Jockerai 

Over the last few days, NVIDIA released an incredible new 4K upscaling model called PixelID. The results are genuinely impressive: outstanding detail, excellent identity preservation, and extremely fast performance. It can even run on GPUs with as little as 4GB of VRAM and upscale a 1K image to 4K in just a few seconds.

So far, it has received excellent benchmark scores and, in my opinion, it's currently the best upscaling model available.

I'll be sharing the original workflow with you, but while testing it I found a few limitations. That's why I built a custom workflow to solve those issues, which you can see in the preview images. I'll be uploading it very soon.

Default workflow for 1024 and 512 input images :

One of PixelID's main weaknesses is that it struggles with very small or extremely low-quality images. But in our community, "impossible" isn't an option. We always find a way.

To overcome this, I designed a workflow that combines Qwen, Flux Klein, and PixelID together, allowing you to upscale even the smallest and lowest-quality images while maintaining excellent quality and consistency by just a one click!

Hey my dear friends,

I wanted to give you all a quick update about my channel situation. After the hack and everything that happened, the whole recovery process has become much more complicated than I expected. At this point, I think I may need to go through a much longer legal process with my lawyer if I want to fully recover the channel, and honestly I still don’t know if the amount of time, energy, and money required will even be worth it in the end.

For now, I’m leaving the legal side of things to my lawyer and focusing on continuing the work itself. Because of that, we’ll start uploading tutorials directly here on this platform while I wait to see what finally happens with the channel. There’s also a chance that I may eventually move to a completely different platform for uploading videos, because YouTube’s recent policies and strict moderation have honestly become exhausting to deal with.

At the same time, several of my YouTuber friends told me that during the last couple of months their channels were either removed or demonetized because YouTube has started pushing a new policy wave where they claim certain creators are “not providing valuable content anymore,” and then suddenly remove monetization. In reality, for many creators that feels almost the same as losing the channel completely. These newer policies are becoming so aggressive that I genuinely think creators need a strong alternative platform, because years of hard work shouldn’t disappear overnight because of one new policy change.

I’m still moving forward with full power, even though mentally these past weeks have been extremely stressful and heavy for me. I hope all of you are doing well. We’ll start the tutorials again very soon, and I’m sorry for disappearing these past weeks.

Love you all and please take care of yourselves ❤️

🔶Full Tutorial for Upscale Workflow :

The Z-Image upscale workflow is officially ready — and this is not something you will find anywhere else.

💫Important note: the "No plastic-look skin" lora became free for all paid subscribera (Exclusive Access Tier) but I can't share it here. Please Dm me I will send you the safetenaors file to you.

More important than the workflow itself are the deep, precise explanations included in this tutorial. These details are exactly what allow you to push image quality far beyond the default limits and achieve extremely high-quality results up to 4K, even though Z-Image does not support 4K natively.

As you know, recently ComfyUI cannot be updated through the Manager and the process fails with an “Update Failed” error.

I have prepared a quick tutorial showing exactly how to fix this without reinstalling a fresh portable version. 

If you are experiencing failed updates, missing nodes, or workflows refusing to load (including Z-Image), make sure to watch this video. I will guide you step-by-step through a safe fix that updates your current ComfyUI.

Z - image workflow and tutorial : https://boosty.to/jockerai/posts/c4704123-ec7d-4d29-9525-4356d2621f37?share=post_link

The workflow and all detailed tutorial for insane Z-Image Turbo model has just been uploaded — and this one is a major upgrade across every angle.

main video is here: https://boosty.to/jockerai/posts/c4704123-ec7d-4d29-9525-4356d2621f37?share=post_link

In this tutorial, I walk you step-by-step through installing and running the Z-Image offline model on your personal computer. The model is 100% free, open-source, Uncensored only 12GB, and produces some of the most realistic photos you’ve ever seen. Even better:

you can run it smoothly on GPUs with just 6GB VRAM. 🚀

Inside the video, you’ll see:

✨ Ultra-realistic, shockingly accurate results
Its prompt understanding is unbelievable for a model this small.

🔍 Default 2K image generation + the best method to upscale to true 4K
I teach the exact process for getting maximum clarity and cinema-level detail.

🔄 A groundbreaking Face-Swap hack
By combining Z-Image + Nano Banana Pro, you can achieve a hyper-accurate, seamless face swap — something no previous model could do before this workflow.

🔓 Fully uncensored model support
Z-Image is completely uncensored and fully capable of generating uncensored content with high accuracy.

▶️All links and resources mentioned in the video are included below :

🔗Link for models are in the workflow so I don't mention them here again.
🔗 Download workflow :

➡️Note: My voice in this video was generated with the Vibe Voice AI, so in some parts it may pronounce certain numbers incorrectly. I just wanted you to know that it wasn’t intentional. If you also want to generate your own voice for any script on your computer, you can watch my Vibe Voice tutorial here : https://boosty.to/jockerai/posts/d46c2a15-936a-485f-afba-3d304f6b38c9?share=post_link

In this video, I demonstrate the new Outpainting (Image Extension) Workflow V3 using Qwen Image inside ComfyUI.

You will learn how to extend any image in all directions with clean, seamless results using FP8/FP16 or GGUF versions, ControlNet, VAE, and Lightning LoRA.
I show how to install the required models, adjust resolution, bypass resizing, modify pixel values, and write effective prompts for better anatomy, hands, and background consistency.

🔗Try Nano banana pro Full access (4K, aspect ratios, no limitations)

You’ll see how Nano Banana Pro handles:

• 4K photorealistic portraits with perfect skin texture

• Ultra-clean upscaling from destroyed low-res photos

• Professional lighting and background blending

• Realistic character interpretation (Mortal Kombat demo)

To celebrate Black Friday, you can access the biggest discounts of the year on Whop subscriptions from November 24th to November 28th:

• 60% OFF the 1-year subscription 🎉
• 40% OFF the 6-month subscription 🎉

• 20% OFF the 3-month and 1-month subscriptions 🎉

This offer is especially valuable for the 1-year plan, where you get a full year of unlimited access to all YouTube and Whop VIP content (Such as VIP workflows specific files, tutorials and more) from Jocker AI — for less than half the original price. 💎

Start & End Time ⏰
Starts: Beginning of November 24th
Ends: 11:59 PM on November 28th

For current subscribers 🤝
You can also use these discounts to renew your subscription. Just message me directly during the Black Friday period, and you will receive a discount code based on your current plan (1 month, 3 months, 6 months, or 1 year).

Introducing Change Background V2 and Put Any Subject in Any Background V2 — two brand-new, upgraded workflows for ComfyUI.

These versions are far more powerful than V1.

In the previous workflows, background replacement was done using Flux and Stable Diffusion 1.5, which sometimes caused loss of detail or altered the original product/subject.

But in V2, everything changes.