Would you have the scope to catch up in discord voice chat? Just wanted to have some intelligent conversation, since you are more experienced in AI tooks
Thanks for the guide, but when I downloaded it, the 'run bate ' option is not visible. Instead, only the 'Run' option is visible. After clicking on it, the file leads to a similar page, but an error 'connection timed out' is showing. I got a GTX 1650
This was a very helpful tutorial. In particular, I really liked the detailed explanations you did for the various image prompt options. I'd love a guide on more advanced usage, like complex prompt writing with weighting and multi-prompt, and examples of how to use Styles and LoRAs in combination to make a great image. Thanks again for the tutorial!
Oh, that is nice feedback. Thanks a lot, Jake. I am happy to make such thing as a follow-up tutorial. But please know, every style for example is 'nothing more' as written prompts predefined to reach a certain look without the user finding it out himself. Also, Fooocus is already using similar to Midjourney a GPT2 conversion. Means, no need for complex prompt engineering. Otherwise, weights work the same as in Stable Diffusion. Imagine you want to see a space soldier and a spaceship, if your results are ignoring the spaceship, give it a weight "(spaceship visible:2)" for example.
Hey @@Axel.Peters Please do make a tutorial on how to use Styles and Loras in combination as I am a complete beginner. And if you could show a way to fine-tune the faceswap feature
@@Axel.Peters Oh, that's interesting. So do you rely on the styles or do you try to accomplish it all in the actual written prompt. I imagine using both is a lot of trial and error.... is there any way *we can* find out ourselves how the style is actually predefined? Right now I am trying to wrap my head around best practice when it comes to styles, refiners and loras and how they play in conjunction with each other.
Thank you for the tutorial. My PC almost caught on fire from the exces stress I put on it when I diden't need to.Thanks to you I finnaly how to optimise the program and make full use of the options. THANK YOU
Appreciate your kind words I hope you could take something out of it, and if you have ideas or interest in other tutorials, just let me know, or check my channel.
Thanks for the tutorial it was helpful. Some of the shake screen transitions during editing were a bit jarring and made it harder to watch but overall it was made well.
Thanks for sharing this tutorial, I was tried to use the new Defooocus to change clothes using both image prompt & inpaint function but I never able to matching over 70% on some patterns or any patterns like patterns on T-shirts or any printed patterns. Would you make a video on how to mach a pattern as close as possible. Thanks in advance.😊
I've been using Foocus for some time and it's very good but the native interface (very small black text on a white screen) is niaive, poorly laid out and VERY difficult to use! What display interface are you using. It seems to be exactly what I've been looking for!
Hey there Nice informative video! Any tips to get ‘better’ results with face swapping as im trying to have a consistent model and whenever i try to get the desired face in different scenarios it always looks very fake and like the face does not fit on the body
Hello, it really should! Do you really waited, long enough? The file is rather big, so depending on your Internet connection, it might take a while. Can you check again?
Hello, if you have FOOOCUS installed on your D drive, the models and generated images are located as well on the D drive. However, this doesn’t alter your standard Windows configuration. Typically, paging files are located on the C drive. This means if your VRAM or system RAM reaches capacity, the system will utilize paging files, which are likely on the C drive where Windows is installed. After a fresh restart, do you notice an improvement in the available space on your C drive? If so, my assumption might be right. You could change your paging files to another drive via the Windows system settings.
Hello, it has both already available. AMD support as well as low system requirements. All that needs to happen for the low system is an adaptation to the original run.bat file. Let me know how f you need help.
@@Axel.Peters i actually tried using fooocus on my brothers pc and work pc as well.a my brother has amd and i received some error messages saying that I need nvidia. his pc is pretty powerfull i might say. yet, on the work pc, i see nvidia when i right click on the desktop, but i am not sure what graphics card it has. maybe the drivers are installed incorrectly or something like that. it is a weaker pc than that of my brothers and even mine. in every cases it shows various errors, mostly saying that I need nvidia graphics card. at work, i opened foocus and it just says error. i cannot generate something. i even tried google collab, but i an new to it and i might used it wrongly. it says nvidia graphics card error as well. what a shame! I need to make this fooocus work. i find it very lucrative for some projects that i have envisioned. i would love some help from someone who knows more programming than i do. how did you made it work?
I understand you’re curious about whether you can use Fooocus AI with an AMD GPU and how to go about it. After looking into this, it seems that there are significant compatibility issues when trying to run Fooocus on systems with AMD GPUs. This is primarily because Fooocus AI relies heavily on CUDA, a technology developed by NVIDIA, which isn’t supported on AMD hardware. One user on GitHub noted that despite installing torch from the AMD package index, Fooocus AI still attempted to use CUDA, leading to errors and incompatibility . Unfortunately, the Fooocus AI GitHub repository doesn’t offer specific guidance or solutions for AMD GPU users . Given these challenges, it appears that currently, there isn’t a straightforward method to run Fooocus AI effectively on an AMD GPU. If you’re keen on using Fooocus AI, you might need to do so on a system with an NVIDIA GPU, or consider other AI software that’s compatible with AMD hardware. I understand this might not be the news you were hoping for, but I hope it helps clarify the situation. If there are any updates or workarounds in the future, the Fooocus AI GitHub repository is a good place to keep an eye on.
Thanks for the video. How did you get all of the information about different settings? Just by playing with it? I'm trying to figure out how to (1) see what the different "styles" add to the prompt and (2) learn what the different image prompt settings do
Hello, (1) you can see the style prompts in the running terminal when rendering or in the history log file. (To open the log file, activate advanced bottom left and then you see history log file and a link for it) Not sure what you mean by your second point. Would you elaborate a bit further?
I downloaded the zip file, extracted and then downloaded the 6gb file later it also downloaded 222mb and the window just closed. in many videos i have seen an offline window opening on the web browser immediately after the 6gb file is downloaded. is there any way out of this or does my laptop not support fooocus?
The file you DoubleClick to install is the same file you use to start the program. So what is happening when you start it again, will it go back to download files or will it open Fooocus (on the browser)?
