- 20
- 69 807
reflekkt
เข้าร่วมเมื่อ 27 มี.ค. 2013
TouchDesigner Tutorials - Digital Art - Creative Technologies
Content Creation for Immersive Spaces - comfyUI & Colab Tutorial 2/2
The second part of this series is all about advanced AI content creation using comfyUI on Google Colab. The Colab Notebook is set up with everything you need for this workflow. Note that you need ~12GB space on your Google Drive.
For the workflow file, extra masks and style references check out my Patreon patreon.com/reflekkt and Instagram reflekkt_
Colab Notebook link: drive.google.com/file/d/1IrixrlWpMvNnfhqtJBYSHyVIMvo-wcix/view?usp=sharing / Mirror: www.mediafire.com/file/61hxlj87w8wtht7/comfyUI_colab.ipynb/file
AbsoluteReality checkpoint model: huggingface.co/digiplay/AbsoluteReality_v1.8.1/resolve/main/absolutereality_v181.safetensors
FlowFrames: nmkd.itch.io/flowframes
0:00 Intro
1:04: comfyUI general Information
2:12 Google Colab general information
3:22 Workflow & nodes explanation
12:40 Install & start comfyUI on Google Colab
19:08 Build the comfyUI workflow
31:57 Test render in 512x512
34:21 Longer test render
36:06 Make videos loopable
37:08 Change to more powerful L4 GPU for high resolution generation
43:49 Frame Interpolation with FlowFrames
46:03 Topaz for upscaling (if you don't have it, in my Patreon there is an intergrated upscaler included in the workflow)
49:05 Immersive room simulation
52:15 Outro & general information
For the workflow file, extra masks and style references check out my Patreon patreon.com/reflekkt and Instagram reflekkt_
Colab Notebook link: drive.google.com/file/d/1IrixrlWpMvNnfhqtJBYSHyVIMvo-wcix/view?usp=sharing / Mirror: www.mediafire.com/file/61hxlj87w8wtht7/comfyUI_colab.ipynb/file
AbsoluteReality checkpoint model: huggingface.co/digiplay/AbsoluteReality_v1.8.1/resolve/main/absolutereality_v181.safetensors
FlowFrames: nmkd.itch.io/flowframes
0:00 Intro
1:04: comfyUI general Information
2:12 Google Colab general information
3:22 Workflow & nodes explanation
12:40 Install & start comfyUI on Google Colab
19:08 Build the comfyUI workflow
31:57 Test render in 512x512
34:21 Longer test render
36:06 Make videos loopable
37:08 Change to more powerful L4 GPU for high resolution generation
43:49 Frame Interpolation with FlowFrames
46:03 Topaz for upscaling (if you don't have it, in my Patreon there is an intergrated upscaler included in the workflow)
49:05 Immersive room simulation
52:15 Outro & general information
มุมมอง: 308
วีดีโอ
Immersive Space Previsualization - TouchDesigner Tutorial 1/2
มุมมอง 66214 วันที่ผ่านมา
In part 1 of this tutorial series, we build a previsualization setup for an immersive space and learn what format your content needs to have, to be displayed correctly. You get to know what a pixelmap is and be prepared to set this up for your own venues or possible gigs. Part 2 will cover a more advanced content creation technique using comfyUI, an user interface for StableDiffusion. For the p...
Synthetic Sculptors - Interactive Media Installation (TouchDesigner & AI)
มุมมอง 4222 หลายเดือนก่อน
Throughout history, the essence of human beings and the way of its creation has always involved two humans. Yet, artists possess a unique ability to transcend this necessity, forging only one bond between creator and creation. In our exploration, we want to question the creation process, this time engaging the intersection of human and artificial intelligence through the medium of sculpture. He...
Interactive Eye Projection with Webcam or Kinect - TouchDesigner Tutorial 009
มุมมอง 8K4 หลายเดือนก่อน
In this TouchDesigner tutorial, you learn how to create an interactive projection, where an eye follows people passing by. I use a webcam and the bodytrackCHOP to track persons, but you can also use a Kinect or other data like mouse cursor position, noise or any other object. With free AI image generation tools like Bing Image Creator, you can create your own texture, I used PromeAI: www.promea...
