Video není dostupné.
Omlouváme se.
Depth Anything ControlNet Create AI Image That Looks Stunning In ComfyUI
Vložit
- čas přidán 6. 08. 2024
- In today's video, we have an exciting topic to discuss - Stable Diffusion and the groundbreaking Depth Anything model. We'll dive deep into the world of AI image and video creation, exploring how ControlNet and Depth Map technology are revolutionizing the field.
Depth Anything:
depth-anything.github.io/
huggingface.co/spaces/LiheYou...
Depth Anything Test Lab ComftyUI Workflow:
www.patreon.com/posts/depth-a...
Stable Diffusion ComfyUI - How To Create Workflow With Clean Layout Less Lines
• Stable Diffusion Comfy...
Chapters:
00:00 - Introduction Of Depth Anything
02:48 - Installation Of Depth Anything ControlNet
04:40 - Update ControlNet In ComfyUI
05:03 - Apply Depth Anything In Workflow
06:00 - Depth Map ControlNet Testing And Comparison
12:54 - Testing In Hugging Face Demo
Join me as we take a closer look at the collaboration between TikTok and the University of Hong Kong in their project, Depth Anything. This innovative model has been trained on a whopping 1.5 million identified images, along with an additional 62 million random images. The extensive training data has equipped this model with the incredible capability to detect elements in an image and accurately recognize the distance between objects and the camera view.
If You Like tutorial like this, You Can Support Our Work In Patreon:
/ aifuturetech
Discord : / discord
For those who want to dig deeper, we'll also provide a link to the research paper in the video description. You can explore the detailed metrics, formulas, and other technical aspects of the project.
Now, let's move on to the practical side of things. We'll navigate to the GitHub page where you can access the ControlNet models for Stable Diffusion and Depth Anything. We'll guide you through the process of downloading and integrating these models into your own projects. From the compatibility with Automatic 1111 to Comfy UI, we'll cover all the essential steps to get you up and running with this cutting-edge technology.
In the latter part of the video, we'll demonstrate the power of Depth Anything in action. Using Comfy UI, we'll showcase the results of applying the Depth Anything model to various images. From animated workflows to stunning piano compositions, we'll witness the model's ability to accurately identify the distance between objects, create layering effects, and bring out intricate details in the generated AI images.
Additionally, we'll compare the results of the Depth Anything model with other Depth Map models like Sui and Midas. You'll see firsthand the superior performance of Depth Anything in terms of object recognition and detailed depth mapping.
So, if you're a content creator, a tech enthusiast, or simply curious about the latest advancements in AI image and video creation, this video is a must-watch for you. Join me on this journey as we explore Stable Diffusion, Depth Anything, and their potential to revolutionize the way we create and experience visual content.
Don't forget to like this video, subscribe to my channel, and hit that notification bell to stay updated with future tech-related content. And as always, all the relevant links and resources mentioned in this video can be found in the video description below.
Thank you for watching, and I'll catch you in the next one! Stay tuned for more exciting tech content on this channel.
#comfyui #stablediffusion #depthanything #controlnet
Depth Anything:
depth-anything.github.io/
huggingface.co/spaces/LiheYoung/Depth-Anything/tree/main
Depth Anything Test Lab ComftyUI Workflow:
www.patreon.com/posts/depth-anything-97555225?Link&
Stable Diffusion ComfyUI - How To Create Workflow With Clean Layout Less Lines
czcams.com/video/L0dy4hcHP-A/video.htmlsi=58yuU9NcOwWU--BT
Very clearly informative. Love your coontents!
❤great tutorial, thanks!
thanks for this comparison
thank you this was helpful and infromative
Glad it was helpful!
What is the problem with the 2 unsafe files?
Great!
You can use ICAT to easily compare all six of them both the bw and the colored pictures.
nice content.
Great video thanks. It was amazing, even tho it's not looking as good as Merigold Depth (visually in human eyes) but the depth created is really good and better than Merigold for controlnet and can be used for SDXL sort of.
Depth Anything pre-proccessor (at any SDXL resolution depending on what you need) + MiDaS model for controlnet + SDXL = win. it works extremely well with Pony models but hard time doing complex poses on other models.
Great comparison of different Depth Maps. Where do you get your stock photos and videos?
It is not stock photo😄 , those image are my previous gen Ai image. But thank you you think those are stock images😁
I try to apply Depth Anything but I got an error.
I know it casue by my location Internet fault. I try to creat the folder path manually that like your video 5:51 emerge（as same as I console displayed but can‘t connect）but in windows folder name "/" is banned，so would you display your folder name in “ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts\ "LiheYoung/Depth-Anything" name for me?
I am trying to get along your video, but i am stucked cause i cant find a lora that is in your workflow file, called add detail.safetensors. Where can i find it please?
You can find it here : civitai.com/models/58390/detail-tweaker-lora-lora
, Or just use other lora you prefer.
The "depth anything vit l14", etcetera, from the preprocessors didn't automatically downloaded for me, instead I got an error showing I didn't had those files. Do you know how to do it manually?
Should be able to check Download Model in Comfyui Manager. Try this route 😉
@@TheFutureThinker I did but I didn't find them. :( Thanks for replying. Do you have any other ideas?
is depth anything comaptible with sdxl? or just sd 1.5?
It train with SD1.5 model only
hey so before you had responded i decided to try it with SDXL models in ComfyUI and it works pretty good! I just changed the resolution from 512 to 1024!@@TheFutureThinker
Yes, change models for all components and resolution! Very easy right?
This space has 2 files that have been marked as unsafe. 💀
discord link not working
discord.gg/s8kp4P2d
Try this one 😊
still doesn't work
@@TheFutureThinker
I followed your instruction, but when I run depth anything, it does not download but gives me an error saying "processor:
[Errno 2] No such file or directory: 'C:\\Users\\meeee2\\Desktop\\SD\
ew_ComfyUI_windows_portable_nvidia_cu121_or_cpu\\ComfyUI_windows_portable\\ComfyUI\\custom_nodes\\comfyui_controlnet_aux\\ckpts\\lllyasviel\\Annotators\\.huggingface\\download\\ZoeD_M12_N.pt.c97f94c4d53c5b788af46c5da0462262aebb37ea116fd70014bcbba93146c33b.incomplete'"
do you know how to fix it?
Try moving your ComfyUI_windows_portable folder that contains everything in comfy directly onto your local disk so it should be C:\ComfyUI_windows_portable and run the prompt again. This worked for me for whatever reason.