PICTURE TO 3D - FREE, FAST & AWESOME! -🤩 [RE-UPLOADED]
Vložit
- čas přidán 20. 08. 2024
- In the wake of making 3D characters fast and easy the Facebook AI research, Facebook reality labs and the University of Southern California has developed an amazing image-based 3D human shape estimation which is driven by the significant improvement in representation power afforded by deep neural networks
---------------------------------------------------------------------------------------------------
*This was re-uploaded due to previous copyrights issues with CZcams
Get BLENDER Addons: bit.ly/3jbu8s7
---------------------------------------------------------------------------------------------------
CATCH UP WITH THE PAPERS:
shunsukesaito....
PHOTOPEA
photopea.com
REMOVE BACKGROUND
remove.bg
---------------------------------------------------------------------------------------------------
SCATTER ADDON 3.O
blendermarket....
COUPON CODE FOR 5% DISCOUNT: ASKNK
SCATTER 3.0 DOCUMENTATION:
sites.google.c...
GET TREE BOTANIQ ADDON!
blendermarket....
GET TRAFFIQ ADDON:
blendermarket....
GET TRUE TERRAIN:
blendermarket....
GET ANIMATION LAYER FOR BLENDER
• BLENDER 2.8x - ANIMATI...
blendermarket....
GET FREE 3D CONTENTS HERE:
• FREE & PREMIUM 3D RESO...
BLENDER 2.90 IS HERE!
• BLENDER 2.90 ALPHA IS ...
GET 3D UPDATES HERE:
• UNREAL ENGINE NOW SUPP...
If you want to help me create more videos, you can by supporting me with anything on patreon.
Patreon:
/ asknk
Gumroad:
gumroad.com/asknk
#2DTO3D
#neural
#network
#b3d
#3d
#BLENDER
#asknk
asknk,max,keyshot,Texturing,3d Tutorials,Cinema4d,Twinmotion,Arnold,Sketchup,BLENDER,substance painter 2019.1,asnk,FACEBUILDER,2D TO 3D,CREATING 3D HEADS,CREATING CUSTOM 3D HEAD,HOW TO CREATE 3D HEAD IN BLENDER,BLENDER 3D MODELLLING,BLENDER 2.82,BLENDER FREE ADDONS, face builder,
What a time to be alive!
THIS IS SOOOOO AMAZING, like god daayummm, what a time to be alive
This is exactly what I was looking for!
Thanks 😊
That's pretty cool, not ideal for game characters but like you said if you just need some characters to fill the space off in the distance this could work.
You can redo the topology easily. Good quick baseline
It dosen't work anymore on Collab. There is some problem while importing pretrained model. Some skilled pythonist required ;)
i wish there was a front end for this so it would be user friendly
this only support standing pose
character.
i hope they will modify in future
This tool give access topart of your google account to the tool's creator. There is a warning when clicking on the "Clone PIFuHD repository" button (1:48).
Always get an error message in Colab and nothing is generated.
So who's gonna turn this into user friendly app for the rest of us noobs?
i agree. I need that.
Yes Our App
@@whiterabbitrambo7666 Yes Comrade
@@whiterabbitrambo7666 communism
it doesnt work gives error at a certain point and result file is empty.
Do keep uploading such contents
I’m fairly new to Blender, and I find this very useful. Tho I’m having difficulty on the mirror part, could you explain it to me how to do it in a more detail way? Thanks in advance
and this process takes... well, depending on some stuff it will take a while.
wow thats specific!
Haha, I know that bodybuilder guy from the thumb :) Visited the same gym.
Cooool
name?
If the photos are in t pose and you also have a photo of the back is there a way to project the original image textures onto the models
im having an error when clicking runtime after. any help ?
The only thing I think is - WOW (in Owen Wilson voice)
THANKYOUUUU 🥺🥺😩😩
did anybody have some luck with this so far? show me results please
This is so powerful! Great video 👍
im not getting results folder!!
broo idk why i want able i didnt get the result folder nd ive been trying for the longest 😭 can u help me do one for mee ??
I'm looking forward to a time when this has a better UI.
Thank-you for the very interesting video! But is there a way for PIFuHD to create the model keeping the colours of the original?
use this on a comic book
I wish that they were textured from the image too.. are they?
Thanks for this information and Tutorial.
Tech and AI.....
What is going to happen to Software like Iclone, Character Creator?
Brilliant cheif, there are some grade one tossers on you tube who cannot explain anything - you are different very very clear - good work....
Mine is taking too long to run, how did you speed up yours?
What means "you cannot currently connect to a GPU due to usage limits in Colab."...😶
Can you make videos on depth map extraction from single images? I have seen a lot of research videos of those but never understood where to start from.
