Ive been using comfy for quite some time. This is very useful for those starting out. I dont know how many hours I've spent installing diff addons and models. Great work. Thank you.
@@myandrobox3427 basically it's gpu intensive (CPU is waaaay to slow) so the more vram on your GPU the better. Nowadays, 12 gig min. You can use 8 for sure but it will be slower and generating large images(upscaling) may result in out of mem errors.
@@haripopstar whatever ur budget allows for 12 or more gigs of ram, there are lots of reviews out there for mid-tier cards. Nvidia may be slightly faster.
I'm new to the node based and really don't know a whole lot. but after watching this video and downloading the studio it was very easy to follow and do, everything seems to be working correctly as you showed in your video. I am 100% satisfied with the studio you did a lot of work to it and you did a very good job
Good looking and very useful for actual designers! This is exactly the kind of stuff Patreon was made for. Reward a talented workflow designer for his time and effort in putting together this great set of workflows.
First of all….thank you for putting in this much work to create a ComfyUI user interface that is a lot easier to handle! But to my understanding, the interface along with everything else you have put in, works only on a machine with an Nvidia card (2060) for now? Hope you can create ComfyUI studio for us Mac owners who would like to try the same. 😉 Kudos for your work!
Thank you so much for your kind words! Yes, the current version of ComfyUI is optimized for machines with Nvidia graphics cards, such as the 2060 or higher, for faster image generation. A Mac version is possible, but right now, we are focusing on advancing the Windows version to higher levels. Image generation on Mac graphics cards would be slower, so for serious image creation, an Nvidia card is recommended.
@@ComfyUIstudio Thanks for the update on your work! I fully understand Nvidia as a first choice to bring the new interface out in the wild ;) I might be considering a Windows machine in the near future, just because of this! Thanks again for your reply and keep up the good work!
Can you please show how to change SDXL nodes to Flux? I don't care about time of generations, I need the maximum quality. SDXL does not do it for me, unfortunately.
I've had exactly the opposite experience. Using Forge with same model, Forge is faster. The "advantage" of Comfy is the possibility of much more complex workflows. Like generating an image in Flux, then upscaling in SDXL all in one go. The only reason I am using ComfyUI is it seems to get the newest functions faster than anything else. That's the benefit of node based systems, you don't have to reorganise the interface. Having said that, I use Fooocus and Forge more than ComfyUI, because I rarely experience errors. It's VERY easy to introduce errors in ComfyUI, because of the lack of structure.
This is amazing work! Question. Would I be able to use custom Loras and models from civitAI for both SD and Flux? Also is there a way to change the face expressions when working with any of the ID workflows?
Unfortunately, ComfyUI might not run smoothly on a GTX 1650 due to its limited VRAM. You may experience slow performance or crashes with larger models or more complex workflows.
Another question…can you create discount promo images (like an “50% discount in any order before Monday” with a product image background ) for an e-commerce? Thanks!
Im in your patreon, downloaded the comfyUI studio but i had error on checkpoint, do you know how can help me? i have to click on download all models or its already in my folder?
@@victorandrade2160 The checkpoints have not been downloaded. They need to be downloaded using the download all script. Sometimes the checkpoints don't fully download due to network errors. Therefore, the faulty checkpoint must be deleted and downloaded again.
@@ComfyUIstudio Thank you for fast response! I am doing it again, i noticed there are 2 checkpoints, flux nf14 and realistic vision, i dont understand very well, i must use realisticvision40 right? if i do, is it "connected" with flux features?
