Image to 3D in ComfyUI is Fantastic

Image to 3D in ComfyUI is Fantastic
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
SOCIAL MEDIA LINKS!
✨ Support my (*・‿・)ノ⌒*:･ﾟ✧

character available at ko-fi.com/impactframes/shop
SD related civitai.com/user/impactframes
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
VIDEO LINKS📄🖍️o(≧o≦)o🔥
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
to watch tutorials
Join the Impact Frames fam! Subscribe now: youtube.com/@impactframes?si=...
ko-fi.com/impactframes
patreon.com/ImpactFrames
/ @impactframes
------------------------------------------------------------
please star the repos bellow.
Workflows
IF 3D CRM Playgrounds Alternative
comfyworkflows.com/workflows/...
IF 3D CRM Playgrounds
comfyworkflows.com/workflows/...
custom Node
github.com/MrForExample/Comfy...
github.com/flowtyone/ComfyUI-...
github.com/cubiq/ComfyUI_esse...
github.com/rgthree/rgthree-comfy
github.com/if-ai/ComfyUI-IF_A...
My extension for SD webUI
github.com/if-ai/IF_prompt_MKR
ollama
make sure ollama versions 0.1.25 for image to prompt or wait for them to fix the bug: github.com/ollama/ollama/rele...
github.com/ollama/ollama
Ollama WEBUI
github.com/ollama-webui/ollam...
ollama Models
ollama.com/brxce/stable-diffu...
ollama.com/impactframes/mistr...
ollama.com/impactframes/stabl...
ollama.com/brxce/stable-diffu...
CRM
huggingface.co/Zhengyi/CRM/tr...
Playground
huggingface.co/playgroundai/p...
Enjoy
ImpactFrames.
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
🔥NOTES
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Ollama commands
ollama -h
ollama -v
ollama list
ollama run name of the model

[@islamfoda]

Thanks a lot the video and the comment below helped me solve my issues after 2 weeks trying!

[@impactframes]

Thanks 🙂 glad you find it useful

[@LaminarRainbow]

Thank you!! 🙏

[@impactframes]

Thanks for the comment :)

[@Duckers_McQuack]

Nice! Had to make a separate comfy instance for it to work. How do i increase the mesh and texture resolution for the generations? And saw on the pages for the CRM that it can do lego's too? How do i accomplish that?

[@electronicmusicartcollective]

muchas gracias. Very good explaning

[@impactframes]

Gracias a ti por ver el video 🙂

[@offcutstudio6381]

Comfy UI 🤔 I should definitely check it out

[@impactframes]

Yes is the best stable diffusion tool available

[@PeteStueve]

never in my life such trouble with a module wow hate this nvdifrast so mcuh

[@impactframes]

Do you have the toolkit Cuda 12.1? The Mr forexample repo comes with a folder with the precompiled well for windows. It also has a bat file that install everything nicely.

[@umbifx7232]

Is there a method to improve the quality of the textures on the model? I uploaded a high resolution photo, my face is missing. Is it possible to pass the photos generated by the model through an upscaler or reconstruct them manually and insert them?

[@impactframes]

I think there is a new one called instant mesh that does the texture part better I think is called instant mesh. By the way you can also do what you are suggesting it will increase the quality. If you are not getting textures out is because your 3d app doesn't support imports I forgot the name I think GLIF I mentioned it on the video a couple of times that format includes a texture but some apps can't import it I show the addon you need in the video if you are using blender but yes both the texture and model need a refinement pass by hand

[@vivigomez5960]

Thank you for your videos. I got not problem installing on Linux

[@impactframes]

I am glad you got it working I would say is a hard installation but you made it that's a big deal since you can now use it in your projects congrats 🙂

[@nhannguyenphuc9146]

My boss sees the thumbnail of this video and proceeds to give me (a 3D modeler) a lecture on how dumb I am, how far behind I am on new tech, and that I'm not worth the money he's paying me for

[@impactframes]

Nah he needs you more than ever, at this stage the AI can't finish a ready for production model. It just helps to speed up the process since you can work over a more clear result than simple model sheets. On the thumbnail you can see the actual blender file too and the retopo process is started.

[@shiccup]

Lol this will be the case for everybody in a year.