Audio Reactive Instancing - TouchDesigner Tutorial 008
มุมมอง 7K6 หลายเดือนก่อน
Learn the technique of instancing combined with audio reactivity. This tutorial guides you step by step and explains easily how to manipulate the instance data with CHOPs for high performance and fps. Color Lovers Picker: derivative.ca/community-post/asset/colour-lover-palette-picker/62697 For more content check out my Patreon patreon.com/reflekkt and Instagram reflekkt_ 0:00 Int...
AI Projection Mapping Tutorial
มุมมอง 14K7 หลายเดือนก่อน
Learn an easy way of creating content for projection mapping with AI and how to map it on an object with TouchDesigner. This method is beginner friendly, quick and gets you impressive results. Have fun experimenting! PromeAI: www.promeai.pro/?vsource=i_8fr8d0f2m7 For more content check out my Patreon patreon.com/reflekkt and Instagram reflekkt_ 0:00 Intro & overview 0:45 Projecto...
Vibrant Color Tunnel - TouchDesigner Tutorial 007
มุมมอง 1.9K8 หลายเดือนก่อน
Beginner friedly tutorial to create a vibrant color tunnel in TouchDesigner. Simply said we instance a noise pattern and give it an interesting appearance by using several patterns for position, rotation and scale of the instances. Add some ambient occlusion and we get this beautiful color tunnel. For more content check out my Patreon patreon.com/reflekkt and Instagram reflekkt_ ...
Northern Lights - TouchDesigner Tutorial 006
มุมมอง 6K9 หลายเดือนก่อน
In this quick and easy beginner tutorial, we recreate the Northern Lights in TouchDesigner in under 5min. For more content check out my Patreon patreon.com/reflekkt and Instagram reflekkt_ 0:00 Intro & Overview 0:20 Base Lines 1:17 Post Effects 4:18 Outro #touchdesigner #northernlights #creativecoding
AI Projection Mapping Show
มุมมอง 8849 หลายเดือนก่อน
A short clip of a projection mapping I showcased at University of Applied Sciences Offenburg during the annual "Werkschau". The contents are mostly generated with help of different AI tools, the mapping is done in TouchDesigner. Thanks to my university@HochschuleOffenburgHSO for providing the technical equipment, projection surface and opportunity to show my semester work :-) For more content c...
Star Galaxy Vortex - TouchDesigner Tutorial 005
มุมมอง 9K9 หลายเดือนก่อน
Learn how to create a stunning starfield animation in TouchDesigner. It is easy and beginner friendly, as the network and techniques are very simple. For more content check out my Patreon patreon.com/reflekkt and Instagram reflekkt_ 0:00 Intro & Overview 0:23 SOP preparation and render setup 01:00 Noise texture as instancing source 01:40 Instancing 03:41 Post effects 05:15 Outro
Audioreactive Kinect Dancer - TouchDesigner Tutorial 004
มุมมอง 10K10 หลายเดือนก่อน
In this tutorial, I'll show you how to use the magic of a Kinect camera and audio reactivity to turn dancers into vibrant, dynamic lines on the wall. For more content check out my Patreon patreon.com/reflekkt and Instagram reflekkt_ 0:00 Intro & Overview 01:12 Basic grid and render setup 02:50 Kinect player index / silhouette 04:30 Line point data manipulation with CHOPs 07:45 Ad...
Interactive Nvidia Flex Fluids Inside 3D Objects - TouchDesigner Tutorial 003
มุมมอง 2.1K10 หลายเดือนก่อน
This tutorial shows how you can create interactive particles/fluids inside 3D objects with Nvidia Flex Solver and render them with a glassy looking surface. You can download the Om symbol I used here: sketchfab.com/3d-models/om-symbol-e73b69b1d7fb4d38be62e23a626842cb Of course it's possible to use your own 3D models or SOPs. For more content check out my Patreon patreon.com/reflekkt and Instagr...