@@ComfyUIstudio i cant make it works, the following error appeared: !!! Exception during processing !!! Error while deserializing header: MetadataIncompleteBuffer Traceback (most recent call last): File "C:\Flux\App\ComfyUI\execution.py", line 323, in execute output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Flux\App\ComfyUI\execution.py", line 198, in get_output_data return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Flux\App\ComfyUI\execution.py", line 169, in _map_node_over_list process_inputs(input_dict, i) File "C:\Flux\App\ComfyUI\execution.py", line 158, in process_inputs results.append(getattr(obj, func)(**inputs)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Flux\App\ComfyUI odes.py", line 539, in load_checkpoint out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=folder_paths.get_folder_paths("embeddings")) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Flux\App\ComfyUI\comfy\sd.py", line 524, in load_checkpoint_guess_config sd = comfy.utils.load_torch_file(ckpt_path) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Flux\App\ComfyUI\comfy\utils.py", line 34, in load_torch_file sd = safetensors.torch.load_file(ckpt, device=device.type) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Flux\App\python_embeded\Lib\site-packages\safetensors\torch.py", line 311, in load_file with safe_open(filename, framework="pt", device=device) as f: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer Prompt executed in 0.12 seconds
if only these studio have specialized workflow for low Vram, like lcm loras, etc... i guess that will be add more viewers and users since many have low Vram
@@ComfyUIstudio can you cancel my patreon so i dont get charged anymore because they said they found on my other account i did not follow the guide line. im the one who you ask me to goto your facebook page
I have a 12th generation Intel core i7 pc with 16gb of RAM. And a Geforce RTX 3060 graphics card with 6gb of dedicated RAM. It will work or it is insufficient in terms of capacity.
@@TheColonelJJ there's a free pack called Epic pack. Try it out. But it's just workflows. The studio contains the ComfyUI app. Preinstalled workflows and model download script.
Ofc it wont vs a 3090 but i have a m2 pro macbook pro with 16gb of ram and i use forge rn (not for flux) and Draw Things application (for flux) and it works great especially if you use 4 step lora for flux dev it takes a minute or just under per gen ill take that speeds on a 16gb shared memory silicon chip any day @@PhillipMarquez
I was wondering the same thing as well but his response always seems to be the same, so I'm guessing the answer is "there is no Mac version" since he's convinced ComfyUI doesn't run well (IMO it's totally acceptable, even with FLUX.1 model).
This is a 1-click install version with 20+ workflows, including models. Just like shown in the video, it includes Flux (11GB), RealVis (7GB), and several smaller models for ControlNet and other features. Plus, we've just added a new workflow that’s not available anywhere else-the two-person Instant ID face-based generation. Check out the latest video on our channel to see it in action!
There will be a workflow for image-to-video generation soon! I need to test it first before sharing, as some nodes could potentially cause issues with ComfyUI if not set up correctly. Stay tuned for updates!
yeah thats the best organised noodles spagethi interface maybe... you realy have to love them... even if comfy is realy extremly powerful, i think it s unusable for teams... because each time you create something... the noodles are entangled in another way.. dont get me wrong... comfy is powerful and i like it.. but it doesnt work intuitive enough. even to use tthe word intuitive in one sentence with comfy... doesnt work..
![I Replaced ALL my ADOBE APPS with these [free or cheaper] Alternatives!](http://i.ytimg.com/vi/5EfqHg49kMk/mqdefault.jpg)
All the tools needed to make a consistent character with many variations of poses and clothing. Very useful, great job.
Ive been using comfy for quite some time. This is very useful for those starting out. I dont know how many hours I've spent installing diff addons and models. Great work. Thank you.
What's the specs machine require to get this going? Did you run into any issues?
@@myandrobox3427 basically it's gpu intensive (CPU is waaaay to slow) so the more vram on your GPU the better. Nowadays, 12 gig min. You can use 8 for sure but it will be slower and generating large images(upscaling) may result in out of mem errors.
@@GrantLylick thank you for sharing that.. 👍
@@GrantLylick kindly suggest GPUs that are value for money having 12gigs, just researching for now.
Thanks in advance.
@@haripopstar whatever ur budget allows for 12 or more gigs of ram, there are lots of reviews out there for mid-tier cards. Nvidia may be slightly faster.
I'm new to the node based and really don't know a whole lot. but after watching this video and downloading the studio it was very easy to follow and do, everything seems to be working correctly as you showed in your video. I am 100% satisfied with the studio you did a lot of work to it and you did a very good job
Good looking and very useful for actual designers! This is exactly the kind of stuff Patreon was made for. Reward a talented workflow designer for his time and effort in putting together this great set of workflows.
great stuff, worth the $ just for the script. I ported it to linux, and use my own ComfyUI. Can't wait for more.
This mustve been A LOT of work!! very nice job
Thanks for all the work ya'll have done!
