Generative AI Examples from GTC25 XR Pavilion (#278)

* Create README.md

* Create README.md

* Create README.md

* Create README.md

* Update README.md

* Create tmp.txt

* Create tmp.txt

* Add files via upload

* Update README.md

* Delete GTC25/VirtualAthlete/images/tmp.txt

* Update README.md

* Add files via upload

* Create tmp.txt

* Add files via upload

* Update README.md

* Delete GTC25/PersonalizedCharacter/images/tmp.txt

* Update README.md

* Create tmp.txt

* Add files via upload

* Delete GTC25/EnvironmentDesign/images/tmp.txt

* Delete GTC25/DrawWithAI/images/tmp.txt

* Renamed from "GTC25" to "VFX", added spdx to all files

* Removed spdx from README.md files

* Delete VFX/.DS_Store

* Delete .DS_Store

Author: DMW-XR

VFX/DrawWithAI/README.md

@@ -0,0 +1,20 @@
+![](images/DrawWithAI-HeaderImage.png)
+
+# **Draw with AI**
+
+
+## **Demo Overview**
+
+This workflow showcases how NVIDIA GPU supports AI generations with control nets, Loras, and the connection between different applications all at the same time. Accelerated by NVIDIA Blackwell, drawings come to life interactively!  
+
+
+## **Key Points**
+
+
+
+* The ability to connect between AI generative processes and the sketch input from other applications like Adobe Photoshop, makes it easy to integrate with users’ current workflows.
+* Combined with text prompting, Control Nets give us the control we need over the generations. This is what makes our sketch guide the design of the generation
+* To turn the sketches into any style, Loras are applied. An amazing example of how finetuning can help the user define a style.
+
+NVIDIA GPUs help run all these processes simultaneously in order to generate interactively, as the user sketches their vision.
+

VFX/DrawWithAI/images/DrawWithAI-HeaderImage.png

@@ -0,0 +1,18 @@
+![](images/EnvironmentDesign-HeaderImage.png)
+
+# **Environment Design**
+
+
+## **Demo Overview**
+
+This workflow is using Generative AI to allow users to retexturize, relight any place objects into a new environment or the same one. Powered by NVIDIA Blackwell, this workflow accelerates design iterations with unparalleled speed and efficiency.
+
+
+## **Key Points**
+
+
+
+* Generate a specific material over a preselected object from an image - using CosXL for retexturing and SAM2 for accurate selections
+* Reinforce texture/material onto object if necessary with IPAdapters using strong style transfer
+* Relight the object in the same scene or a new designed space. Selected light colors and direction - Using IC- Relight and Stable Diffusion 1.5
+* Upscale your final design to add extra detail - Using SAM ( Segment anything Models), In Context Lora, And Clip VIsion with Flux Redux, Flux Turbo and Flux Dev Fill. 

VFX/EnvironmentDesign/README.md

@@ -0,0 +1,22 @@
+![](images/PersonalizedCharacter-HeaderImage.png)
+
+# **Personalized Character**
+
+
+## **Demo Overview**
+
+This workflow leverages Generative AI to turn any person into an illustrated character. It chains together 5 distinct AI workflows to produce character sheets and an animated turntable video. On an NVIDIA GPU, running on a powerful workstation, all of this content can be composed in  just under 1.5 minutes. 
+
+Finally everything comes together though compositing automation and SAM . Customized background elements, text and generated assets come together to create the final composition.
+
+
+## **Key Points**
+
+
+
+* Generate a specific character in a T-pose, based off of real life images - Using Instant ID & IP Adapter Face, Controlnet Open Pose and Stable Diffusion XL \
+
+* Generate a full body character in multiple angles using Diffusers based Multi View
+* Upscale Illustrations and create high quality faces with Generative AI upscale methods, Face Detailer and Face Detection Ultralytics , and SD Ultimate Upscale. 
+* Create unique character expressions for your character sheet using Live Portrait
+* Extract Key colors and elements from the scene using AI color extractors 

VFX/EnvironmentDesign/images/EnvironmentDesign-HeaderImage.png

@@ -0,0 +1,15 @@
+![](images/VirtualAthlete-HeaderImage.png)
+
+# **Virtual Athlete**
+
+This workflow is using Generative AI and automated compositing to allow users to design a custom esports jersey based off of their own colors, style selections, identity and team logo design. This will ultimately composite all together into a nice esport trading card
+
+
+## **Key Points**
+
+
+
+* Generate a specific logo based on user input name - using FLUX and FLUX LoRA
+* Create a custom realistic Jersey and put it on the person - Using SAM ( Segment anything Models) , In Context Lora, And Clip Vision with Flux Redux, Flux Turbo and Flux Dev Fill.
+* Relight the scene to the user selected colors Using IC- Relight and Stable Diffusion 1.5
+* Creating an export card layout  with user name and random number, as well as their custom logo