Not sure I follow entirely. But if you want to keep the person and generally speaking the position, then you could just do some sort of inpainting by selecting the face and create a prompt by describing a happy, sad or angry face, lips, etc.
Hello, what do you mean with error? Can you be a bit more precise? Error during installation, usage, image creation, missing models, etc. ? Where and in what process you have an error and what is the error message?
Can you please make a tutorial about negative prompting in fooocus? With adding weight to it. I keep struggling with a t-shirt WITHOUT text or logo. Same for a baseball cap. It seems fooocus doesn't know those even exist. 😁
For example: 1. Initial Image Generation: Begin by creating a front portrait of the person. This image serves as a reference for subsequent modifications. Utilize features like Extreme Speed in Fooocus AI for rapid iteration and selection of a preferred face. 2. Template and Image Prompts: Once you have the front portrait, use a template designed for generating portraits in multiple positions. This process involves using two image prompts: the original front portrait and a template that specifies positions (e.g., front, side, etc.). 3. Adapt Text Prompts: Along with the image prompts, modify your text prompts to indicate the desired positions or angles, like “multiple portraits in different angles.” 4. Image Splitting and Upscaling: After generating the composite image with multiple poses, split it into individual images. These can be upscaled (e.g., 1.5x) to enhance quality. 5. Faceswap Adjustments: You may further refine the images by using the Faceswap feature. Adjust the weight of the images to change the angles or perspectives. 6. Handling Pose Variations: For changing the pose direction (e.g., making the photo look in a different direction), consider using a combination of PyraCanny (for edge mapping) and Faceswap, along with appropriate prompts.
if i increase the aspect ratio / higher landscape resolution will the photos take longer to generate? cuz what im using rn takes 51 min for 10 images( 1:1 1024x1024 and other factors involved as well)
Hello. Thank you for this important tutorial video. It is clear how much time you spent in order to let us execute all your instructions precisely specially concerning all the note you wrote below. For me, I have an issue. After extracting the Focus Zip file , you can read ( run ) but (bat) is not showing and not able to run this file. Thank you in advance.
Hello, thanks for this great feedback. Concerning your question, I’m not sure if I really fully understand your problem. For me it sounds like you see the run file, but you are having the problem that you can’t see the file extension (.bat), correct? if so, it just means that windows is hiding the extensions for all your files, but this is just an easy system fix. Generally speaking, it doesn’t matter whether or not you can see the extensions in theory you would just have to double click the run file. Can you try double click the file?
I’m sorry, great video but I have a question. I’m running this on colab and everything is working but I would like to know why every time I run it looks like the memory “disk” is decreasing a lot. Why is that and how can I fix it, do you have any suggestion? I really can’t explain myself better but I’m referring to the thing in the left and down part of the screen where it says disk n/100gb
Hello, I can't help a lot with colab space. I assume you are limited on the cloud space available. When you say decreasing, I assume you refer to the available memory and if so, that is normal. The generation process takes quite some compute CPU/ GPU and memory power (VRAM / RAM) ;-)… That is the reason (one out of many others) why people use online services (Midjourney, Dall-E, etc.) that do all the compute work for them and no high-end hardware required. Anyhow, as long the decreasing memory it is not hindering you in your creative process, don't bother.
Hi , I've installed the package, but when I run it, it doesn't install the sd_xl, and when I run the program I cannot generate an immage it just says error. I tried to install them separetely but still does says error, can you help me?
OK, help me a bit here, what is cmd saying? Is it stopping the download or doesn't even start, does it take long and feels like there is no progress? Just be aware, the download of the checkpoint model takes quite a time since you download around 7 GB and depending on your internet connection quite a ride. You said, you manually downloaded the checkpoints, have you renamed them? Have you dropped them into the correct folder inside Fooocus? Have you once the web server (Fooocus frontend) started activated advanced on the bottom left and manually selected your downloaded checkpoints?
Hello! Yes, it should operate effectively on Colab Pro. The term “smoothly” is subjective, so experiences may vary. It’s unlikely that you’ll have access to an RTX 4090, which is a top-tier GPU, but you’ll certainly be able to generate images with ease.
Thank you @Axel.Peters . I dont have enough GPU and i am getting NVIDIA error. I will go with Collab Pro then. If that works. Thank you so much again. I am waiting for your next vids.
Hello, the same way you opened it the first time. One of the three bat files. Double click - done. The first time it downloads certain models and the second time it just opens since everything else is already in place. Hope this helps.
Hir Sir Just wanted To ask last night my pc got crashed shutdown while generating The Model. I have 12gb of ram installed with nvidia gtx 1650. And it takes too long to generate the model. Any help would be really appreciable ?
Hello, AI is rather hardware intensive and I fear not being able to help you in regard to the speed. You could try to reduce the steps to generate an image (use speed instead of quality in the advanced settings). You could decrease the output resolution and only when you are happy with your outcome you could start to upscale.
Hello bro can you please give me solution of my problem Now the google Collab is not giving me permission to use fooocus ai They are telling that i need to buy their subscription and getting error Please answer me 🥺
Hello! It sounds like you’re accessing Collaborative cloud resources rather than your own hardware. It’s reasonable for the service provider to request payment for their hardware offerings. If you prefer not to incur these costs, consider using your own equipment. For this, you’ll need a reliable computer equipped with a capable NVIDIA graphics card with sufficient VRAM. Please feel free to reach out if you need further assistance. Wishing you the very best!
Hello, 128 MB of VRAM is significantly less than what is common these days, where people usually have 4GB or more. This could be the reason for your problem. To adjust for this, you might have to alter the run.bat file that you use to start Fooocus. You have to add the command to use your CPU only, without overall GPU usage. The command is: -always-cpu Please note that running Fooocus AI in CPU-only mode can be considerably slower than using a GPU, so you should expect longer processing times.
I have 16GB of Ram and gtx 1660 with 6GB VRAM. It currently uses low vram mode and gets to around 5.1 GB Vram usage. Can I optimise by replacing low vram mode with medium vram mode?
