Hi ICEKIUB! I'm a huge fan of your work and have learned a ton from your tutorials. I'm currently experimenting with the Wan 2.2 animate workflows and I'm really trying to push the quality on facial micro-expression transfers. What’s the best approach or node setup for achieving the highest quality results there?

Anyone here managed to keep a consistent background in ComfyUI while changing poses/outfits? What workflow are you using?

Couple of updates following the poll I made this week (thanks for your insights!) I tried X; It's kinda abysmal for my type of content, I will try to use it but won't be my main thing, you can follow your boy over here https://x.com/icekiubAI Regarding the videos, my most popular videos and workflows have mostly been re-uploaded on the skool community, I omitted some of my older less relevant videos but I can reupload whatever you guys might need. My twin brother will also be taking recommendations for future workflows/videos. For the people coming from discord after my post: if you were a prior sub dm me proof on skool to get a couple free days on the sub tier as a thanks for the support OR resub like a chad, roles and tiers will be reset to real active subscriptions around may 6th. Discord server link for the skoolers: https://discord.gg/6yTkbzGz Appreciate the support either way 🙏 Because I am banned on youtube, I am technically banned for life from creating a new channel. Luckily, my twin brother FireTriangle has the exact same knowledge and content creation motivation as me and decided to start his own youtube channel... How crazy is that! The odds are his channel will also get banned eventually for ban evasion, but hopefully it can last a while. The videos will be double posted in skool too just in case You can follow my twin brother's channel here, he said he'll post his first vid this week https://www.youtube.com/@FireTriangle-v1 my selfhosted platform will not be going live right now but is ready to go in case we get banned on skool for whatever reason (Don't wanna fragment the community too much)

As you may have seen my youtube channel with over 45 videos of educational ComfyUI / AI tech content has been terminated by youtube due to censorship rules regarding ai influencer content. This is the reason why I have been more unresponsive in the last couple days, trying to figure stuff out in the shadows. We have now been banned on 4 platforms without even showing one nipple 🤣 For those who have been asking, yes I will be re-uploading what I still have on some alternative platform but my question for you is which platform should I use/ would you watch on, in the future? Ultimately, my main goal is to be remain discoverable organically which was mainly happening through youtube before the great purge... Let me know your opinion 👇

Hi everyone, I'm having trouble creating the dataset. Could you briefly explain how you do it? Thanks to Icekiup, I've made some progress, but I'm not getting consistent training. Whether running a runpod or locally, what parameters are used for training? Specifically, what about the dataset in terms of identity, body type, and number of images? And if we want to create a commercial version of LoRa for FanVue or other sites with this type of content, we need consistency and coherence in anatomy, as well as realism. I've tried SDXL, which is more or less manageable for this type of content, but I'm not getting good results because ZImage or Flux Klein, among others, make it difficult to create NSFW content for a character. I appreciate your answers, guides, and opinions.🙏👍