You can download from ComfyUI from here: github.com/comfyanonymous/ComfyUI And extension from here: github.com/ltdrdata/ComfyUI-Inspire-Pack github.com/ltdrdata/ComfyUI-Impact-Pack github.com/shingo1228/ComfyUI-SDXL-EmptyLatentImage Workflow: (you can drag drop this image) github.com/ltdrdata/ComfyUI-extension-tutorials/blob/Main/ComfyUI-Inspire-Pack/workflow/seed-explorer.png
In theory, one can consider it in that conceptual framework. The seed is used to generate noise. The variation seed is a parameter that generates additional noise. Overlaying additional noise on top of the existing noise with a certain strength is what variation does. In other words, strength is a one-dimensional shift that determines whether it is closer to the base noise or the variation noise. The seed explorer adds the ability to change the direction in which the noise moves. If you move from 'noise a' to 'noise b' by 0.3, and call that position x, adding 'noise c' allows the noise to move again between the new direction of 'noise c' and 'noise x'. Through this, if you discover a favored noise point, you can move towards other noise points in the vicinity of that noise.
This is next level stuff that is beyond my comprehension. So, rather that inpainting or image to image, you make slight variation by manipulating the seed?
He won’t. He’s been doing this for a LONG time and he’s not JUST making tutorials. He codes the majority of the best comfyui nodes. People ask him all the time to add voice. Trust me. He won’t. We LUCKY he does this to begin with. Just appreciate and learn and be grateful.
Unfortunately, he's right. In fact, just refining the workflow and adding subtitles takes a considerable amount of time. I'd prefer to invest more time in the development aspect rather than video production. In truth, there are many areas in the documentation that need improvement, but I'm currently unable to address them.
@@drltdata I feel you. I too make silent videos and audience has been asking for voiceover. I'm aware that audio will improve thing but in the end I'm just one person and the development aspect takes priorities.
You can download from ComfyUI from here:
github.com/comfyanonymous/ComfyUI
And extension from here:
github.com/ltdrdata/ComfyUI-Inspire-Pack
github.com/ltdrdata/ComfyUI-Impact-Pack
github.com/shingo1228/ComfyUI-SDXL-EmptyLatentImage
Workflow: (you can drag drop this image)
github.com/ltdrdata/ComfyUI-extension-tutorials/blob/Main/ComfyUI-Inspire-Pack/workflow/seed-explorer.png
Awesome vid. Thanks for putting this out and all the hard work.
It's crazy how many good nodes are slept on because we may not even know about their existence :( - thank you for the video though!
Hello! Please ask: How to check the Seed value of the generated image in comfyui? Thanks!
It seems amazing but why the gem belt can be fixed by increasing the additional strength?
Maybe I got it now.With the Seed Expolorer I can apply variation seed as many times as neccessary until the result is satisfying, right?
Is this basically manual seed travel? I'm guessing the node is overlaying the additional noise with a very low oppacity?
In theory, one can consider it in that conceptual framework.
The seed is used to generate noise. The variation seed is a parameter that generates additional noise. Overlaying additional noise on top of the existing noise with a certain strength is what variation does. In other words, strength is a one-dimensional shift that determines whether it is closer to the base noise or the variation noise.
The seed explorer adds the ability to change the direction in which the noise moves. If you move from 'noise a' to 'noise b' by 0.3, and call that position x, adding 'noise c' allows the noise to move again between the new direction of 'noise c' and 'noise x'.
Through this, if you discover a favored noise point, you can move towards other noise points in the vicinity of that noise.
This is next level stuff that is beyond my comprehension. So, rather that inpainting or image to image, you make slight variation by manipulating the seed?
Yup. And there is one more feature. I'll upload tomorrow.
I really like your videos, could you add a voice, even AI/synthetic ?
He won’t. He’s been doing this for a LONG time and he’s not JUST making tutorials. He codes the majority of the best comfyui nodes. People ask him all the time to add voice. Trust me. He won’t. We LUCKY he does this to begin with. Just appreciate and learn and be grateful.
Unfortunately, he's right. In fact, just refining the workflow and adding subtitles takes a considerable amount of time. I'd prefer to invest more time in the development aspect rather than video production. In truth, there are many areas in the documentation that need improvement, but I'm currently unable to address them.
@@drltdata I feel you. I too make silent videos and audience has been asking for voiceover. I'm aware that audio will improve thing but in the end I'm just one person and the development aspect takes priorities.