Do you have an old computer or no NVIDIA graphics card? I also have a tutorial where I show how to use FLUX AI in the cloud with Google Colab for free (with English subtitles): th-cam.com/video/62wytdlrvBE/w-d-xo.html
I'm in the same boat, I tried to install on my little laptop and I started getting RuntimeErrors in line 11 saying my device does not support the current version of Torch/CUDA! Consider download another version". Then line 12 shows the ...webui-forge/releases.... I assume I need to try the Google Colab as well. My question is though, will it still be private and have unrestricted, uncensored functionality?
thank you! Finally someone who explains this topic in the simplest and most efficient way without talking nonsense.We are waiting for more tutorial videos.
I've thought about creating videos like this where it's no bullshit and very clear highlighting all the steps in a straightforward manner. Excellent video man, I'll have to check out your other videos, you just earned a subscriber.
Extremely helpful!! I'm up and running now! From first clicking on your video to generating an image within an hour or two (forgot to look - surprisingly fast) Thanks very much! Subscribed. I'll check your listings for other vids on the subject. ; )
Thank you! Not only did you easily show how to install Flux into Stable Diffusion, but I didn't know Stability Matrix even existed, much less being a nice interface, bringing the best AI UI programs together. Looks like I'll be purging some programs I already have and using them in Stability Matrix.
If you haven't downloaded stability matrix as portable then there will be no data file, instead just go straight into the models folder: e.g Stability Matrix\Models\VAE
Hey, I feel kinda dumb but I download stability metrix as you mentiont but it did not install and didn't make a single folders as you have in video. Also if I tried to install any model it massaged to me download failed. Do anyone know what I messed up? Thanks.
Basically just a somewhat recent (last 5 years or so?) pc, with a rather beefy GPU. In the case of the GPU, don't go for anything 8gb or less. You'll run into a memory bottleneck sooner rather than later, even if you've got a fast GPU. Mine's a rtx3070, fast enough to run almost any game on max settings (at 1080p at least), but the speed doesn't translate into running AI's because of all the memory swapping.
Hi, i have followed your clear instructions, but when starting, i get this error : "Connection errored out.". I have changed web browser but there's still this issue. Does anyone have a solution? Thanks!
is this the dev model or schnell youre running? im having issues with img2img im trying to train img model locally but cant really find much information except for links to replicate and using it in hugging face
Thankyou for creating this video, it really helped. To understand what kind of power it takes to generate these images, what is the graphics card that you use?
Sure 😁 Create a colorful and whimsical cover for a children's book. The main theme of the book is about a playful and curious cat who goes on adventures in a magical forest. The cat is fluffy with bright green eyes and a striped orange and white coat. The background should show a vibrant forest filled with flowers, butterflies, and sparkling lights to give it a magical feel. The title of the book, written in large, friendly font, should be "Cathies Adventures in the Magical Forest". Include the author's name, "Dommi Bessle," at the bottom of the cover in a smaller, elegant font. The overall style should be cheerful and engaging, appealing to young children. A captivating contemporary romance book cover presents a dashing blonde man in a sleek blue suit and bold red tie, standing proudly beside a stunning woman with long red hair. The elegant woman dons a stylish green knee-length dress and black heels embellished with malachite stones. The nightclub and city street backdrop create an alluring atmosphere. The title "City of Love" stands out in pretty script, with author Dominic Beauchamp’s name in bold letters below, promising a thrilling and entertaining read. Create an image of a crazy professor standing in front of a large blackboard. The professor should have wild, messy hair, glasses, and a lab coat, looking eccentric and excited. He is writing complex mathematical formulas on the blackboard. In large letters on the blackboard, the phrase "The Crazy Professor" should be written. In the foreground, a fluffy cat with bright green eyes is sitting on the teacher's desk, watching the professor curiously. The scene should be set in a cluttered classroom, with books, papers, and scientific equipment scattered around.
thanks for the guide ... it didn't work for me on Apple M3 Pro I am getting an error a minute after I press "Generate": "RuntimeError: linear(): input and weight.T shapes cannot be multiplied (961x64 and 1x98304). Time taken: 1 min. 1.9 sec." I didn't change any settings except on the top as described in the video.