First of all….thank you for putting in this much work to create a ComfyUI user interface that is a lot easier to handle! But to my understanding, the interface along with everything else you have put in, works only on a machine with an Nvidia card (2060) for now? Hope you can create ComfyUI studio for us Mac owners who would like to try the same. 😉 Kudos for your work!
Thank you so much for your kind words! Yes, the current version of ComfyUI is optimized for machines with Nvidia graphics cards, such as the 2060 or higher, for faster image generation. A Mac version is possible, but right now, we are focusing on advancing the Windows version to higher levels. Image generation on Mac graphics cards would be slower, so for serious image creation, an Nvidia card is recommended.
@@ComfyUIstudio Thanks for the update on your work! I fully understand Nvidia as a first choice to bring the new interface out in the wild ;) I might be considering a Windows machine in the near future, just because of this! Thanks again for your reply and keep up the good work!
Hi, The app looks very useful. Am I able to take a comic character and adjust the posture? Thanks in advance.
Can you please show how to change SDXL nodes to Flux? I don't care about time of generations, I need the maximum quality. SDXL does not do it for me, unfortunately.
@@1Know1tHurts there is flux workflows in studio. No need to change
Genuine question, what kind of advantage comfy UI have if compared with FORGE? Can it generate faster with low ram?
I've had exactly the opposite experience. Using Forge with same model, Forge is faster. The "advantage" of Comfy is the possibility of much more complex workflows. Like generating an image in Flux, then upscaling in SDXL all in one go. The only reason I am using ComfyUI is it seems to get the newest functions faster than anything else. That's the benefit of node based systems, you don't have to reorganise the interface. Having said that, I use Fooocus and Forge more than ComfyUI, because I rarely experience errors. It's VERY easy to introduce errors in ComfyUI, because of the lack of structure.
Hi, did you have a Lora workflow in this ComfyUI Studio?
@@NutSuksamai Hi. Yes. Next update will have lora, controlnet, ip adapter, and other workflows. In 1 day. Stay tuned.
Very new to comfy ui but WOW fucking amazing man, I sub and downloading
Will this break my current installed comfyui ? I have previous installed comfyui (NOT portable one)
@@cgpixol No. It will not brake.
@@ComfyUIstudio Thanks for the reply. Would love to get a try. How can I buy this?
@@cgpixol Thank you. Follow the link in the description.
Hi, bro! Is it possible to use flux dev fp8 or in your studio I can use only NF4?
@@Lifejoy88 You can use the dev version if you have a workflow, but it significantly slows down your PC, making the time investment not worthwhile.
What, wait, when?? Just saw this. This is so cool.
how to install the comfy studio? there is no apps installation in the link.
@@firdausferd no need to install. Just download, run the model download script. And its ready to run.
I would like you to add ControlNet for Flux NF4 as well as other checkpoints SD1.5, SDXL and video workflows.
@@chaleesae1322 Thank you. ControlNet is on the rise, followed by the video workflows. Stay tuned
Are you bound to the square image format, or can you go any aspect ratio? Also, will any prompt work, i.e. if I wanna make Sci-fi imagery.
@@Niwon-g3s Hy! You can make any resolution. Any style.
@@ComfyUIstudio Great, thanks! 😄
This is amazing work! Question. Would I be able to use custom Loras and models from civitAI for both SD and Flux? Also is there a way to change the face expressions when working with any of the ID workflows?
@@pray4fabe Thank you for your feedback. Will make it as soon as possible.
Can work on GTX 1650 Max design ?
Unfortunately, ComfyUI might not run smoothly on a GTX 1650 due to its limited VRAM. You may experience slow performance or crashes with larger models or more complex workflows.
Do you need to be conected to the internet for this to work?
no
very nice 👍
amazing..subscribed...is it safe to update All and/or update ComfyUI in Manager ?
@@henkedrax4998 Thank You! It's better not to update. ComfyUI is already updated in every pack. The nodes and the manager should not be updated.
@@ComfyUIstudio Opened Workflow "FluxNF4" and there is a note: Setup:
1. Update ComfyUI in the manager menu.
Hello…this studio works with osx? M3 MacBook? Thanks
@@mitocondrianos no. Because it will be slow. Only Nvidia 2060 or higher cards are the optimal choice.