Hello, you can surely do that. There’s no harm done figuring out that it might not work.. your system might unstable or unresponsive, but after a quick restart, it should work normal again. And if you noticed that it doesn’t work, change the command back to what it was before. But if you ask me, there is not much room for improvement knowing that you have used already 5.1 GB versus 6 GB you have available - there’s not much room… Also, think about that you might want to upscale your image afterwards or that you want to inpaint, all that might require more VRAM.
Thanks for the tutorial, it was really helpful. I only have a question. If I already have SDXL models downloaded and don't want to download the ones in the Installation. How could I skip this step?
Thanks, happy to hear you have found it useful. To answer your question, you have two options. (1) It will download default models to the folder "Fooocus\models\checkpoints" --> If you have them already, just copy them over. (2) If you have a folder elsewhere with all your checkpoint models, change the path inside the config.txt file with a simple text edit tool to your own folder. (config.txt can be found inside the main Fooocus folder)
I have been having fun with this prompt Character 1 (Spider-Man) is talking to Character 1, Character 2 (Deadpool) is listening to Character 1.. With interesting results. I have even been able to get up to Character 3. But that's really rare. Also if you prompt more than 2 characters it starts to blend them very funny. I want to know how to control the camera. As I am wanting to use this to start my Comic book Series. And so far NOTHING I have tried has given me consistent results.
That sounds like real fun! The camera you could control with Loras for example. Also, to control the camera angle and position in a prompt for image generation, you can specify these aspects explicitly. Here’s an example prompt structured to include these details: Prompt Example: Subject: Victorian-era street scene, bustling with people Camera Angle: Low angle view, capturing the grandeur of the architecture Position: From the perspective of a pedestrian on the street Lighting: Soft afternoon light, long shadows Modifiers: Vivid colors, high detail Breakdown: Subject: Clearly describe what the scene or subject is. Camera Angle: Specify the angle like low, high, bird’s-eye, etc. Position: Indicate the vantage point or perspective, such as pedestrian view, overhead, close-up. Lighting and Modifiers: Detail any specific lighting (e.g., golden hour, diffused light) and other modifiers like color scheme or focus. I hope that helps.
@@Axel.Peters Danke. I’m really liking this and once I think I got some good pages I’ll share some. But give me a bit as now that this tech is avail, I have a very good outline just need to flesh the script out. This is going to be fun.
@@Axel.Peters Question how can I make the Ship from Flight of the Navigator show up in my render and/or the robot from that movie.. ? Is that possible or do I need to Train that into a Loras?
@@Axel.Peters when i press generate a prompt, it appears next to the loading bar. Or it used to. It seems that it got fixed on its own? I dont know. Anyways thank you for the quick reply. I still have no idea what the problem was lol.
@@VTick712 the first time it takes a bit to load the checkpoint into the memory. That can take a while… glad you have solved it. Happy prompting, and use the styles as described in my tutorials. Simplifies your prompting…
Okay, I think it's installed but how do you start the program?? There's no icon on the desktop or and .exe file to be found. Can't navigate the "simple" screens if no one tells you how to access the program.
Hello, it’s the same bat file you used for installation. There is no exe… Start the bat file - and it will open if all is installed correctly. Have fun
@@Axel.Peters well, I have done just that multiple times and it only reinstalls and then nothing every time. Even using the IP address does not work. Says something like it doesn't exist. And so far, it is NOT fun.
Hello... Try to add one of them as text into the bat file (the batch file you use to start Fooocus)... in your case, I assume (3) or (4) would work best: 1. --highvram 2. --normalvram 3. --lowvram 4. --cpu 5. --gpu-only
Hello, I really don't know. You could try to start Fooocus with the 'lowvram' or 'cpu-only' command. Let me know if you are unclear about the commands, happy to help out.
@@ManniBOY Try to edit run.bat file (the one you normally doubleclick to start fooocus) adapt the first line: \python_embeded\python.exe -s Fooocus\entry_with_update.py You could add behind one of the following things: 1. --highvram 2. --normalvram 3. --lowvram 4. --cpu 5. --gpu-only In your case use -lowvram
Good morning, I wanted to ask once I turn off the computer. When I turn it back on, how do I launch Focus? Without having to reinstall it. Is it possible to create a shortcut on the desktop?"
Hello, you don’t install it over and over again. All you do you is to execute the bat file. It finds all the formerly installed components such as the checkpoint models, Lora’s, etc automatically… I hope this helps.
Hello, to start Fooocus, you’ll need to use the same .bat file that you initially used for its installation. This file is located in the main folder of the Fooocus program. Just as you did during the installation process, open and run this .bat file to get Fooocus up and running. If you need further guidance, please refer back to any instructional materials or videos provided during the installation. Have a great day!
@@user87546 for consistency you can only use 1 model at the time. About the Eyes no solution until now. I know some people that used some eyes Lora's. I'm using A1111 and with Adetailer all eyes are solved.
What checkpoint do you use when experiencing it? Standard in that case Juggernaut or the realistic.bat version? If not, try the realistic.bat version, it uses a great checkpoint for photorealistic things. I am having no clue what difficulties you are experiencing, but all I would do is inpainting the ear of the person and adapt the prompt to serve your needs. With your needs, I mean - better explanation of the earring or accessories you would love to see. Make sure you use the standard inpainting feature in Fooocus, not the one that adds or changes the background. In case you have deformed ears, use the Fooocus 'enhanced' style, it is a negative prompt collection that solves such problems very well.
Hello, no… I was speeding it up for the benefit of the viewer to focus on the things that matter rather than waiting for something to render. I have a RTX4090 with 24GB VRAM and with system RAM of 100GB. It takes around 30sec + to generate 4 images as a batch.