I have a problem, the Stable DIffusion WebUI Forge shows me the error "Runtime Error: Your device does not support the current version of Torch/CUDA! Consider download another version: ". I tried installing the VS Build Tools with the msvc package but it still didn't work. Do you know a fix? Thank you (AMD Gpu)
I ran into the same thing, tried a bunch of fixes I found online but none worked. I finally just turned off the DreamShaper launch option to tell it to check the Torch/CUDA compatibility - and bang, it started working perfectly like a champ! I also have an AMD GPU, which works fine - so it is not dependent on having overpriced NVIDIA GPU's to work.
I've been usieng forge for a few months and prefer it over A1111. I decided to get flux dev and nf4 along with the very and text encoders. When I get home ill past the whole error log but I keep getting the error "Expecting value: line 1 column 1 (char 0)" the dev model sometimes even crashes my pc, the nf4 never crashes but it always gives me that error when I try and make anything. I updated my pytorch and forge web ui but I still can't do anything with flux. Other SD models work perfectly its only flux and yes I have it set to the flux
It also works with the mix of the both files, I also used it in my tutorial. Access to the autoencoder file at FLUX.1 [dev] is restricted and you can't download it just like that? 🤔
My understanding is except the pro, you can't use the other models for commercial use, so if it is true then the others are a little bit useless except to play with!?
The dev model is licensed under flux-1-dev-non-commercial-license. You can only use it for non-commercial purposes. The Schnell model is licensed under Apache 2.0 and on the official page of Black Forest Labs on Hugging Face it says: “Released under the apache-2.0 license, the model can be used for personal, scientific, and commercial purposes.”. But the Schnell model is just not that good 🤷♂
@@jaouadbentaguena8340 That's a good point 👍 Forge is also apparently very easy to install and is therefore suitable for beginners. But StabilityMatrix is a good program in itself for managing models, loras, etc., but if you don't need it, you can just use Forge directly.
@@DominikBoesslthank you I am going for the stability matrix😅. Is there a way to also use the trained Lora. I’ve been training Loras on replicate Flux, but I don’t know how to use it in stability matrix
@@jaouadbentaguena8340 There is a folder „Lora“ in the data folder of StabilityMatrix. You can place your lora in this folder and then you can use it with your prompt with “”.
I AM GETTING THIS ERROR EVERYTIME D:\fLUX\Data\Packages\Stable Diffusion WebUI Forge'... error: RPC failed; curl 18 transfer closed with outstanding read data remaining error: 3635 bytes of body are still expected fetch-pack: unexpected disconnect while reading sideband packet fatal: early EOF fatal: fetch-pack: invalid index-pack output
1. Checkpoints are more than a "rip off", maybe don't display your obvious lack of knowledge for all the world to see 2. Flux Schnell is free, why would someone "rip it off"
![FLUX AI: Lokal installieren, kostenfrei nutzen (ohne ComfyUI) - Einfaches Tutorial zu FLUX.1 [dev]](http://i.ytimg.com/vi/kKf0ljUKxEM/mqdefault.jpg)
Do you have an old computer or no NVIDIA graphics card? I also have a tutorial where I show how to use FLUX AI in the cloud with Google Colab for free (with English subtitles): th-cam.com/video/62wytdlrvBE/w-d-xo.html
This is awesome tutorial thank you so much
I'm in the same boat, I tried to install on my little laptop and I started getting RuntimeErrors in line 11 saying my device does not support the current version of Torch/CUDA! Consider download another version". Then line 12 shows the ...webui-forge/releases.... I assume I need to try the Google Colab as well. My question is though, will it still be private and have unrestricted, uncensored functionality?
Hi Domique, GREAT tutorial, thank you mate. Can you do one for midjourney too please?
@@darrenlewis8273 Thank you ✌ Nope, sorry, I don't use midjourney.
@@DominikBoessl commercial use ?
Simple. Straightforward. Direct to the point. Shows step by step. Easy to understand. Clear English. Well presented. Perfect!
Thank you ✌ Yes, clear English thanks to AI 😜
@@DominikBoessl 🤗🤗
I disagree, I never saw he install stable matrix.. is that warning from Windows defender normal?
@@AndrewTSq Yes, the warning is normal.