You mean like some sort of depth extraction for stereaocopic purposes? There is a Google Colab project you can use. Google for "creating depth maps from 3d images using Google Colab" is the Stereo Maker video in You Tube
@@ErikRicardoLC no actually i was talking about AI Based depth generator which can accurately output the depth from 2d video or images
@@AB-mv1mb Im interested in this as well
will it work with close up face portrait photo?
merci beaucoup c'est très utille
But what about texture?
Wow! Amazing!!! Thx for sharing! :D
Bro I was waiting for this! I managed to create amazing full body characters in Stable Diffusion of more than 6000px PNGs this is F. Amazing thanks to share!
can you share more about your process?
What is the first option to remove backgrounds called?
Very nice
Thanks for this. But every time I get a Cuda out of memory error. even with 256x256 image. It produces 3 image maps but no obj file.
This is amazing!
Is it like auto sculpting or something???
Great video but how can I apply the textures of the photo I'm using to the model?
UV project
I met a lot of issues. How to reset l start all process again from scratch? Because I closed more time the page trying to star from scratch but the folder and images are still in the root...
When running I get this on the Render the result part:
AttributeError: 'NoneType' object has no attribute 'shape'
Why am I getting this? I've been searching all over and can't figure it out.
Hi.
can you confirm that the image you used a clear and can be recognized.
@@askNK Yeah. I tried my own image but after that didn't work, I used the test image which gave me the same results. I deleted the Colab and copied it again following the same steps but didn't have any luck that time either.
@@askNK i have the same error:
net = PoseEstimationWithMobileNet()
checkpoint = torch.load('checkpoint_iter_370000.pth', map_location='cpu')
load_state(net, checkpoint)
get_rect(net.cuda(), [image_path], 512)
why my png always refused?
Nice
Hmm interesting. I might use this.
Link?
This is brilliant!
how we can do it in unreal engine pls tell
Hard to setup...doesn’t really work well
Thank u so much sir for sharing this video
The good thing about tools like this? It makes it easy to make quick rough reference models and such to work with.
The bad thing about tools like this? You get lazy individuals use it and then plug these unoptimized meshes into things like VRChat or Second Life or w/e else and it causes issues.
No need worry about bad things, this image->mesh ML research field is moving fast, just wait for better application.
What's the name of the program?
How do you make results appear?
This is fantastic
This is very good!
Mine won’t let me drag my sample image
great, but it sucks that people still are making all this to-be-compiled stuff all the time! Hello geeky people, it's time to wake up and make it user friendly!
How can i get a copy with colors? I have no experience on Modelling ,but will it be hard or realistic to paint the model from scratch?
Please say 🥺
What is the name app ?
does this only work with human bodies?
Facebook AI Research?
before i know it every user in that site has 3d model of themselves even after their death
Wow awesome tutorial, thank you so much
U are my hero.
how can i drag it bruh
So useful even after a year
Is it possible to use this program or another program that would make that image more close to the original picture? I mean having that picture animated?
So, next Games come with 3D scanned models and they will be animated by a AI. No longer need a long production with all this trackers on the actor’s face. Ecstasy 🤪
🤩
Link to the website pls???
link is in the description
Thanks so much 😝
Website name ?
Both amazing and scary. A lot of people will lose their jobs in the future.
This is my concern as well
lol. no this will make some 3d artists more powerful and games cheaper/faster finished :)
@@antiRuka how exactly will it make 3D artist more powerful?
@@silentwraith3464 same question.
Is this model free copyright? Can we use it for free?
Interesting stuff!
what about texture :(
This is Good
this is similar to the technology pixar uses
Simply
it keeps saying this could someone explain, please
net = PoseEstimationWithMobileNet()
checkpoint = torch.load('checkpoint_iter_370000.pth', map_location='cpu')
load_state(net, checkpoint)
get_rect(net.cuda(), [image_path], 512)
Same here
Its awesome
Just think of the fanart, so amazing.
Cool stuff!
how will u apply the texture
You can do some projection mapping.
Works cinema 4d?
Cant we do this offline without uploading online?Or a better question, when we copy to drive, can they see our pictures that we used ? the pictures are shared with the developers or not?
Would this work to 3D print myself ?
Noice
sick
Hey NK couple questions or anybody I’m don’t know what exactly turns the 2D image to 3D is it the external website that does it? Or blender that does it? Also I missed the part on how to actually import the pic into blender? Thanks Jake
-The external webpage makes it 3D, blender visualizes it
-To import to blender: File -> Import -> (choose same file type as the one created)
App name
Can this be used with video?
doesn't work anymore
very cool
Interesting. Nice. It works well. Thanks.
Holy shit what????
how i can install this? 🤔
...Very simple ?...o my.....rsrrsrs