Take that link here and scroll down until you come to the MAC section. Installation instructions can be found there. Have fun --> github.com/lllyasviel/Fooocus
I have macOS 14 Sonoma & I folloewd the instructions from your link, everything installed & updated correctly, but I think I'm missing something because I can't seem to find the correct prompt to launch or open fooocus. I'm a pc person but right now only my macbook is available😣@@Axel.Peters
Hello. Not sure what are you considering good specs, but it should not take 10 min. I can only imagine that your VRAM is full, and therefore it slows down your Fooocus activities . You would have to check your VRAM usage during rendering via the task manager (performance). Furthermore, I would as well try to use Fooocus with a --medvram or --lowvram command. Enter it to the existing text inside your bat files.
Hello, please verify if this issue is related to your VRAM. When VRAM is full, the system tends to slow down and may become unresponsive. If this is the case, you should modify the Fooocus AI startup command to start with lower VRAM usage. Add the following command-line argument: --lowvram This command enables the low VRAM mode in Fooocus AI, which is designed to reduce VRAM usage and optimize performance on systems with limited VRAM. Please be aware that while this adjustment can help in running the application more efficiently, it might also impact performance.
I am not an expert in that area, but here are the licenses that are being used by Stable Diffusion: www.licenses.ai/blog/2022/8/26/bigscience-open-rail-m-license
Hi again.Thank you for your prompt reply on my comment. I did a double click on the file and you were right. Running the file is tacking time and suddenly stopped on 49 % from a few minutes ago. So can you please tell me what to do ? Thank you very much.
I’m am not a magician, but I assume the 49% is referring to the download of the checkpoint. It’s not related to the software as such, but for whatever reason, your download stopped, might be that your internet connection dropped… the file/ checkpoints that the program is going to download are rather big, around ~ 6GB… just try again, and ensure your connection stays stable. If that doesn’t work, I can provide you the link to the first file, and you can download it separately…
@@georgedeeb4483 thanks a lot. Let me know if that worked out. As said before, if not let me know, I can provide you otherwise the link for the checkpoint.
Hello and thank you very much for your great feedback. concerning the deinstallation is rather easy… it does not need to happen, all you need to do, is to delete the Fooocus folder. You have a portable version, that means everything that Fooocus requires is within the folder.
Is the connection error preventing you from using fooocus? I am asking because cmd sometimes shows that message but it still function correctly… might also be that you loose connection when your VRAM is full. You eventually need to start your run.bat with an extra command towards low VRAM availability or CPU usage only. If you don’t know the extra commands, check the other comments or get back to me, happy to provide them to you.
I understand you’re experiencing connection error messages with Fooocus AI on your NVIDIA 1660ti, which might be related to VRAM limitations. One way to address this is by modifying the run.bat file to use less VRAM. Specifically, you can try adding the --always-low-vram argument to your run.bat file. This command switches the program from normal VRAM mode to low VRAM mode, which could help run Fooocus more stably on your system. Here’s how you can modify the run.bat file: 1. Open the run.bat file in a text editor. 2. Add --always-low-vram at the end of the command line. 3. Save the file and execute it again. This adjustment should allow Fooocus to operate in a mode that’s more compatible with the VRAM capacity of your GTX 1660 Ti. However, keep in mind that running in low VRAM mode might impact the performance or speed of the application.
Great video, but it doesn't work for me for some reason. It stops at model_type EPS adm 2816. I downloaded 3 base models, and it still doesn't work. Do you have any idea? / GTX 1660 TI / 531.18 - 16GB RAM / WINDOWS 10
Hi, it might be related to your system specs. Try to edit run.bat file (the one you normally doubleclick to start fooocus) adapt the first line: \python_embeded\python.exe -s Fooocus\entry_with_update.py You could add behind one of the following things: 1. --highvram 2. --normalvram 3. --lowvram 4. --cpu 5. --gpu-only In your case, use lowvram. Should work.
@@Axel.Peters it didn't work . yes, i add it correctly. It says 'Set VRAM state to: LOW_VRAM' in CMD .i looked up on the Fooocus github page and found out that another person has the same issue as me.(they posted about it 14 hours ago). should I download the previous_old_xformers_env.7z version? It might work .
Hello, help me to help you 😀... What do you mean with your own images? I thought that my tutorial was exactly about generating your own images… But I assume that you mean something else, right? Are you possibly referring to your own face?
@@Axel.Peters Tysm peter for ur reply, actually i was talking about how we can generate our own face image in own style? Bcz i don't wanna use someone else's style, i don't wanna do that, remove someone else's face and put your own. That's my point brother.
Hello, yes you can. The download is exactly the same, but the run.bat file need to be adapted. You would have to open the bat file with a normal text edit tool and add some lines to it. It's easy, the text you have to add can be found here --> github.com/lllyasviel/Fooocus (Scroll down until you come to the section of AMD GPU's)
Hello. I'm just curious to know if anybody else is experiencing Fooocus running VERY slowly? I don't know if I did something but half the time it crashes and won't run.
Hello and good morning. I’m sorry to hear that It doesn’t work very well at your end. A slow and unresponsive program might come from not having enough VRAM available. May I ask what kind of graphic card + VRAM you have installed?
@@Axel.Peters Hello and thank you for taking time to reply. My Vram is: 6 GB My Graphic card is: Geforce GTX 1660 Until now I've only been able to successfully run Fooocus from a link in a video tutorial that launches it on my PC. Sadly, Fooocus only runs at night but tends to crash periodically until I reload the program. Any thoughts you may provide will be appreciated. This is the only AI program that really I really enjoy and best of all, it's free. I would hate to lose use of it. Thank you again.
I really believe the problem is related to your limited amount of VRAM. You should start fooocus with a different setting so that it doesn’t use all of your available VRAM. For managing VRAM usage, options like --lowvram or --normalvram, which can be added to the command line or batch file to adjust resource allocation.
@@Axel.Peters Hi. Thanks for replying. Today I installed a new Graphic Card with 12 GB of vram. I started to use Fooocus and is began issuing "error" warning running down the screen again. In all honest, due to my lack of experience with this software I have no idea how to begin trouble shooting. Any thoughts on where I can look to resolve this? Thanks again.
