"But same as always, with generative AI, it's not about letting the machine do all the work, it's about helping you out finding new ways of integrate the new tech in your existing use case" That's what I'm talking about dude!
Hey man, im new to ComfyUI, is there a way to livelink the image from blender directly comfyUI instead of rendering the image in blender and then importing the image to comfyUI? that would be a really strong workflow i think with stable difussion.
I’ve always tried not to delve too much into animating with SD, mostly because it’d take much longer to record and edit the tutorials - and having a full time job doesn’t help. But if there’s demand for it sure, I can do it every once in a while.
My current rig is: - CPU AMD 5900x - 64GB RAM - GPU Nvidia 3080ti (bought during the Great GPU Shortage, if I were to buy it now I would prioritize VRAM over anything else, but at the time models weren’t quite as demanding). Generations are a bit slower than usual on video, because I’m recording my screen with OBS and myself with a Sony A7R IV.
There’s some ways to trim down my workflows. For example, in this one if you switch the photoshop to comfyUI node with a load image node, remove the pop up node, and don’t auto queue, the overall load on the GPU decreases a lot. You can also choose to sacrifice the weight of one of the two controlnets (I would lower the openpose, personally) so that the generations run faster.
Hey there! Until we get a node that does the same that Photoshop to ComfyUI does but for non-Windows OSes I’m afraid there’s no solution to this, apart from ditching the “real time” aspect of working in PS and using a simple load image node. I’m testing out a node that screen grabs, from any active window, but that one uses a lot of resources and I’m not sure it can run on macOS as well. I’ll look more into alternatives for macOS because it’s frustrating to not have a cross platform solution, but as of now I have found nothing yet.
@@risunobushi_ai Hi Andrea! Thanks a lot for your quick answer! I noticed some other custom nodes are not installed as well on Mac OS, like the OpenPose pre-processor or Depth Anything :( Your workflow is really interesting for me :)
@@risunobushi_ai Morning Andrea! Thanks! Ah, I was doing from scratch to learn! it was not working, opening your template worked fine, Only ComfyToPhoto node is not working :(
@@eias3d yeah, that node is windows only because of the way it is coded. I think a Mac version should be possible, but unfortunately I’m not a dev by any stretch of the definition, so I can’t change that
"But same as always, with generative AI, it's not about letting the machine do all the work, it's about helping you out finding new ways of integrate the new tech in your existing use case" That's what I'm talking about dude!
Thank you for the kind words!
Excellent workflow! 🤯
Thanks for the great contribution!! 🙌
Thank you!
Would love to see an animation workflow from this base tutorial.
Hey man, im new to ComfyUI, is there a way to livelink the image from blender directly comfyUI instead of rendering the image in blender and then importing the image to comfyUI? that would be a really strong workflow i think with stable difussion.
Well, my latest video shows how to do exactly that!
brilliant. thank you!
rly cool
The title is so confusing I didn’t know or understand what i was getting into but all i saw was ai aside that i wouldn’t have clicked on it🤷♀️
It’s kinda hard to make everything fit in 120 characters (without resorting to clickbait), but I’ll try to think of something a bit more clear
Would you create an animation workflow from this base tutorial? Amazing workflow by the way!
I’ve always tried not to delve too much into animating with SD, mostly because it’d take much longer to record and edit the tutorials - and having a full time job doesn’t help. But if there’s demand for it sure, I can do it every once in a while.
@@risunobushi_ai I want!!
@@risunobushi_aiit is
dont forget to refine those 16 fingers lmao
Ahahah yeah, I know!
Hi, do yo describe the specs of your gear in some video? If not, would you like to share?
My current rig is:
- CPU AMD 5900x
- 64GB RAM
- GPU Nvidia 3080ti (bought during the Great GPU Shortage, if I were to buy it now I would prioritize VRAM over anything else, but at the time models weren’t quite as demanding).
Generations are a bit slower than usual on video, because I’m recording my screen with OBS and myself with a Sony A7R IV.
@@risunobushi_ai Thanks a lot. I manage to run simple workflows on my current computer, but I'm in need of a better set up.
There’s some ways to trim down my workflows. For example, in this one if you switch the photoshop to comfyUI node with a load image node, remove the pop up node, and don’t auto queue, the overall load on the GPU decreases a lot. You can also choose to sacrifice the weight of one of the two controlnets (I would lower the openpose, personally) so that the generations run faster.
@@risunobushi_ai Of curse, I'm aiming for Ip-Adapter and that is a hungry one. I just placed my order. Thanks a lot for your help.
Hi Andrea! cool tutorial, is it possible to have a similar workflow for Macs? Thanks
Hey there! Until we get a node that does the same that Photoshop to ComfyUI does but for non-Windows OSes I’m afraid there’s no solution to this, apart from ditching the “real time” aspect of working in PS and using a simple load image node. I’m testing out a node that screen grabs, from any active window, but that one uses a lot of resources and I’m not sure it can run on macOS as well. I’ll look more into alternatives for macOS because it’s frustrating to not have a cross platform solution, but as of now I have found nothing yet.
@@risunobushi_ai Hi Andrea! Thanks a lot for your quick answer! I noticed some other custom nodes are not installed as well on Mac OS, like the OpenPose pre-processor or Depth Anything :(
Your workflow is really interesting for me :)
Are you importing the workflow .json file and installing the missing custom nodes via the manager?
@@risunobushi_ai Morning Andrea! Thanks! Ah, I was doing from scratch to learn! it was not working, opening your template worked fine, Only ComfyToPhoto node is not working :(
@@eias3d yeah, that node is windows only because of the way it is coded. I think a Mac version should be possible, but unfortunately I’m not a dev by any stretch of the definition, so I can’t change that
This is a really excellent approach to use 3-D software and generative AI as render output. Thank you for that amazing tutorial.
Thank you! Mixing AI with traditional 3D workflows is always fun.