@@ComfyUIstudio hi! But…even the speed…exist any way to install it on my Mac?
@@mitocondrianos yes there are videos on UA-cam. Then you must ask chat gpt how to run scripts on mac.
@@ComfyUIstudio ok..thanks…I will be part of your patrons soon…thanks
Another question…can you create discount promo images (like an “50% discount in any order before Monday” with a product image background
) for an e-commerce? Thanks!
Im in your patreon, downloaded the comfyUI studio but i had error on checkpoint, do you know how can help me? i have to click on download all models or its already in my folder?
@@victorandrade2160 The checkpoints have not been downloaded. They need to be downloaded using the download all script. Sometimes the checkpoints don't fully download due to network errors. Therefore, the faulty checkpoint must be deleted and downloaded again.
@@ComfyUIstudio Thank you for fast response! I am doing it again, i noticed there are 2 checkpoints, flux nf14 and realistic vision, i dont understand very well, i must use realisticvision40 right? if i do, is it "connected" with flux features?
@@victorandrade2160 Every workflow is set to the checkpoint it works with. You don't need to change anything except for the prompts.
@@ComfyUIstudio i cant make it works, the following error appeared: !!! Exception during processing !!! Error while deserializing header: MetadataIncompleteBuffer
Traceback (most recent call last):
File "C:\Flux\App\ComfyUI\execution.py", line 323, in execute
output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Flux\App\ComfyUI\execution.py", line 198, in get_output_data
return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Flux\App\ComfyUI\execution.py", line 169, in _map_node_over_list
process_inputs(input_dict, i)
File "C:\Flux\App\ComfyUI\execution.py", line 158, in process_inputs
results.append(getattr(obj, func)(**inputs))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Flux\App\ComfyUI
odes.py", line 539, in load_checkpoint
out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=folder_paths.get_folder_paths("embeddings"))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Flux\App\ComfyUI\comfy\sd.py", line 524, in load_checkpoint_guess_config
sd = comfy.utils.load_torch_file(ckpt_path)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Flux\App\ComfyUI\comfy\utils.py", line 34, in load_torch_file
sd = safetensors.torch.load_file(ckpt, device=device.type)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Flux\App\python_embeded\Lib\site-packages\safetensors\torch.py", line 311, in load_file
with safe_open(filename, framework="pt", device=device) as f:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer
Prompt executed in 0.12 seconds
There is the flux schnell model workflow?
@@Aristocle No, because it's very slow.
@@Aristocle The NF4 model can generate anything in seconds, while the dev version takes minutes.
patreon is keep suspending my account. is there any other way to buy this?
Thanks :)
Tell me how to run this software on Linux?
install ComfyUI and insightface. Copy the models, workflows. Done
Does anyone know if it's got reactor or group video face swap?
@@WallyMahar it got reactor faceswap, and reactor faceswap 2x for two faces.
Does this also work with an amd card?
@@rockefella4life Hi. No.
@@ComfyUIstudio thx for your reply :)
Hi, if I subscribed to your Patreon, it's ready to download or I have to wait for the release? because I want to use it as soon as possible
@@Omarippo Hy! Ready to download.
@@ComfyUIstudio thank You 👍
Anything for MAC silicon?
Nice! Is the fee for this one-time or monthly?
@@zdziwslawy You can purchase it for a one-time fee, and you'll receive updates for one month.
@@ComfyUIstudio Thank you! it's great, I'll buy it soon :)
if only these studio have specialized workflow for low Vram, like lcm loras, etc... i guess that will be add more viewers and users since many have low Vram
All workflows are designed for low VRAM, which is why I don't use the Flux Dev model.
@@ComfyUIstudio good to know thanks, oh now i see it's nf4 flux
I’m already a comfyui user. Kudos for putting this together, it looks very good. Will you be replacing NF4 with GGUF in future versions?
@@DavidDoyleOutdoors Thank you. Yes gguf will be availible too.