Hello, don’t give up. I thought I covered all cases somehow already in the video but if you followed it 1:1 I really don’t know what else it could be. Couple of questions though… 1. Did you used a NVIDIA card? If you have a AMD card, it might not work… 2. Have you waited long enough during installation process to download the necessary models and Lora’s for proper functioning? It’s quite a lot of GB, so depending on your internet connection it can really take a while. 3. You said you have a new Card, have you installed as well the correct WINDOWS drivers for it? I know it might appear as functioning well, but a newer generation of cards require newer drivers. All can be found at NVIDIA.com. 4. What is the error message? I don’t mean inside the FOOOCUS UI, but on your command line tool (CMD) which is running in parallel when having started Fooocus. Also, if you close the cmd tool, be aware the UI of Fooocus stays active, but it can’t do anything and will not generate any images. (Cmd is required, pls ensure it’s not closed)
Hello, yes it does for both, Windows and Linux . Which operating system do you have? You would have to adapt the run.bat file. Happy to help, but for this you have to answer my question.
@@gielh Go to this site here and scroll down until you find "Windows(AMD GPUs)". There is the text you need to add to your run.bat file. It's not complicated. --> github.com/lllyasviel/Fooocus
Thank you for this. I've been using Fooocus for a few months and I love the simplicity compared with A1111. However, at 12:06 we're doing variations. I always get an error "Connection errored out" when trying to do a version or upscale. It also does this on initial generations until I close and reopen. Weirdly, I have a similar issue in A1111 when trying to do anything with any existing image. My computer is more than capable with I think 12 GB VRAM. Any ideas? Not expecting any but hey, maybe you've encountered this :)
Hello, 12GB VRAM is great but might reach the limits when trying to deal with other images. With others I mean your own images. For example, dragging an already bigger (in size and resolution) image into SD/ Fooocus requires quite a lot of compute power when starting to inpaint/ out paint/ upscale, etc. I have a RTX 4090 with 24GB VRAM, but I can also only upscale to 5500 pixels before Fooocus says good night to me. Even thought it’s slower, try to run Fooocus once with the command -cpu-only to see if the problems keeps coming. If it’s gone you know it’s surely related to your 12GB VRAM.
I have less then the recommended amount of vram, I'm sitting at 4, The only thing I have issues with is the size of the image I can make and the use of other extensions that require (8,12) and more Vram. If you need help with some tips and tricks. Ill gladly help a fellow A.I. designer
I do not have a GPU. I have a AMD CPU 12 GB VRAM. Can I run this on my system? Please help lol I've been trying to install stable diffusion a 1111 for like a year haha
Hello, it should work with CPU only. To do so, install as per video and add following line to your run.bat file (open with any text editor) ‘-always-cpu’ and don’t add (‘)…
@@Axel.Peters So I am sorry This is just how it goes for me... So what I have done is, Downloaded the windows file from the Repo the link you providedafter i Unzipped the files then i Double clicked on the run.Bat file and it started installing it downloaded the Big image model all of the different things for the web UI and then it got to the GPU section and it said I do not have a Nvidia GPU and basically stopped so then you gave me the code -always-cpu To add to the Run.Bat file unfortunately that did nothing so I added This line of code to all three batch files now in the folder the anime the realistic and the rundot bad all three now have that code still does not work. So then I went back to the repo and got the code to pip uninstall and pip install all of the dependencies that detect the Nvidia and installed the dependencies that detect AMD GPU after doing both of these things I uninstalled the Nvidia dependencies installed the AMD dependencies Because I have a AMD CPU not a GPU, after doing all of this in the web UI I was able to Give a prompt press create and it showed the progress bar of it creating the image then the whole computer shut down LOL I have a 6 month old computer that's got pretty good specs but not a GPU am I just in a dream world that I could ever make images on my computer? There's nothing else I can think of to do. Do you have any suggestions I tried to give you all of the info I could so you would be on the same page I guess
@@TheGeneticHouse Also take a look here. I want one step further and did something with AI in videos. I demonstrated here what can be done already. If you like it, I am happy to make a tutorial about it: ua-cam.com/video/Jd_Zc_a7SsE/v-deo.htmlsi=w7cFeEMwXHCwDLMa
Take that link here and scroll down until you come to the MAC section. Installation instructions can be found there. Have fun --> github.com/lllyasviel/Fooocus
Hello, difficult to say, might be that you do not have enough VRAM, that you have no checkpoint in place, etc. check the CMD tool for more details. What does it say when hitting generate? (Just in case, cmd is the console that opened when you double-clicked the run.bat file)
Hello, what do you mean it's not working? I tried it just as we speak, and it works for me. You can generate any image with the face of your desire… 1. Activate 'Input Image' 2. Head over to the 'Image Prompt' tab 3. Activate 'advanced' on the bottom left 4. Click on 'FaceSwap', drag and drop an image of a face, and… 5. Hit 'Generate' - Done
Just drop them into your checkpoint folder inside Foocus and in the advanced settings on the Fooocus frontend(Webbrowser) choose a different one. You could also change the settings inside the txt file in the main Fooocus folder to your stable diffusion checkpoints. Fooocus will then search in your specific folder for checkpoints, but still in both ways, you have to select them manually via the advanced tab.
Hello, mhm… does the website starts alone after double clicking the bat file or do you try to enter the IP yourself into your preferred Webbrowser? If it does not open automatically, that’s already the first indicator that the service is not running. What is your command tool saying, what is the message?
Hey, I'm not typing it in myself, when running the bat file it sends me to it. so that part goes well. my command tool doesnt say anything is wrong either. it does tell me that it's trying to enable lowram mode. but I don't think that's the issue@@Axel.Peters
I am exploring Fooocus and it is amazing, the results are great! However my "input image" doesn't work. Any idea how I can activate it? I check the box but nothing happens :-( Thanks! Great video :-)
That is strange, when you say activated, what do you mean precisely? Nothing opens where you could enter an image, or your image is not being taken, or ignored, etc? Let me know, I am sure we come to a good end here
Thank you! Nop, nothing happens when check the input image box...I read the documentation page and search on the internet but can't find anything...I am using windows...perhaps something I should run on PowerShell? Weird, thanks a lot anyway
@@Axel.Peters I want to do it like in the video below, but I couldn't understand it because it was in Japanese:( Could you please make a video about it? ua-cam.com/video/2eZqJLG1dpU/v-deo.htmlsi=u-IBWuhqAVntHNPz
hey it is enabling low vram mode and uses on 2 gb of vram of the gpu i have 4gb vram rtx 3050 in my laptop how to set it to normal vram mode and any tips on performance optimization as advance features lipe face swapping and input image donot work like the terminal says the the cusa is out of memory ??