@@AndrewTSq It's safe, I would send it
thank you! Finally someone who explains this topic in the simplest and most efficient way without talking nonsense.We are waiting for more tutorial videos.
I've thought about creating videos like this where it's no bullshit and very clear highlighting all the steps in a straightforward manner. Excellent video man, I'll have to check out your other videos, you just earned a subscriber.
Also, do you think you could do a follow-up tutorial for Loras? Thanks!
Thank you! This was sooooo fast and clear tutorial, was easy to jump to needed parts. Love it, this kind of tutorials are best what TH-cam have
Hehe, thank you ✌😁
i'm an actual caveman and even i could follow along
Extremely helpful!! I'm up and running now! From first clicking on your video to generating an image within an hour or two (forgot to look - surprisingly fast)
Thanks very much! Subscribed. I'll check your listings for other vids on the subject. ; )
Got an error installing Stability Matrix, but it seems to be working anyway. Great guide!
Thank you! Not only did you easily show how to install Flux into Stable Diffusion, but I didn't know Stability Matrix even existed, much less being a nice interface, bringing the best AI UI programs together. Looks like I'll be purging some programs I already have and using them in Stability Matrix.
I can't find the "data" file in stability matrix. Can you help?
If you haven't downloaded stability matrix as portable then there will be no data file, instead just go straight into the models folder: e.g Stability Matrix\Models\VAE
Hey, I feel kinda dumb but I download stability metrix as you mentiont but it did not install and didn't make a single folders as you have in video. Also if I tried to install any model it massaged to me download failed. Do anyone know what I messed up? Thanks.
concise and straight to the point, thanks
Works like a charm, tyvm!
Thank you for this vidéo, very helpful ! (hello from France)
Thank you for making it in English , beautiful video , Thanks for sharing.
You're welcome ✌😁
Super easy, great video, very precise! Thank you!
what's the best stable diffusion matrix / interface or forge? please
the stable diffusion webui forge keeps downloading and has no end to its size, is something wrong with mine?
Take my like, you deserved it
Good stuff man, thanks! Nice and simple, you should make more english content too.
I followed every step, but after entering a prompt I just get "TypeError: 'NoneType' object is not iterable". I am using a GTX4080Super.
how do i run it as a portable folder?
Do i need to install another software to unzip the file?
When you install it, you can choose to install it as portable.
@DominikBoessl I found out my computer was blocking me from making a portable folder due to security reasons but i just unblocked it and it made one
Flawless tut, thank you :)
Can you use this to create videos? Otherwise what download is best for highest quality free videos?
Can you make a tutorial of how-to train a flux model?
Awesome tutorial, thanks! I am now going down the rabbit hole of AI image creation and may never return! Please send cookies.
What kind of pc configuration so does it need to get started?
Basically just a somewhat recent (last 5 years or so?) pc, with a rather beefy GPU. In the case of the GPU, don't go for anything 8gb or less. You'll run into a memory bottleneck sooner rather than later, even if you've got a fast GPU. Mine's a rtx3070, fast enough to run almost any game on max settings (at 1080p at least), but the speed doesn't translate into running AI's because of all the memory swapping.
Thank you!! Great Video, 10/10!!
Hi, i have followed your clear instructions, but when starting, i get this error : "Connection errored out.". I have changed web browser but there's still this issue. Does anyone have a solution? Thanks!
Thank you for this detailed tutorial! Is there a way to remove the safety filter?
is this the dev model or schnell youre running? im having issues with img2img im trying to train img model locally but cant really find much information except for links to replicate and using it in hugging face
thank you very much, that was so easy to follow. im up and running now!
thank you
after i installed it can i use it via APi?
Thankyou for creating this video, it really helped. To understand what kind of power it takes to generate these images, what is the graphics card that you use?
I use a RTX 3060 6GB Mobile, so kind of low end. More power is always better 😁
Thank you for the tutorial!
can you give us the prompts you used to procure those images?
Sure 😁
Create a colorful and whimsical cover for a children's book. The main theme of the book is about a playful and curious cat who goes on adventures in a magical forest. The cat is fluffy with bright green eyes and a striped orange and white coat. The background should show a vibrant forest filled with flowers, butterflies, and sparkling lights to give it a magical feel. The title of the book, written in large, friendly font, should be "Cathies Adventures in the Magical Forest". Include the author's name, "Dommi Bessle," at the bottom of the cover in a smaller, elegant font. The overall style should be cheerful and engaging, appealing to young children.