Trippy Visuals - TouchDesigner Tutorial 002
มุมมอง 5Kปีที่แล้ว
Trippy Visuals - TouchDesigner Tutorial 002
Interactive Fire Particles - TouchDesigner Tutorial 001
มุมมอง 4Kปีที่แล้ว
Interactive Fire Particles - TouchDesigner Tutorial 001
Shifting Nature - Interactive Art Installation
มุมมอง 4832 ปีที่แล้ว
Shifting Nature - Interactive Art Installation
Very interesting! great ideas!
when i add my kinect information it doesn't follow the kinect. tried a few things, but when i add it to switch it displays the torus over and doesn't follow kinect chop
really great tutorial, thank you sooo much
How to setup the camera view in the right side_?
Great Content ! thank for sharing your work.
what gpu or cpu we need for make it run?
i get it, i need google colab, can you check the link of the collab ? It dont work
@@MaxiCrows I will upload a mirror in some minutes, the Dropbox link had too many downloads, didn't know there's a limit.
link should work now
This is an amazing tutorial man, thankyou so much.
awesome video! how did you group together nodes inside these blue, green and purple rectangles?
Thank you :) you can use shift+b or right click somewhere and choose 'add network box' to create the 'group'.
@@reflekkt_net simply life changing. subscribed.
hola algiuen me puede decir como saco el cuadro de Annotate por favor. soy nuevo en esto muchas gracias de antemano
hola buenas estoy hacinedo tu tutorial por cierto esta buenicimo pero tengo un problema en mi cuadro de CHOP no me sale body track por favor me ayudas muchas gracias
How to use the interaction with MediaPipe addon? what should be replaced in your network to use the addon?
Already sorted hahaha
Looks so cool, would love to try it out!
hello sir how to use web camera tracking effects in this tutorial video eye flow
my grid shows like duplicated, with instances on the "floor" and the others going up and down and it has a rogue pixel in the corner its fully way up! can you please help me identify what i need to change? great tutorial I suscribed!
Thank you so much for this tutorial! I'm a beginner and when i place "kinect1" with skeleton seated it doesn't show anything in its window, infact i can't move the fire tornado. How can i fix this? Thank you :)
Simple yet effective my favorite!
Great tutorial! Thanks for sharing
What projector are you using?
It's a NEC ME372W with a NP47LP lamp. I got it used from ebay for ~300$. It is not FullHD, but for hobby stuff or smaller parties it is really good, as it has a brightness of 3700lumen.
Hello, how can I prevent geometry COMP from displaying the final screen?😇
You can just turn off the render flag (the purple one in the bottom right corner of the operator) for the stuff that you don't want to show up in the render.
@@reflekkt_net Got it. Thanks.
Well done 👍🏾
Good Morning my name is Likhith calling from crossworks i have recently gone through your video in TH-cam named interactive eye projection it was wonderful can you please let me know how did you create the tail glow effect in it
many thanks!
what exactly do you do around 6:12 ? i see you drop the pallette but then it cuts and all of a sudden it turns into a ramp
Hi, you can go inside the dropped container from the ramp-palette-container and cut out the 'real' rampTOP. I don't know why it comes in a container when dropped from the component, as it's not so good to work with, but going inside, you will find the original rampTOP.
thank you so much for the tutorial and @FunctionStore for the complexop. i'm enjoying it fully..
Thank you!
Hello! How could i track two or more person? And how to add new sphere/eye for new person in the scene?
I'm getting stuck at about 2:30 because it's not animating for me at all, just a still image. How do I fix this?
Hi, sounds like you didn't animate the translate z axis parameter of the point transform with absTime.seconds It just happened 10 sec before, at 2:20.
Hey thanks for the tutorial I am facing an issue though My render never rendered anything… I finished the rest of the program to try and fix it later but so far haven’t been able to Thoughts?
To add to that my geo occasionally flickers a yellow exclamation mark And I’ve narrowed the cause to the VZ in instancing 2
This is an awesome concept and good for quick projects Please continue sharing projection mapping tutorials!
Hello! I'm on a mac, and BodyTrack COMP is unfortunately not compatible with a MAC! What would you recommend for subbing in the bodytracking aspect with a webcam on a MAC, so I can get my program to work like you have it here? Thank for the wonderful tutorial!