Hi, you could try the following. Go to the bat file (the one you use to start Fooocus). Edit it with a simple text tool. You should see something like this: .\python_embeded\python.exe -s Fooocus\entry_with_update.py You could add behind that one of the following things: 1. --highvram 2. --normalvram 3. --lowvram 4. --cpu 5. --gpu-only
do u speed the generation time or is it real time, i have a gtx 1660ti 6gb and it takes quite some time. i know this process is based on the graphics card but does it affect the time this much
Hello, yes I did speeding for the benefit of the viewer. Indeed, GPU power is making the difference. I wish my answer would be a different one but your limits are related to your GPU compute power and available VRAM. You can also use low/ mid-end cards but have to live with the time it takes to render. But, you can reduce steps, switch to SD1.5 checkpoints (models). Happy generating
bonjour et merveilleuse année pensez vous que c est possible de crée un model et lui faire porter une robe venant d une photo ? en utilisant l'invite image ou autre ? merci
L'intelligence artificielle a des limites, mais selon le degré de précision nécessaire, je pense que cela peut être fait. Vous pourriez utiliser votre photo avec la robe et simplement échanger les visages pour remplacer la tête existante par celle que vous avez générée. Avec openpose, vous pourriez changer la position générale de la personne. Mais comme dit, cela a des limites.
The minimal requirement is 4 GB Nvidia GPU memory (4 GB VRAM) and 8 GB system memory (8 GB RAM). But the min requirements don't make fun anymore. So, a good GPU with 6 GB+ VRAM and 32 GB of System RAM would simplify things a bit. Otherwise, you have online services, but then you are again in a subscription mode and have to pay for the usage of cloud computation.
1. www.fal.ai/models/fooocus 2. aws.amazon.com/marketplace/pp/prodview-7rd6v2ujpn54e Disclaimer, I don't know how good they are or if the offering is correct and fair. Otherwise, what I mean before was more related to the usual suspects. You can pay for DALL-E, Midjourney, etc.
@@Exscape_Reality easy, just go to your main Fooocus folder. Inside you will find an output folder… inside there folders separated by dates with all generated images per day. In these folders is the log file.
You indeed need to have a rather strong GPU with higher VRAM. But you can limit the amount of VRAM used by FOOOCUS to keep something for playing around while generating images…
Hello, its supposed to work but is not extensivly tested. here are the steps to do so. 1. Install the conda package manager and pytorch nightly 2. Open the macOS Terminal app and clone this repository with: git clone github.com/lllyasviel/Fooocus.git 3. Change to the new Fooocus directory: cd Fooocus 4. Create a new conda environment: conda env create -f environment.yaml 5. Activate your new conda environment: conda activate fooocus 6. Install the packages required by Fooocus: pip install -r requirements_versions.txt 7. Launch Fooocus by running: python entry_with_update.py (Some Mac M2 users may need python entry_with_update.py --enable-smart-memory to speed up model loading/unloading.) More info here: github.com/lllyasviel/Fooocus
Fooocus is Awesome, but I want to know how can I make my Image turn into any style I want (For Eg. Neonpunk) without changing anything in the Image, it should only act as a filter on my Original image without changing the face, body, etc?
two options I figured out so far. 1. Drag your own image into the bottom canvas, under advanced into the upscaling window. Upscale with 1.5 or 2 and your image will appear right after in the main canvas, and you can apply to it whatever you want. 2. Drag your image into the inpainting canvas and paint a very little, almost invisible point somewhere in the image. Then say generate, and it will appear in the main canvas, and you can do whatever you want. Let me know if you understood my explanation.
@@Axel.Peters I tried them both but none of them worked 1) As you mentioned, I upscaled the Image , then changed the style as per my liking and left the prompt Tab blank and hit generate, it generated images but not at all related to my image, it were Random images from that style. 2) Did the Same as you Mentioned in Inpaint, created a Tiny dot, selected a style and hit generate, and it generated my images same as Original nothing changed. If you could find a way, please try to make a video!
@@inspiredactions19 now I understand what you want. Drag and drop an image of your liking to the advanced image prompt tap, then activate PyraCanny, change the weight slider all above to the right (2) and change the style to your liking, for example, neon punk. Make sure you enter a resolution/ aspect ratio that is similar to your current image and hit generate. Done, similar image different style 😀 I just tried it and it definitely works. Please be so kind of let me know if that was helpful. If you want that neon punk image, even closer to your image, you can also take the ‘stop at’ slider and push it all the way up to the right (1)…
@@inspiredactions19 And in case you have Photoshop at hand and the generative fill functionality activated, you can watch another Video from me about different styles and how to apply them to your images. But, as said, it would require PSD. ua-cam.com/video/nKl5p6kLDFc/v-deo.htmlsi=4ptQP9XKkMdECP7m
Hello, sure. Easy.. You can generate any image with the face of your desire… 1. Activate 'Input Image' (bottom left) 2. Head over to the 'Image Prompt' tab 3. Activate 'advanced' on the bottom left 4. Click on 'FaceSwap', drag and drop an image of a face (in your case, your dad) 5. Prompt an image (whatever you want) 6. Hit 'Generate' - Done 😀
Another thing ive seen is that fooocus really struggle when it comes to using a pyracanny from image A and faceswap from image B> although im running a 4070ti with 64gb of ram
@@holetarget4925 you could take an existing picture of a man model, drag it into the image prompt but use for the first slot PyraCanny (to take the pose of the model), in the second slot your dad‘s photo and let AI do something that comes at least close to the original picture. Help guide the AI by saying what you want. Otherwise there are free online services that do exactly what you have ask for.
