Use these tricks and techniques to take your ComfyUI diversity to another level!

One of the most important things when dealing with AI generated images is the versatility and the diversity of your generations. It usually takes several attempts and generation until you manage to get your desired outcome that can then be used as you base for the next steps.

In this video I will present you with 5 tools (or techniques) inside ComfyUI that will make you a master of variation! a legend of diversity and the king (queen or what in between) of quantity.

So enough with the chit chat.. lets get working.

P.S. Don't forget to like, subscribe, and share the good vibes ❤️

00:00 Intro
00:22 The power of the Batch
03:40 Auto Queue for a better view
07:50 Prompt Pro Prompts and a Text!
14:10 Prompt with template files
21:04 Saving the right way
23:55 Done and Done

#generativeart #ai #aiart #texttoimage #theaIart #the-ai-art #comfyui #stablediffusion #generativeai #generativeart #art #image
#tutorial #training #customnodes

[bertilorickardspelar]

When using SDXL Lightning models or LoRas I have found that sgm_uniform as scheduler works really well. I also always use more steps, 2step = 3-4 steps, 4step = 5-6 steps, 8step = 10-12 steps. When following this with Ultimate SD Upscaler it gives some great results fairly quickly. With sgm_uniform I find that there is always a sweetspot in the number of steps for a level of denoise which is usually around 0.2 denoise = 10 steps but it varies a little depending on the subject and detail level in the image. More details = more steps.

[Mehdimontahw]

Thank you for your explanation. It is easy for a beginner to understand and very accurate at the same time 😀

[Mehdimontahw]

We want a video about generating prompting professionally according to the model used an lora 😅