Видео

FLUX has vegan pixels; SD3.5's pixels are tested on kittens.🙄

KSampler AttributeError: 'UserDefinedObjectVariable' object has no attribute 'proxy' from user code: File "F:\AAI\ComFYui\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-GGUF\ops.py", line 139, in get_weight weight = dequantize_tensor(tensor, dtype, self.dequant_dtype) Set TORCH_LOGS="+dynamo" and TORCHDYNAMO_VERBOSE=1 for more information You can suppress this exception and fall back to eager by setting: import torch._dynamo torch._dynamo.config.suppress_errors = True

Hi, where are the links to the workflows?

Are u aware that your video keeps breaking almost as if a game were running on extremely poor FPS. I thought it was something on my end but regardless of the browser I view it in, I have the same problem.

very well explained, thanks

Great video, even if I am watching 8 months later. Gonna try this out ….wonder how it compares to the tea cache and wavespeed. Question: how can I maximize the ram usage on my Mac. Using a Mac Studio with an M2 Ultra with 128g ram….i see that most of my runs in comfyui only seem to be using 60-70. Stats usually are cpu 2-4%, Mem 60-75%, and disc, 92%. I am running comfyui in python3 main.py so any tips for another Mac user would be appreciated.

I get an error: no module named 'gradio' dunno what to do :(

Can you use it on a video instead of an image

Is it free?

How do I fix hands of an existing image?

Where to get workflow ??

my text gen is ran on my ally rog x but i need silly tavern from my computer to use it, the text gen that is, i figure this be free verse runpod because 7b models run super fast on my rog x

Hello, thanks a lot for your video ! I wanted to ask if you occasionally made the runpod template for this workflow? And is it possible to launch this workflow with that local installation guide as it is, on runpod template which you provided in a description ? Thanks again :)

how to get it to detect faces with parts blocked by an object? (peeking behind wall, arm in front of face)?

If I wanted to run mistral model for writing content do you have any similar videos on setting up in runpod? Thank you for your video series thumbs up

I'm having trouble connecting to the UI my ports aren't coming up. Is this because the models are downloading?

No json file, I just checked Discord! Is it available anywhere or is there an update to this workflow?

❤

Comfy UI just looks like some node red crap, I would rather like a form or pure coding solution XD Give me an API to work with

Interesting. Glad that they're not the only game in town. I find myself increasingly antagonistic towards the idea that knowledge and information are commodities. Again we see moves that show more of a desire for domination than to truly assist and raise up humanity from the slime we still live in. Greed is a sure path to dystopia. Not surprised they pursue the most lucrative markets, and ignore the 'unwashed masses' of Android and Windows. Definitely some choice features coming up though (shared coding environment looks cool).. eventually. I would also add that Hunyuan Video came out and is working locally on 8G RTX cards (in my case 3070 No Ti.). It works, and looks ... ok for a free 1st crack at it. Not having used Sora myself I can't offer any direct comparisons, but I'll take a free and local model over setting up a new PaySoul account anytime. The idea I've been playing with lately is a personalized agent using some form of knowledge base (RAG or similar) embedded within the system prompt. By allowing the agent to keep notes and records in a vectorized form, it can slowly learn how I work, and the best ways to help me complete my mad science projects. In the future such agents will likely know us more intimately than any biological, since they will be with us from birth and throughout our entire lives. ☮, 💓,n 🎸, 🖖😎👍 P.S. 1st!

hailuo is my favourite ai video gen.. seriously img to video is so good even the newest kling stand no chance..

Your use of the multiply sigmas node is actually reducing detail. From the detail daemon repo: "factor: the amount that you want to multiply the sigma (noise level) by at each step. So, for example, if the first step has a sigma of 1, then using a factor of 0.95 would make this sigma 0.95. If a step has a sigma of 0.7, then a factor of 0.95 would make it 0.665. You probably want to keep this factor between 0.95-0.99. Lower values increase detail, but might also increasingly change the composition of the image, or introduce noisy grain. Setting it to 1.0 effectively disables the node." Also, you definitely should not be using a CFG of 8 with vanilla Flux. That needs to be set to 1.

Anyone got any tips for getting it to work in forge? 😥

Nothing of them look like anime, its not look like normal animes and even not like ghibli animation. And the example with blue lock doesnt make sense. I can also use kling ai and can use blue lock picture, it will happen the same. When i have already a picture like your blue lock example , i can do this with every ai video program

You should use ultimate sd, its much better !!!! This video is not helpful....

Let's go! 💡

Loved the examples!!

image-to-video feature looks so smooth and cinematic, cant wait to try

game-changer for AI creators 🔥

this video also could have been improvec by showing the end result of using your workflow. it seems like you showed us a bunch of stuff, an original super ai-ey photo, but never actually got a good result image that looked real. kind of a bummer.

a really great video would be if you went through from the very start how to created this workflow. your thought process, how you decided what nodes to pick next, how you decided on the settings within that node. there are not many tutorials on showing how people create the workflow, which would be really helpful to those who want to learn how to create workflows for ourselves. thanks for this video, definitely addresses a real problem

Nice Video! For anime movements the best model for now seems to be vidu, but with less consistency.

👋 hi

PLEASE see my comment on your LORA video regarding helping mac users. NOW after this one I really like your videos and gonna sub in the hopes that you start making videos geared to MAC users.....cus thats where all this stuff will eventually end up with the rest of my film industry mac using brethren.

What could best be used in a faceswapping workflow to retain facial hair of the original image? I'm using canny as controlnet but it shaves off mustaches and beards partially. Eager to find a solution.

thx! how can i use lora with your workflow?

did you add all that empty screen at the end to make your video meet minimum ideal length? or was it an accident?

Thanks bro, couldn't find any info on how to connect runpod to omnigen without a paywall. I wish you great fortune and success in everything you do.

comfy ui again. need a1111 tutorial.

perfect tip for text caption thanks

mhh, are you just showing the informations from the websites? I'm not seeing you hitting "Queue Prompt". Did you ever made it run? I'm just getting a black "video". Curious how it looks... the run through.

it just AI there is no better. you got billions seed. change it result change LOL. remember this when do yet again other hunyuan vs somthing new

imagine in 50 years we will be able to do, what photoshopper already can. until then, please pay us at least 12 bugs a month

so this is like an ad for that website?

👋 hi

Good video! Is it able to run on google collab note book?

hi friends....i cant select destination in the training page...pls help me

We don’t want ai in movies

Using a 1080ti 11gb, comfyui for windows, video 256x256, 16 ateps 45 frames, it works but it is too slow, if try using wsl it crashes all the time

Is Kaiju Gen a website? I don’t see a link for it.