This looks incredible! Been waiting for something that would enable this kind of workflow. Looking forward to following how it develops! ( that install process complexity feels like it is going to be a barrier to people trying this still)
@@Sibanamush You are right it is a hard installation but is well worth investing a morning or even a day installing all the needed reqs this thing rules.
this looks awesome, but i think ill wait to try it until it becomes a node in the comfy manager...this set up is a lot and execution policy unrestricted is scary.
You have to be a little careful while installing via manager sometimes. As it could overwrite or remove dependancies. I think that's why they keep showing us the manual install.
@@juncandonoob here, can confirm, many times broke dependencies by installing nodes thru manager willy nilly. Fixed most by going into custom node manager, filtering broken nodes and just clicking on the 'attempt fix' on them. I gotta thank ComfyUI team for been really accommodating for us mentally regarded script kiddies 😂
@@omegablast2002 the unrestricted policy is more for when installing CUDA and some of the compile libs things like flash attention you can turned back off after otherwise you can't compile stuff and is even more dangerous using unsafe pre compiled wheels.
Also "NOT tested or compatible with PORTABLE comfy embeded python env" What other version of comfyui is there that can be run locally, I thought it was only the Portable version?
No you can install comfy on it's own environment and you have more control of the packages. But this works on embedded I just helped someone and it works you follow almost the same steps but from the embedded python env
I'm having some issues with my PC at the moment so forgive the lazy question. What ultimately is the output format of this? What can they be imported into? I am a 3ds Max user, can I use this process to create obj or some sort of compatible mesh that can be imported to 3ds Max?
@@clray123 you can use load 3d after updating comfy add node / 3d you need to connect something to those outputs a display text node or something so the wf run
@@Gounesh this creates angles for you what you can do to be perfectly following the a character sheet is make a model for each angle and then merge them I'm Maya delete the unused faces merge all the usable ones
This is incredible but I am sorry to say your written guide is all over the place. Please do a slimmed down installation procedure guide or a video doing all the steps at once.
@@NoPhilospher I haven't tried on Linux but is supposed to be easier. Just need the same packages nvdiffrast, xformers, flash attn all is available for Linux too
This looks absolutely amazing - great work, I can't wait to try it out!
@@pixeladdikt thanks yes I also blow away with the quality is this 👍
This looks incredible! Been waiting for something that would enable this kind of workflow. Looking forward to following how it develops! ( that install process complexity feels like it is going to be a barrier to people trying this still)
@@Sibanamush You are right it is a hard installation but is well worth investing a morning or even a day installing all the needed reqs this thing rules.
Why not just pay a artist...
this looks awesome, but i think ill wait to try it until it becomes a node in the comfy manager...this set up is a lot and execution policy unrestricted is scary.
You have to be a little careful while installing via manager sometimes. As it could overwrite or remove dependancies. I think that's why they keep showing us the manual install.
@@juncandonoob here, can confirm, many times broke dependencies by installing nodes thru manager willy nilly.
Fixed most by going into custom node manager, filtering broken nodes and just clicking on the 'attempt fix' on them.
I gotta thank ComfyUI team for been really accommodating for us mentally regarded script kiddies 😂
@@omegablast2002 the unrestricted policy is more for when installing CUDA and some of the compile libs things like flash attention you can turned back off after otherwise you can't compile stuff and is even more dangerous using unsafe pre compiled wheels.
How much vram is needed to run that!?
Awesome 😊
@@swannschilling474 thank you ❤️
This is very cool, but is there an easy way to give them bones for animation and physics? Can I import them to UE5 ?
Change format to fbx in blender and upload to mixamo for autorig and various animations
@@Ocean_Ftwyes that's a good way or you can use auto rigger pro or some other auto rig plugins
@@impactframes lazy bones could be a option too? just guessing
@marcluzon5007 yes that too lazy bones are very easy option
It would be nice to have a step by step tutorial video. It would make it easier. Thanks for this though!
Also "NOT tested or compatible with PORTABLE comfy embeded python env" What other version of comfyui is there that can be run locally, I thought it was only the Portable version?
No you can install comfy on it's own environment and you have more control of the packages. But this works on embedded I just helped someone and it works you follow almost the same steps but from the embedded python env
I'm having some issues with my PC at the moment so forgive the lazy question. What ultimately is the output format of this? What can they be imported into? I am a 3ds Max user, can I use this process to create obj or some sort of compatible mesh that can be imported to 3ds Max?
Looking at the ComfyUI output node in the video it says '.GLB' file format.
iirc, that's a widely compatible format for most modeling software
I wish you could show us the wireframe for these models...
What do you mean? They are shown at the beginning of the video
@@kunemann Oh I see, the dino robot, right? Sorry, well, I wish you could show us more then...lol
Amazing work! Just a quick question, would you mind making a tutorial on how to set it up on google collab? I would like to run it there
Yes I am adding some big updates it will take a couple of days to go through then I will make a notebook
@@impactframes Awesome! Looking forward to it
So do I get it correctly that the GLB preview and FFmpeg preview nodes are not working at all yet?
@@clray123 you can use load 3d after updating comfy add node / 3d you need to connect something to those outputs a display text node or something so the wf run
how much vram do need? Is 12 enough?
@@stas_lu it needs 16 but I am looking into adding some optimizations over this week
yes now works on under 12GB Vram
@@impactframes i have 3060 can i use this
@arash5550 is it 12 GB vram works no problems I don't know what vram is 360 but maybe it works on less than 12
Can this handle it multi-angle photos?
@@Gounesh this creates angles for you what you can do to be perfectly following the a character sheet is make a model for each angle and then merge them I'm Maya delete the unused faces merge all the usable ones
Impact frames ❤❤
Hi, theres any way to run this easily on a Google Collab? I'm bad at code 😢
cool!
@@DREX-Institute thanks this one is really good. This has to be one of the top 10 things in AI right now.
Is the text to 3d image implemented?
@@TheSORCERER-p9l Trellis hasn't released that code for text editing yet but I will make a workflow that does text to 3D at some point soon.
This is incredible but I am sorry to say your written guide is all over the place. Please do a slimmed down installation procedure guide or a video doing all the steps at once.
@@MotuDaaduBhai thanks I am not sure about doing a video on the guide but I'll try.
the future is now... omg!!!!!!!!! time to get back to Blender; the last I had was 2.6 I think
@@CoqueTornado good time to hop in
Lol vray was just released for blender 4.2! I just started getting into blender a month ago for animations! I have incredibly good timing!
@@shiccup super lucky
Bro!!!
Hi bro 👋 Hope you enjoying this one 🙂 sorry made awful typo wowsirry
aaaand lastly, can you extract the textures and normal maps?
Thankyou man.
why not making script that will do every thing automatic in 1 click so much steps i got confused
u have free sites u put image u get a mesh
thanks
thank you soo much, also how do i make it work on linux?
@@NoPhilospher I haven't tried on Linux but is supposed to be easier. Just need the same packages nvdiffrast, xformers, flash attn all is available for Linux too
Thanks! That looks amazing. Could you please provide a tutorial on how to run it remotely like in RunPod for example?
Thank you so much. Hope you make it optimized for low vram user . Its the best open source 3D AI
Yes there is some optimisation possible for this.
need google colab to run on 16 gb.
please learn in a video how to install step by stem, this video is not good to understand