You can Google it. For the traditional jobs. For AI , one of the task is dataset prep. it save lots of time for video dataset. And this one provide lot more detail description for video if we prep video dataset.
Setup and using their single image workflow. Everything goes through fine, no errors in Comfy or the Command Line but the Display Text node stays empty. Odd thing is if I hook up a show text node to the String Output of the Display Text node, I get a description in Show Text. Any ideas?
Tried building this into my workflows but it's nodes aren't passing in a format that any other node likes. Have you built any working workflows with this?
Yes , I find that out too. The text output are not the same String as other Comfyui String. I mod the code and return a normal String variable from the node function, after the try out of this custom node.
Oh man... Previously , I did a project , create stock assets website selling images and videos. I wish this AI model existing at that time. All tedious work are gone.
Thanks for your video, is really amazing I got this error after run install requirement CUDA_VERSION = "".join(os.environ.get("CUDA_VERSION", default_cuda_version).split(".")). What is this mean, do I miss anything? Thank you
Could u or anyone maybe help me out. I have a Intel 14900k CPU and a AMD Radeon RX 7900XTX GPU. Ive download comfyUI and it runs on my Intel CPU and not graphics card. Then I watch a video on using Flux AI model and when I went to use it it said I needed a Navidia GPU. So does anyone have a easy reliable work around for this? If so please let me know and I'll send u my email or if u have a good video/link I would be greatly appreciative. Thanks and be safe
@@TheFutureThinker aww man, so basically right now it's a no go for my gpu on anything like this? Also when they say Navidia gpu can if be any cheap brand 4080 or 4090? Or does it have to be specifically the Navidia brand? Say like a PNY GeForce RTX 4090? Thanks
ComfyUI works fine in Linux, and if you want it to run in Windows use WSL, and create a venv environment in your Linux distro to install ComfyUI with the necessary libraries I have gotten a 7900 to work in both environments with most of the major Image Gen and LLMs, notably InvokeAI, SD-Next, ComfyUI, msty, and so on. Look up the AMD instructions to create a WSL on their website, you have to follow the final step of copying the library file otherwise it'll spit a dummy out. I can run all the major models, though Flux is a resource hog and you really need 64Gb system RAM alongside the VRAM. Once you have used the AMD instructions you can use PyTorch 2.4 but do not use ROCm 6.2 in WSL as it is not supported yet, but it is under Linux.
@@ThirdEnvoqation man thank you for this, this is all new to me so I only understand about 1/2 of it. But I will definitely use your reply and Google everything u said to try to figure this out. U wouldn't happen to have any videos u made on this would u? Also if I had other questions would u mind helping a little? If not I completely understand. Thanks and be safe
Amazing, it does giving more detail than Florence 2.
Wow, btw could you give more specific examples of video captioning usefulness in industries?
You can Google it. For the traditional jobs.
For AI , one of the task is dataset prep. it save lots of time for video dataset. And this one provide lot more detail description for video if we prep video dataset.
Setup and using their single image workflow. Everything goes through fine, no errors in Comfy or the Command Line but the Display Text node stays empty. Odd thing is if I hook up a show text node to the String Output of the Display Text node, I get a description in Show Text. Any ideas?
In ComfyUI, it can't run this model with AI agent or functional calling.
Those features should use their code pipeline to execute.
Cool, we can use it for OCR.
Really amazing 😏
the model is censored? thks for your work!!
Haven't try with extreme content. Lol
Greattt! Thanks bro
You're welcome
Tried building this into my workflows but it's nodes aren't passing in a format that any other node likes. Have you built any working workflows with this?
Yes , I find that out too. The text output are not the same String as other Comfyui String. I mod the code and return a normal String variable from the node function, after the try out of this custom node.
@@TheFutureThinker I rewrote the Qwen2 VL nodes this afternoon and now I have working ComfyUI workflows using the Qwen2 VL nodes and models.
👏👏👏nice, the real power of Open Source community is to improve things , not only for freebie zombie.
@@TheFutureThinker Yes!
possible to generate subtitle from qwen?
That is audio to text.
It's like Google Gemini, and run locally not a problem.
Similar yes
@Benji @TheFutureThinker what about minicpm... it can do video too, minicpm v2.6
What is di video?
@@TheFutureThinker sorry typo
Oh man... Previously , I did a project , create stock assets website selling images and videos.
I wish this AI model existing at that time.
All tedious work are gone.
stock videos review and process on marketplace? 😄
Import is not working for me...
LLMoconception... AI genberated videos for AI related content.
Being able to understand Korean is as important a skill for work as understanding English.
Alibaba is not just any🙉🙉🙉
they deleted the original repository
video made by qwen2 xD
Thanks for your video, is really amazing
I got this error after run install requirement
CUDA_VERSION = "".join(os.environ.get("CUDA_VERSION", default_cuda_version).split(".")).
What is this mean, do I miss anything? Thank you
Looks like your Cuda need to update. No sure the detail of it. But mostly related to the Cuda toolkit update.
@@TheFutureThinker Thanks! I thought the same at the beginning, I try to download and install the latest toolkits but still don't work
Could u or anyone maybe help me out. I have a Intel 14900k CPU and a AMD Radeon RX 7900XTX GPU.
Ive download comfyUI and it runs on my Intel CPU and not graphics card.
Then I watch a video on using Flux AI model and when I went to use it it said I needed a Navidia GPU.
So does anyone have a easy reliable work around for this? If so please let me know and I'll send u my email or if u have a good video/link I would be greatly appreciative. Thanks and be safe
Yes, thats a problem for the most AMD users. Cause , many nodes and other ai softwares, they write code that taking care of Cuda architecture only.
@@TheFutureThinker aww man, so basically right now it's a no go for my gpu on anything like this?
Also when they say Navidia gpu can if be any cheap brand 4080 or 4090? Or does it have to be specifically the Navidia brand? Say like a PNY GeForce RTX 4090? Thanks
I still remember back then we use AMD for gaming pc.
ComfyUI works fine in Linux, and if you want it to run in Windows use WSL, and create a venv environment in your Linux distro to install ComfyUI with the necessary libraries I have gotten a 7900 to work in both environments with most of the major Image Gen and LLMs, notably InvokeAI, SD-Next, ComfyUI, msty, and so on.
Look up the AMD instructions to create a WSL on their website, you have to follow the final step of copying the library file otherwise it'll spit a dummy out. I can run all the major models, though Flux is a resource hog and you really need 64Gb system RAM alongside the VRAM. Once you have used the AMD instructions you can use PyTorch 2.4 but do not use ROCm 6.2 in WSL as it is not supported yet, but it is under Linux.
@@ThirdEnvoqation man thank you for this, this is all new to me so I only understand about 1/2 of it. But I will definitely use your reply and Google everything u said to try to figure this out. U wouldn't happen to have any videos u made on this would u? Also if I had other questions would u mind helping a little? If not I completely understand. Thanks and be safe