1. Before you tell people how to download a program, you MUST tell them the minimum requirements for the hardware they must have to use it. Otherwise, many people will simply waste time and nervous energy. You need to think about all viewers of your videos, not just those lucky ones who have new video cards with 4G Nvidia or 8G AMD memory.😑 2. When you show the process of creating images, you MUST mention that you are using fast playback - don't give users the false impression that the images were created within a second.😑
Thank you for your valuable feedback. My aim is always to produce informative videos that keep viewers abreast of the latest technology and trends. This particular video is available at no cost, and I regret that I overlooked the two key points you mentioned. I am more than willing to incorporate your suggestions, but I also kindly ask for your understanding and empathy towards the effort that goes into creating this content and your way of approaching what you think is your right for free content.
Only thing I don't like about Fooocus is, everytime I hit generate it says "loading the model to GPU" and waste few seconds to generate images. Does anyone else facing the same issue? any fix for that? (I have a RTX 3060 12GB VRAM and 16GB RAM)
Hi there, there are indeed a couple of seconds, but really just a couple. I don't found this very noticeable. On the other side, you don't spend hours in creating convincing prompts and feeling right after like a real prompt engineer. Fooocus has a GPT2 conversion model in place. You keep it simple, and the results are always convincing for me. On top, plain and simple style guides and everything else you need to get something quickly with top-notch quality. 🙂
Hello, you seem to have minimum system requirements, but might really be related to both VRAM and System RAM. But it’s a guess only. Try to edit run.bat file (the one you normally doubleclick to start fooocus) adapt the first line: \python_embeded\python.exe -s Fooocus\entry_with_update.py You could add behind one of the following things: 1. --highvram 2. --normalvram 3. --lowvram 4. --cpu 5. --gpu-only In your case use (3) or (4)
Whenever I try running the bat file. I get the following error Fooocus>. \python.exe -s Fooocus\entry_with_update.py The system cannot find the path specified. Anyone else get this?
Hello, that really should not happen. Fooocus, when complete, should have standalone python integrated and the path should work without the need for an offline and local python installation. Have you downloaded everything correctly, and is the fooocus folder still complete (incl. python embedded)?
you’re curious about whether you can use Fooocus AI with an AMD GPU and how to go about it. After looking into this, it seems that there are significant compatibility issues when trying to run Fooocus on systems with AMD GPUs. This is primarily because Fooocus AI relies heavily on CUDA, a technology developed by NVIDIA, which isn’t supported on AMD hardware. One user on GitHub noted that despite installing torch from the AMD package index, Fooocus AI still attempted to use CUDA, leading to errors and incompatibility . Unfortunately, the Fooocus AI GitHub repository doesn’t offer specific guidance or solutions for AMD GPU users . Given these challenges, it appears that currently, there isn’t a straightforward method to run Fooocus AI effectively on an AMD GPU. If you’re keen on using Fooocus AI, you might need to do so on a system with an NVIDIA GPU, or consider other AI software that’s compatible with AMD hardware. I understand this might not be the news you were hoping for, but I hope it helps clarify the situation. If there are any updates or workarounds in the future, the Fooocus AI GitHub repository is a good place to keep an eye on.
As promised: here is the link: github.com/lllyasviel/Fooocus
Would you have the scope to catch up in discord voice chat? Just wanted to have some intelligent conversation, since you are more experienced in AI tooks
Thanks for the guide, but when I downloaded it, the 'run bate ' option is not visible. Instead, only the 'Run' option is visible. After clicking on it, the file leads to a similar page, but an error 'connection timed out' is showing. I got a GTX 1650
This was a very helpful tutorial. In particular, I really liked the detailed explanations you did for the various image prompt options. I'd love a guide on more advanced usage, like complex prompt writing with weighting and multi-prompt, and examples of how to use Styles and LoRAs in combination to make a great image.
Thanks again for the tutorial!
Oh, that is nice feedback. Thanks a lot, Jake. I am happy to make such thing as a follow-up tutorial. But please know, every style for example is 'nothing more' as written prompts predefined to reach a certain look without the user finding it out himself. Also, Fooocus is already using similar to Midjourney a GPT2 conversion. Means, no need for complex prompt engineering. Otherwise, weights work the same as in Stable Diffusion. Imagine you want to see a space soldier and a spaceship, if your results are ignoring the spaceship, give it a weight "(spaceship visible:2)" for example.
Hey @@Axel.Peters Please do make a tutorial on how to use Styles and Loras in combination as I am a complete beginner. And if you could show a way to fine-tune the faceswap feature
@@Axel.Peters Oh, that's interesting. So do you rely on the styles or do you try to accomplish it all in the actual written prompt. I imagine using both is a lot of trial and error.... is there any way *we can* find out ourselves how the style is actually predefined? Right now I am trying to wrap my head around best practice when it comes to styles, refiners and loras and how they play in conjunction with each other.
Thank you for the tutorial. My PC almost caught on fire from the exces stress I put on it when I diden't need to.Thanks to you I finnaly how to optimise the program and make full use of the options. THANK YOU
Thanks a lot for your very kind feedback. Have fun and happy creating…
how can the pc catch fire if there is cooling?
wow very detailed yet simple to understand tutorial! thanksss
Appreciate your kind words I hope you could take something out of it, and if you have ideas or interest in other tutorials, just let me know, or check my channel.
This video is simply EXCELLENT. Thank you!
Such great feedback - this is very kind of you. Thank you so much… 😊
For those with not strong PCs with good amount of VRAM or RAM use the colab option to generate images without using your own computer and its faster
Good point - agree to your comment.
Great video. Very clear and easy to follow. It would be great if you could share some tips or do a video on how to generate accurate hands and feet.
Thank you very much for that great feedback. Concerning your question, I will have a look if I can create a video around that topic.
Thanks for the tutorial it was helpful. Some of the shake screen transitions during editing were a bit jarring and made it harder to watch but overall it was made well.
Thanks for your great feedback. Much appreciated.