I second this Except I’m not on a Mac But the “body track” still doesn’t work How can I get it to work with the Kinect without that chop?
@@TheClumsyAcrobat let me know if you find a solution pls!
I just open the video and I recognize your voice directly so fun !
it is amazing 👏👏👏. i would like to know how to use it in fullscreen
You can use a windowCOMP and drag and drop the last null in it, then open as performance window 👍
How do I know what values to input for the range?
It's a good and important question. Check out my reply under ztime8038's comment, that should get you in the right direction. I think I'll also create a little video on that topic soon.
Can I use laptop web cam?
Yes you can 👍
There’s a problem that the eyes would run out of the screen. It just flow out. It seems like its not fully following my neck. Btw Thanks for your tutorial!
You need to limit the position data to the boundaries of your screen. Find out the boundaries by using the mouseinCHOP, move your mouse all the way to the left, note the x value, then all the way to the right, note the x value, all the way down, note the y value, and all the way up, note the y value. Now do the same with your camera or Kinect input, noting the position of your neck for example at the boundaries of the screen. With a mathCHOP just before the "positionNULL", use the Range function to convert your data. Lets say, the neck data all the way on the left is 0.2 and all the way to the right 1, and the screen position (which you checked with mouseinCHOP) is left 0 and right 1.5, then change the range from 0.2 and 1 to 0 and 1.5. That way, the position gets mapped properly on the output. If you move your camera or use a different output screen, you have to readjust these values. So for using it in an installation, make sure the camera won't move after setting it up correctly.
@@reflekkt_netI tried this and so far have not been successful at keeping the eye on screen
For some reason my eye doesnt move. It follows my mouse but it doesn't move from its spot.
Did you initialize and start the simulation? To do that, go to the bsolverCOMP that you created in the beginning (navigate one level up by pressing U or scroll up with mousewheel) and click on Initialize Sim and Collision Shapes, and then Start Sim.
@@reflekkt_net Thankl you so much, lol silly me
please upload tutorial again 😊 thanks
would this work with a video device in or movie file in top?
i’m trying to figure it out now but i’m running into an issue with the grid conforming to the silhouette
Yes but the input video should be the silhouette only, best in black and white.
Cool tutorial. Thank you.
Hello, it's excellent, but I have the jump when the person leaves the frame and returns, the eye makes a sudden movement, is there a way to soften the switch? thanks
Put the lag chop
ouuuu...this is so hot
yo it looks so cool
Hi, thank you so much for this!!
mine didn't turn out like this! hahha don't know what I did wrong..
Hi Reflekkt, thanks for the video great stuff, i have a question, is there a way to add this effect to a custom geo like adding it to a ball or pyramid so it looks like the ball and pyramid are on fire ?
Hi and thank you! Yes you can use any SOP as particle emitter/particle source. Just connect the SOP that you want to come the particles out of to the top input of particleSOP. if you add a sortSOP inbetween and set it to random, you the particles get spawned randomly across the points of your SOP instead of following an order.
thats great. How can we connect this to depth camera so the projection changes with human motion?
Thank you! You could for example use data from a depth camera to drive the switch, so the images projected changes depending on the distance of persons. Or composite the projection with a noiseTOP, which is connected to depth data that drives the transform parameter of the noise or something like that. You can literally use external data on almost every operator by referencing the value of your incoming data to a parameter of the network, almost no limits there :-)
hey, i am a long-term videomapper and it's cool to see how you implement AI into mapping in a very hands-on approach. in my opionen the 'creative fusion' of prome-ai is the key, bc it respets the givin' proportions of the image, which you need for using it in mapping. so far I haven't figured out a way to use stable diffion (or similiar) to work in similar ways. have you?
Thank you! I tried some workflows with Stable Diffusion and ControlNets, but couldn't get high quality results like with PromeAI yet. Right now I'm doing a lot of research tho for a video to video workflow in SD and am getting closer to what I want. Might do a tutorial in future if I figure it out properly!👍
This is exactly what I was looking for! Thank you so much!!
Great tutorial! Clean and simple
👍 nice