and refining my workflow. Thank you for your tutorials-they resonate deeply with me as you patiently cover each property in detail, which perfectly matches my approach to learning.

[@pixaroma]

thanks, I do recommend rtx4090 over quadro, I have rtx4090 and works great.

[@MallikarjunaKasireddy]

@@pixaroma Thankyou

[@MallikarjunaKasireddy]

@@pixaroma Thank you for your response! I really appreciate your recommendation of the RTX 4090 over the Quadro. It's reassuring to hear that it works great for your workflow. I have one more question: When considering a long-term investment, would you recommend a high-performance laptop or a desktop for AI-related tasks? My primary focus is on projects like advanced frame generation and consistent outputs using tools like ComfyUI. Since such purchases are significant investments, I want to ensure that my choice remains relevant and doesn’t lead to disappointment in the future. What’s your take on this, considering factors like upgradeability, portability, and performance?

[@pixaroma]

@@MallikarjunaKasireddy Well you pay a lot more for portability, and you can not put in a think laptop same power like you put on a desktop. So desktop will be more powerful than a laptop, look at the brick size of RTX4090 is huge, to make it think and small you need some compromise, so in my opinion if you dont travel a lot desktop will give you more for the same money. I only use desktop, I can not really work on small screens, I do designs also so I need big screens to see details, and laptop screens are small and not so great :) As for AI I am sure we can do many things with comfyui in the future, but that doesnt means that someone can invent something new that is better than comfyui, things evolve, so we will see.

[@MallikarjunaKasireddy]

@@pixaroma Thank you for your detailed response-it’s very helpful! I completely agree with you that desktops are more practical for performance-heavy tasks, especially with powerful components like the RTX 4090. Currently, I’m working as a professional engineer, and as a dreamer, I’m passionate about designing and creating. I’ve often thought about starting projects but waited for the right moment. Now, I feel it’s time to take the risk and invest in a high-end computer to make things work. Your tutorials have been incredibly inspiring and have given me the confidence to take action. I also appreciate your point about AI tools evolving. While I’m focused on mastering ComfyUI, I’ll keep an eye on future developments to stay updated. Your insight has helped me gain more clarity about my next investment. Thanks again!

[@arnabing]

This is awesome! Thank you. Is there a good way to turn this into an api call that is relatively quick for an app?

[@pixaroma]

Sorry I don't have experience with api, so not sure how can be done

[@Ozstudiosio]

welcome back with new video :), but do you think its will be better than Swapface with Cartoon style cuz i try swapface with 3d cartoon style and result have a lot of issues?

[@pixaroma]

I didn't try swapface, i usually dont stick to one method i try a lot of them and use the one that works for what i need :)

[@Ozstudiosio]

@@pixaroma yeah exactly that's what i did in my pipeline project, sometimes its hard to discover different CG workflow but from other side you have Varity of cards play with them based on project type :)

[@KimikimiYKim]

Next video make image to video on sdxl and flux on comfyui please

[@pixaroma]

I will see what i can do, but usually those need bigger models so you need good video card

[@KimikimiYKim]

@pixaroma are there any scanel flux models that are less than 23 gb in size?

[@manjukeshm4636]

Can you make a video about fluxgym i encounter continous errors using pinokio

[@pixaroma]

I got errors too, that why I used online, plus I can work in that time and is only a few dollars.

[@frankiberlin]

Hi, it's working at my side. The Mods at the Pinokio Discord are very helpful. But it needed a whole night in a local installation, at my tries, and the settings for possible good results still need to be optimized by me, very much. Cheers :)

[@pixaroma]

I am sure I could make it work on my PC eventually but the problem is if i make a tutorial how i explained other people how to fix it, since each pc is different, with different software, so I will be all day trying to fix errors, so too much problems, i will wait for easier to use options

[@amigo516]

Sir, I am deaf. Other videos have subtitles, but this one doesn't. Could you please help me?

[@pixaroma]

other videos has the option to enable CC, this one doesnt have, not sure why, might be a bug from youtube since they added new audio languages and probably messed up something with CC or it will take longer to appear, not sure.

[@studio_20x30]

almost first one here!

[@Vanced2Dua]

Please A1111

[@pixaroma]

sorry I have it used in months, all people i know switched to comfyui, so unless they do something amazing and get better than comfyui I dont think I will do tutorials for it. If it gets better then yes why not, but so far is slower and didnt offer anything extra and updates are very slow, like with comfyui i can use flux from first day, and it took weeks for a1111 to do something