[@impactframes]

@@shiccup Not sure, Modeling is too complex I think the AI will get there but the honeymoon will last a little longer

[@PSA04]

Sounds like your boss is not an artist and it sounds like you will be the boss with these new tools in the coming year. No one should say their employee is dumb. F that guy.

[@impactframes]

@@PSA04 I though he meant a hypothetical situation.

[@TyHoudinifx]

Hi there, where did you get that control button that controls the flow, disabling part of it with a click? This would be extremely useful. Does it work with a list of nodes or any node inside a group node? thanks

[@impactframes]

is Rgthee fast muter github.com/rgthree/rgthree-comfy/tree/main

[@TyHoudinifx]

@@impactframes thank you so much!!

[@osal31991]

Hi, I always learn new things about SD thanks to you. I am trying to use Reference Only Simple node to Animate Diff, but I could not find out a way to let Reference Only Simple fetch more than 16 batches (or frames) from original video data. it returns error like "IndexError: index 16 is out of bounds for dimension 0 with size 16". Could you tell me if there is any way to clear this issue?

[@impactframes]

i.imgur.com/6dydVFt.png without looking at the workflow I am not sure but to select the reference only index I used this setup before now animated diff message may not related to this

[@aa-wd1gy]

oh Fuckin GOD. U are my denisty. Combined Artisrt to Technical. We 'll rule the world!!!

[@PSA04]

This is fantastic, I am running into this error on the last step during the CRM Modeler & Viewer: "Error occurred when executing CRMModelerCuda: Ninja is required to load C++ extensions" I think I followed all of your steps, first timer using ComfyUI so this might be a dependency in other workflows.

[@impactframes]

If you are in windows the best is installing Mr.forexample repo because it has the precompiled wheels otherwise you need to install all the dependencies and compile the wheels and for that you need ninja. first activate your ComfyUI environment and then: pip install ninja

[@PSA04]

@@impactframes thank you for the quick reply, I am on Windows 10, I installed Mr.forexample repo the way you outlined and then running ComfyUI environment, I can't seem to type in the terminal. I have tried install pip install ninja in a standard windows CMD and a terminal from my comfyUI. Any other location that is best? * Looks like it defaults to the Local roaming folder, now I am running into this error: "Error occurred when executing CRMModelerCuda: DLL load failed while importing nvdiffrast_plugin: The specified module could not be found."

[@impactframes]

@@PSA04 you can try using ComfyUI manager there is a experimental pip package install button all you need to do is write ninja on the textbox and restart ComfyUI

[@PSA04]

@@impactframes circling back to let you know the pip installer via ComfyUI didn't work as expected. BUT using chatgpt to guide me through the proper steps to install Ninja to the proper "Path" folder worked. All new to me, great times ahead! Thank you.

[@islamfoda]

@@PSA04 chatgpt is a great idea it helped me solve the same problem too. Thanks a lot

[@PeteStueve]

have tried like so many workarounds and still... 'nvdiffrast not found

[@impactframes]

if you are compiling it yourself you might need First install Visual Studio Build Tools 2022/2019 with Workloads: Desktop development with C++ (There are a few JIT torch cpp extension that builds in runtime) visualstudio.microsoft.com/downloads/?q=build+tools also take a look at this post this looks like a good solution, hope it helpsgithub.com/flowtyone/ComfyUI-Flowty-CRM/issues/4#issuecomment-2001877240

[@GES1985]

Does it work OK with human head/face?

[@impactframes]

Yes it works well you always need to do some cleanup but yes you can plug a real picture even and have some of the projection done automatically it will save some steps specially if many characters are needed for one only maybe is not that big of a saver but for objects fare from the camera can be even be use as is

[@GES1985]

@@impactframes I'd love to see a video showing off how it does on realistic faces. It might be the new best way to bring a forward facing portrait into all of the other angles

[@ValleStutz]

11:53 "ERROR: Cannot find ComfyUI Windows Portable Python Embeded Environment "..\..\..\python_embeded\python.exe""

[@impactframes]

If you are using windows portable version you need to install it with python embedded. 1st You move to the folder of the custom node and then type cmd or open terminal on the current folder and do H:\ComfyUI_windows_portable\python_embeded\python.exe -m pip install -r requirements.txt --user where H is your hard drive letter you also need to do the same for requirements-cuda.txt

[@ValleStutz]

@@impactframes I have the normal version, not the portable. Okay, so I don't use the bat, I make a pip install -r

[@impactframes]

@@ValleStutz Right on the custom environment you just activate your comfy env and then just install the requirements like you normally do you can also try installing from the manager but the torch files get blacklisted so Idk if it works for the 3D pack because it is heavily dependent on the new pytorch libs

[@fujiahermawan8]

can the output fbx/obj?

