Loved this tutorial—simple, practical, and packed with creative inspiration for getting the most out of Image 2. Thank you.

The real test is always the workflow, not just the model. I like using Krea alongside these tools to compare edits, references, and styles without making the process overly complicated.

As a designer this is what I need! The exact youtube shows how to use the features! Instant subscribed!! Next video please teach how to generate 3D assets high quality production for tech/fintech please!

You can add  the following for the exact face consistency - lock the face features, zero alterations, 100% same for consistency. Ensure that the subject front  body pose is fully visible (head to toe including the footwear). Ensure that there is sn invisible camera is meticulously placed at an eye-level all the time. Ensure that the final image should be pristine smooth, clean, and clear. Ensure that the image looks natural and blends with the background perfectly.

What you can also do is ask ChatGPT to create universal prompts and save them. You can then use (Copy and Paste) for all your images. I call it a Universal Master Prompt Library

I learnt a lot from your video. Thanks 👍

Super insightful information and great effort as always. Many thanks!

Thank you so much.  I learned a lot!

Thank you! Amazing content as always! Respect!

Hello friend, I really enjoyed your content. I'm writing from Angola, Africa, and we speak Portuguese.

Nice results. Thank you for the video.

Great video! Thank you!

Thank you, Its so helpful.

This video was great.  The reason ChatGPT will create an image when you are asking them to create a prompt after you've asked them to create a picture is because there are two different models competing.

When you ask them to create a picture the model changes to the image gen.  If you then ask them to write, the img gen tries to create a picture of writing because the IMG gen is not an LLM.

The error is the human, not knowing how model shifts work.  Pretend you're working on a car and you suddenly want to cook a sandwich.  While you COULD technically turn the engine on until it's hot and cook on the radiator (or whatever) your meal will taste better cooked in a kitchen.

Always tell the AI to switch back to what you want.

"Thank you for creating that picture, please switch back to texting so we can work on the next project.  Please do not create any more images."

amaaaaaaaaaaaaaaaaaaaazing 🔥🔥

Pretty awesome, I used it to generate images for a dataset. Good consistency. Tends to over-saturate color like Flux Klein Edit, but I didn't use the thinking mode, so maybe that would've helped.

Nice!

Thanks for the presentation and explanation!  🧡

Your slides are so great! It's amazing how much detail you put into your work! :)

So, to be perfectly honest: In 20-30 years at the latest, I won't be able to verify this anymore, not only because ChatGPT won't exist. ;)

Best 😊

PS: See you in "Marrakech",  on the 3rd. 👋

PPS: Now I need to order some ice cream. 🍦🐰

Thats the problem  u cannot have a vector of a label design .U get an impression. Great at creating  ideas but how do u go to print packaging  or lable  without  still using Adobe .

can you do a seedance learning one if you have time? 😬

Nice one, brother. By the way, I’m a big fan of your channel—especially your discussions about ComfyUI. So far, ComfyUI is the only AI platform/interface that really helps me create and edit AI images and videos using open-source models.

But I’m also interested in closed-source models like Nano Banana, GPT Image, Seedream, Seedance, Grok, Kling, etc. Maybe you could make some tutorials on how to use those closed-source models within ComfyUI? I know it requires API credits—that’s no problem 🙂

I’m just very comfortable and familiar with how ComfyUI works, and I’m sure a lot of people feel the same way. What do you think?