A captivating contemporary romance book cover presents a dashing blonde man in a sleek blue suit and bold red tie, standing proudly beside a stunning woman with long red hair. The elegant woman dons a stylish green knee-length dress and black heels embellished with malachite stones. The nightclub and city street backdrop create an alluring atmosphere. The title "City of Love" stands out in pretty script, with author Dominic Beauchamp’s name in bold letters below, promising a thrilling and entertaining read.
Create an image of a crazy professor standing in front of a large blackboard. The professor should have wild, messy hair, glasses, and a lab coat, looking eccentric and excited. He is writing complex mathematical formulas on the blackboard. In large letters on the blackboard, the phrase "The Crazy Professor" should be written. In the foreground, a fluffy cat with bright green eyes is sitting on the teacher's desk, watching the professor curiously. The scene should be set in a cluttered classroom, with books, papers, and scientific equipment scattered around.
@@DominikBoessl I am really thankful for you!
@@AbdulKhamis No problem ✌
thanks for the guide ... it didn't work for me on Apple M3 Pro
I am getting an error a minute after I press "Generate":
"RuntimeError: linear(): input and weight.T shapes cannot be multiplied (961x64 and 1x98304). Time taken: 1 min. 1.9 sec."
I didn't change any settings except on the top as described in the video.
I'm getting a similar issue, M4 Pro chip
Same problem with MacBook Pro M1
Simple Direct to the point
There is no data file when i downloaded.....
Where can I find a tutorial like this but for amd?
I am AMD user, do you have a tutorial for that?
Exactly the same, just use OpenCL instead of CUDA
Thank you this was exactly what I was looking for
hi, my laptop have GTX 1070 8GB and 32 RAM, will i have any problem running FLUX?
it will run fine
This video def helped me as well. ♥️
Is it possible to host in cloud hosting free tiers like AWS and use the model via API?
Connection errored out.
I have a problem, the Stable DIffusion WebUI Forge shows me the error "Runtime Error: Your device does not support the current version of Torch/CUDA! Consider download another version: ". I tried installing the VS Build Tools with the msvc package but it still didn't work. Do you know a fix? Thank you (AMD Gpu)
It only supports NVIDIA
@@rulisesmendoza2936 No way to get Flux1 to work on AMD?
I ran into the same thing, tried a bunch of fixes I found online but none worked. I finally just turned off the DreamShaper launch option to tell it to check the Torch/CUDA compatibility - and bang, it started working perfectly like a champ! I also have an AMD GPU, which works fine - so it is not dependent on having overpriced NVIDIA GPU's to work.
@@warp00009 how to turn off dream shaper. can you give me a steps?
@@warp00009 yeah, I've the same issue, how did you turn off the dream shaper?
When I unzip its just StabilityMatrix, the data folder is not there
After unzipping, install it.
Ehrenmann :)
Thanks for this tutorial.
I've been usieng forge for a few months and prefer it over A1111. I decided to get flux dev and nf4 along with the very and text encoders. When I get home ill past the whole error log but I keep getting the error
"Expecting value: line 1 column 1 (char 0)"
the dev model sometimes even crashes my pc, the nf4 never crashes but it always gives me that error when I try and make anything. I updated my pytorch and forge web ui but I still can't do anything with flux. Other SD models work perfectly its only flux and yes I have it set to the flux
The two files you have links two are for a dev file and aschnell. Won't that cause errors
It also works with the mix of the both files, I also used it in my tutorial.
Access to the autoencoder file at FLUX.1 [dev] is restricted and you can't download it just like that? 🤔
Good tutorial. Thanks!
4-5 mins on which gpu?
NVIDIA GeForce RTX 3060 Mobile 6 GB VRAM
I'm running Windows 10 with a Radeon RX550 graphics card with 4 Gigs on it. Do you think I would be able to run that without a problem ?
Nope, it would not run. But check out my video running Flux via Google Colab: th-cam.com/video/62wytdlrvBE/w-d-xo.html
Commercial use ?