AWsome but how about matrix?
i cant login to your patreon anymore says deactivated account
@@A-Eye--A please write to patreon.com help center
@@ComfyUIstudio can you cancel my patreon so i dont get charged anymore because they said they found on my other account i did not follow the guide line. im the one who you ask me to goto your facebook page
Clicking the link on my sandbox windows installation... Just in case.
Is it possible to open it on vastai
@@ArsalanVajed yes
Is it Mac compatible?
@@AshNair26 no. Laptop's can't handle the workload.
I have a 12th generation Intel core i7 pc with 16gb of RAM. And a Geforce RTX 3060 graphics card with 6gb of dedicated RAM. It will work or it is insufficient in terms of capacity.
It will work fine.
Seems you must be a Patron in order to down load. I don't see any 'free' or 'trial' options.
@@TheColonelJJ there's a free pack called Epic pack. Try it out. But it's just workflows. The studio contains the ComfyUI app. Preinstalled workflows and model download script.
Does this work on MacOS?
Only with good NVIDIA video cards, like the RTX 3060, because it will be very slow otherwise.
@@ComfyUIstudio a m2,m3 or even a m4could work fine i thinkk
@@RichardPinewood it does. It can't hold a candle to my 3090 at home but for light work? Totally fine. (M2 MBP 16GB ram).
Ofc it wont vs a 3090 but i have a m2 pro macbook pro with 16gb of ram and i use forge rn (not for flux) and Draw Things application (for flux) and it works great especially if you use 4 step lora for flux dev it takes a minute or just under per gen ill take that speeds on a 16gb shared memory silicon chip any day @@PhillipMarquez
@@PhillipMarquezI am also planning to get a used 3090 for £580 GBP refurb to replace my 6600 xt for stable diffusion
Can we make a LoRa with This?
@@chamathzoysa You can use LoRA, but you cannot create it. However, it has Instant ID and IP Adapter to maintain a consistent character.
where is the the mac version you said it was compatible?? uh?
Hy. Comfy UI works well only with Nvidia videocards.
I was wondering the same thing as well but his response always seems to be the same, so I'm guessing the answer is "there is no Mac version" since he's convinced ComfyUI doesn't run well (IMO it's totally acceptable, even with FLUX.1 model).
Sweet!! 😊
58gb? what can it do. deff vversion?
This is a 1-click install version with 20+ workflows, including models. Just like shown in the video, it includes Flux (11GB), RealVis (7GB), and several smaller models for ControlNet and other features. Plus, we've just added a new workflow that’s not available anywhere else-the two-person Instant ID face-based generation. Check out the latest video on our channel to see it in action!
@@ComfyUIstudio i only have rtx 4060 8gb vram. so if its a deff, i wont be able to use it
@@riered9reda101 it will work fine. Because its the NF4 modell.
@@ComfyUIstudio thankyou. my net is slow here. n ive just download about 30gb of comfy flux. n still a bit confused about what i can do with this.
@@riered9reda10158 gb is the size of the installation, not the vram required
minimum system requirements please. Thanks
if you have to ask, you dont meet the minimun system requirements... Buy a RTX 3060 or +
@@RafaelSequera I have a rtx 3060 and 16gb RAM..is that enough? thanks
@@giuseppedaizzole7025 Yes. But you need to run the studio off a SSD with a good IO. Samsung Pro or similar. It is better to keep it portable.
Is there any workflow + module for image to video generation?
There will be a workflow for image-to-video generation soon! I need to test it first before sharing, as some nodes could potentially cause issues with ComfyUI if not set up correctly. Stay tuned for updates!
waiting for an exe file to install 😊
No need to install. Just launch and create.
Is this censored?
You run it locally. You can use whatever models you want, including uncensored ones.
is it nsfw?
@@IkeikeIkeike77 yes
yeah thats the best organised noodles spagethi interface maybe... you realy have to love them... even if comfy is realy extremly powerful, i think it s unusable for teams... because each time you create something... the noodles are entangled in another way..
dont get me wrong... comfy is powerful and i like it.. but it doesnt work intuitive enough. even to use tthe word intuitive in one sentence with comfy... doesnt work..
whats your facebook link i lost it
@@A-Eye--A facebook.com/comfyuistudio
I sent you a message to your Gmail please reply anyway I like your workflows you did a lot of work thank you