filmov
tv
Using Stable Diffusion to upgrade my fashion choices (edited on HP Z8 Fury)
Показать описание
New video for HP! Edited on the Z8 Fury @ZbyHP
Here's our process:
Collab with @noahrobertmiller
1. First we did a bunch of tests. I tried out different outfits, and we tested how we could use Stable Diffusion to reimagine those outfits. We figured out which kinds outfits work best (it's ones with lots of textures and patterns)
2. Then we recorded a video of me dancing different styles. I'm actually changing different outfits in real-life too, and editing a match cut for each one. That's why the outfit at the end is different from the one in the beginning. Each outfit underneath provides the "structure" for us to train on, so I tried to match up the general shapes as best I could
3. Then we rotoscoped me out of the video, to isolate me from the background, providing a cleaner plate for training
4. We used Stable Diffusion to re-imagine the outfits, prompting with different outfits that would match with the dance styles
5. Btw SD tends to do some wonky stuff with the faces (it doesn't handle human faces that well) so as a finishing touch we decided to rotoscope out the faces, and use my real human face instead
A bit about the machine we used -
The Z8 Fury @zbyhp is a monster of a machine. 56 core CPU, 4 GPUs, 128gb of memory, a process that took hours on my old machine took minutes. One of the reasons I like using AI in my editing workflow is the speed boost you get compared to traditional and 3D animation, and the Z8 pushes that speed even further. I'm able to iterate faster and try ideas almost as quickly as I can think of them. It's a super exciting time to be in where the tools are no longer the bottleneck to creative work. #ZbyHP #MadeOnZ #Intel
Client: HP
Editing @noahrobertmiller & @karenxcheng
Choreography @austingumban
Music: Ride or Die by Yarin Primak
Here's our process:
Collab with @noahrobertmiller
1. First we did a bunch of tests. I tried out different outfits, and we tested how we could use Stable Diffusion to reimagine those outfits. We figured out which kinds outfits work best (it's ones with lots of textures and patterns)
2. Then we recorded a video of me dancing different styles. I'm actually changing different outfits in real-life too, and editing a match cut for each one. That's why the outfit at the end is different from the one in the beginning. Each outfit underneath provides the "structure" for us to train on, so I tried to match up the general shapes as best I could
3. Then we rotoscoped me out of the video, to isolate me from the background, providing a cleaner plate for training
4. We used Stable Diffusion to re-imagine the outfits, prompting with different outfits that would match with the dance styles
5. Btw SD tends to do some wonky stuff with the faces (it doesn't handle human faces that well) so as a finishing touch we decided to rotoscope out the faces, and use my real human face instead
A bit about the machine we used -
The Z8 Fury @zbyhp is a monster of a machine. 56 core CPU, 4 GPUs, 128gb of memory, a process that took hours on my old machine took minutes. One of the reasons I like using AI in my editing workflow is the speed boost you get compared to traditional and 3D animation, and the Z8 pushes that speed even further. I'm able to iterate faster and try ideas almost as quickly as I can think of them. It's a super exciting time to be in where the tools are no longer the bottleneck to creative work. #ZbyHP #MadeOnZ #Intel
Client: HP
Editing @noahrobertmiller & @karenxcheng
Choreography @austingumban
Music: Ride or Die by Yarin Primak
0:01:35
UPDATE Stable Diffusion Locally (FAST!!)
0:00:47
how to get better results with stable diffusion
0:08:00
Best Practice Workflow for Automatic 1111 – Stable Diffusion
0:07:38
Take Your Images to the Next Level with Stable Diffusion!
0:14:48
Is Stable Diffusion 2.0 Worth the Upgrade?
0:06:15
Why everyone else's Stable Diffusion Art is better than yours (Checkpoint, LoRA and Civitai)
0:06:48
Stable Diffusion Realistic AI Consistent Character (Instant Method Without Training)
0:09:13
Stable Diffusion IMG2IMG: EVERYTHING you need to know IN ONE PLACE!
0:10:13
SD 3.5 MEDIUM is here, ComfyUI AI usage & comparison with SDXL
0:04:39
HOW TO MASTER AI UPSCALING/UP-RES | Stable Diffusion Web UI
0:15:44
Get Better Results With Ai by Using Stable Diffusion For Your Arch Viz Projects!
0:04:22
Stable diffusion up to 50% faster? I'll show you.
0:05:09
Stable Diffusion 2.0 | AUTOMATIC1111 Update Guide | NEW MODEL
0:10:10
1000% FASTER Stable Diffusion in ONE STEP!
0:08:05
How To Use Stable Diffusion To Improve Texture Of Realism Images (Tutorial Guide)
0:01:16
How to Update Automatic 1111 For Stable Diffusion
0:10:39
Instant Art Upgrade: 260+ Stable Diffusion Styles for A1111 & Forge [Free] #stablediffusion
0:08:27
STABLE DIFFUSION 3 NEW LICENSING!
0:11:09
Easiest way to Install Stable Diffusion XL 1.0 on Your Computer! (Best Tutorial)
0:03:29
Upgrade XFORMERS in Stable Diffusion - Automatic1111
0:11:00
How Stable Diffusion Will Change Your Photography
0:05:02
Stable Diffusion Performance Optimization Tutorial
0:16:08
Stable Diffusion 3 IS FINALLY HERE!
0:08:02
STABLE DIFFUSION 3.5 RELEASED!
Комментарии