[@impactframes]

No fbx but .obj and gltf are available and comes with the texture .mtl too.

[@cristinacolpos1735]

my generated mesh is gray with no image on it, do you know what could be wrong??

[@impactframes]

If you mean in blender you need to enable viewport shading like I mention at the mark 4:16

[@cristinacolpos1735]

@@impactframes I mean in CRM viewer, the mesh is gray

[@impactframes]

@@cristinacolpos1735 maybe try in another browser if you are not generating the image with the workflow make sure the image is square 1024 best an has four channels RGBA. Tbh I have trouble myself with the alt workflow not displaying for me but it loads fine in Blender

[@mostafasamir9472]

Thanks for the video really appreciate your effort to make something like this. It really handy :D After installation for comfy 3d pack custom node I got this error in Comfyui and it marked as red in the comfyui manager. Do you have any idea why this might happend? I'm using windows 11. Traceback (most recent call last): File "D:\Others\Software\ComfyUI_windows_portable\ComfyUI odes.py", line 1864, in load_custom_node module_spec.loader.exec_module(module) File "", line 940, in exec_module File "", line 241, in _call_with_frames_removed File "D:\Others\Software\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-3D-Pack\__init__.py", line 28, in module = importlib.import_module(f".{nodes_filename}", package=__name__) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "importlib\__init__.py", line 126, in import_module File "", line 1204, in _gcd_import File "", line 1176, in _find_and_load File "", line 1147, in _find_and_load_unlocked File "", line 690, in _load_unlocked File "", line 940, in exec_module File "", line 241, in _call_with_frames_removed File "D:\Others\Software\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-3D-Pack odes.py", line 53, in from .algorithms.dmtet import DMTetMesh File "D:\Others\Software\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-3D-Pack\algorithms\dmtet.py", line 6, in from pytorch3d.loss import chamfer_distance, mesh_laplacian_smoothing File "D:\Others\Software\ComfyUI_windows_portable\python_embeded\Lib\site-packages\pytorch3d\loss\__init__.py", line 8, in from .chamfer import chamfer_distance File "D:\Others\Software\ComfyUI_windows_portable\python_embeded\Lib\site-packages\pytorch3d\loss\chamfer.py", line 11, in from pytorch3d.ops.knn import knn_gather, knn_points File "D:\Others\Software\ComfyUI_windows_portable\python_embeded\Lib\site-packages\pytorch3d\ops\__init__.py", line 7, in from .ball_query import ball_query File "D:\Others\Software\ComfyUI_windows_portable\python_embeded\Lib\site-packages\pytorch3d\ops\ball_query.py", line 10, in from pytorch3d import _C ImportError: DLL load failed while importing _C: The specified procedure could not be found. Cannot import D:\Others\Software\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-3D-Pack module for custom nodes: DLL load failed while importing _C: The specified procedure could not be found.

[@impactframes]

Hi, thank you. It looks like you missing some of the wheels packages I think I explained in this video but you need to install some precompiled wheels for your system which will be win11 and matching the pytorch version you have installed or you will need to compile them yourself and you will need visual studio code. I am sorry I don't remember the exact steps but there must be directions on the repo readme on how to do it maybe I mentioned on the video.

[@lilien_rig]

I have RTX 4070, 32 GB of RAM and 8GB of VRAM and when I run the workflow I have "torch.cuda.OutOfMemoryError: Allocation on device" when I load the node "CRM Modeler (Cuda only)", is this normal ? good tutorial, you zre realy accrurate, I like this

[@impactframes]

I Think it needs about 12GB but here is a colab my friend Lovis did colab.research.google.com/drive/1i9wTQhO8MkuXJjiUiIG4b0N4uITiJpX2?authuser=0 it has Mr.forexample nodes which are better than this