See the license file on Hugging Face: flux-1-dev-non-commercial-license
@DominikBoessl ok thanks, but i have now purchased the freepik 1 year plan which also includes this flux realism model
@@okaythanks-sn5ev As far as I know this uses Flux Pro which includes commercial use. If you are not sure, just ask the Freepik support.
How to use the negative prompt box though?
Just enter whatever you DON'T want to see
My understanding is except the pro, you can't use the other models for commercial use, so if it is true then the others are a little bit useless except to play with!?
The dev model is licensed under flux-1-dev-non-commercial-license. You can only use it for non-commercial purposes. The Schnell model is licensed under Apache 2.0 and on the official page of Black Forest Labs on Hugging Face it says: “Released under the apache-2.0 license, the model can be used for personal, scientific, and commercial purposes.”. But the Schnell model is just not that good 🤷♂
@@DominikBoessl Thank you for your answer. So Pro & Schnell are ok for commercial purposes but Schnell is crap, ok I understand now.
how can one install it as a portable version?
You will be asked for this during installation.
Does this work with RADEON cards?
Yes, use OpenCL instead of CUDA
very helpful! please how to do image2image with Flux AI in WebUI Forge?
There is a tab "img2img", where the magic happens 😁 See other tutorials like: th-cam.com/video/uTFdl5s6Vb0/w-d-xo.html
@@DominikBoesslso is it better to directly install forge directly?
@@jaouadbentaguena8340 That's a good point 👍 Forge is also apparently very easy to install and is therefore suitable for beginners. But StabilityMatrix is a good program in itself for managing models, loras, etc., but if you don't need it, you can just use Forge directly.
@@DominikBoesslthank you I am going for the stability matrix😅. Is there a way to also use the trained Lora. I’ve been training Loras on replicate Flux, but I don’t know how to use it in stability matrix
@@jaouadbentaguena8340 There is a folder „Lora“ in the data folder of StabilityMatrix. You can place your lora in this folder and then you can use it with your prompt with “”.
thanks very much ♡
This is awesome tutorial thank you so much
is it comfyui is more speed then this method
I think the command line usage as described on the official github page is still easier.... 😅
I AM GETTING THIS ERROR EVERYTIME
D:\fLUX\Data\Packages\Stable Diffusion WebUI Forge'...
error: RPC failed; curl 18 transfer closed with outstanding read data remaining
error: 3635 bytes of body are still expected
fetch-pack: unexpected disconnect while reading sideband packet
fatal: early EOF
fatal: fetch-pack: invalid index-pack output
Does this work on AMD too?
Nope, not really. But check out my video on running Flux on Google Colab: th-cam.com/video/62wytdlrvBE/w-d-xo.html (with English subtitles)
Does an RTX 3060 even have a chance?
Sure, I have a RTX 3060 Mobile
I got it to work, but following your good instructions it was difficult.
Can it run on 4060 8 gb?
Sure 👌
thank you
Is flux good at face swap?
Not what it's for, don't even think it can. Use roop unleashed or something.
instal failure with stabilityMatrix :(
Matrix is shit, good idea; execution shite. It seems that whatever you do it always fails to find, download or install something.
Oh right, an alternative would be useful. Install Forge, run it on there.
it is only one exe file and no folders
They appear at first run ;)
Is this really using the actual FLUX, isn't it just using a checkpoint, so a rip-off of the original program?
1. Checkpoints are more than a "rip off", maybe don't display your obvious lack of knowledge for all the world to see
2. Flux Schnell is free, why would someone "rip it off"
Installing as a portable option is not a choice, if you want to follow this tutorial.
For my tutorial I installed it as portable version.
Oh good, stability matrix. In case it wasn't hard enough, here's Matrix with it's millions of errors...
Download Forge, install Flux in there instead of this minefield. You'll thank me.
with all ma luve to you .... ♥
of course its not cherry picking... if zou are acctualz interested in this, yout pictures from SD 1.5 looks like 50times better than this...
First of it depends on the prompt, second of sd hardly ever creates what you want.
@@ipellaers no, prompt is just one part of settings... You need much more to make good picture
@PetrBelohoubek-ot5ok hmm you don't say? Oh wise master grant us your entry level wisdom 😂
@@ipellaers i make living with it. And you?
@PetrBelohoubek-ot5ok Sure you do buddy!