This is a dream come true❤❤❤
Happy to hear you’ve got some value out of it 👍🙏
Happy to hear you’ve got some value out of it 👍🙏
I appreciate this tutorial. Thank you very much!
Thank you so much… such kind feedback is motivating me to keep going… 😃
Thanks for sharing this tutorial, I was tried to use the new Defooocus to change clothes using both image prompt & inpaint function but I never able to matching over 70% on some patterns or any patterns like patterns on T-shirts or any printed patterns. Would you make a video on how to mach a pattern as close as possible. Thanks in advance.😊
I've been using Foocus for some time and it's very good but the native interface (very small black text on a white screen) is niaive, poorly laid out and VERY difficult to use!
What display interface are you using. It seems to be exactly what I've been looking for!
Hey there Nice informative video! Any tips to get ‘better’ results with face swapping as im trying to have a consistent model and whenever i try to get the desired face in different scenarios it always looks very fake and like the face does not fit on the body
Thank you, amazing guide.
Thanks for the motivating feedback. If you need something else, please let me know
Great tutorial!!! Very helpful!
Thank you, very kind feedback. Let me know if I can further support. Have a great time.
Nice
Thank you very much… appreciate your feedback 😊
Hello when I use the run file to first start everything I get a HTTP 503 error and it wont download the juggernaught. TY
Hello, it really should! Do you really waited, long enough? The file is rather big, so depending on your Internet connection, it might take a while. Can you check again?
Thank you for this job !
That is very kind of you. I hope you enjoyed and could take something out of it… Have a nice day
Can somebody help? Why this program eating my space on disk C, although installed on D, how change settings?
Hello, if you have FOOOCUS installed on your D drive, the models and generated images are located as well on the D drive. However, this doesn’t alter your standard Windows configuration. Typically, paging files are located on the C drive. This means if your VRAM or system RAM reaches capacity, the system will utilize paging files, which are likely on the C drive where Windows is installed. After a fresh restart, do you notice an improvement in the available space on your C drive? If so, my assumption might be right. You could change your paging files to another drive via the Windows system settings.
Thank you for this!
Appreciate your great feedback! Thank you
THIS IS SOOOOOOOO GREAT!! I JUST CANNOT WAIT UNTIL IT HAS AMD SUPPORT AND LOWER PC SYSTEM REQUIRMENTS SO THAT I CAN USE IT 😁💌
Hello, it has both already available. AMD support as well as low system requirements. All that needs to happen for the low system is an adaptation to the original run.bat file. Let me know how f you need help.
@@Axel.Peters i actually tried using fooocus on my brothers pc and work pc as well.a
my brother has amd and i received some error messages saying that I need nvidia. his pc is pretty powerfull i might say.
yet, on the work pc, i see nvidia when i right click on the desktop, but i am not sure what graphics card it has. maybe the drivers are installed incorrectly or something like that. it is a weaker pc than that of my brothers and even mine. in every cases it shows various errors, mostly saying that I need nvidia graphics card. at work, i opened foocus and it just says error. i cannot generate something.
i even tried google collab, but i an new to it and i might used it wrongly. it says nvidia graphics card error as well. what a shame!
I need to make this fooocus work. i find it very lucrative for some projects that i have envisioned. i would love some help from someone who knows more programming than i do. how did you made it work?
I understand you’re curious about whether you can use Fooocus AI with an AMD GPU and how to go about it. After looking into this, it seems that there are significant compatibility issues when trying to run Fooocus on systems with AMD GPUs. This is primarily because Fooocus AI relies heavily on CUDA, a technology developed by NVIDIA, which isn’t supported on AMD hardware.
One user on GitHub noted that despite installing torch from the AMD package index, Fooocus AI still attempted to use CUDA, leading to errors and incompatibility . Unfortunately, the Fooocus AI GitHub repository doesn’t offer specific guidance or solutions for AMD GPU users .
Given these challenges, it appears that currently, there isn’t a straightforward method to run Fooocus AI effectively on an AMD GPU. If you’re keen on using Fooocus AI, you might need to do so on a system with an NVIDIA GPU, or consider other AI software that’s compatible with AMD hardware.
I understand this might not be the news you were hoping for, but I hope it helps clarify the situation. If there are any updates or workarounds in the future, the Fooocus AI GitHub repository is a good place to keep an eye on.
how about system resources utilized during the usage of this tool?
Not sure I fully get your point. But it uses mainly your GPU (faster = better), and the available VRAM.
Thanks for the video. How did you get all of the information about different settings? Just by playing with it? I'm trying to figure out how to (1) see what the different "styles" add to the prompt and (2) learn what the different image prompt settings do
Hello, (1) you can see the style prompts in the running terminal when rendering or in the history log file. (To open the log file, activate advanced bottom left and then you see history log file and a link for it)
Not sure what you mean by your second point. Would you elaborate a bit further?
the eyes
Little tipp
Make populs like 0:31 bigger or even full screen
I cant see anything on my phone :)
I am happy you said phone and not smart watch 😉… thanks for the Tipp, will care about in the next tutorial. Thanks a lot.
How to setup the same face from input image to another image please help
I downloaded the zip file, extracted and then downloaded the 6gb file later it also downloaded 222mb and the window just closed. in many videos i have seen an offline window opening on the web browser immediately after the 6gb file is downloaded. is there any way out of this or does my laptop not support fooocus?
The file you DoubleClick to install is the same file you use to start the program. So what is happening when you start it again, will it go back to download files or will it open Fooocus (on the browser)?
is faceswapping not possible on free version?
Not sure I follow - There is no paid version. Everything is free and face swapping is absolutely possible. ☺️
very instructive tutorial! do you know the best way to create different emotions for the same person?
Not sure I follow entirely. But if you want to keep the person and generally speaking the position, then you could just do some sort of inpainting by selecting the face and create a prompt by describing a happy, sad or angry face, lips, etc.
i allways get error, if i try to create something. can you help?
Hello, what do you mean with error? Can you be a bit more precise? Error during installation, usage, image creation, missing models, etc. ? Where and in what process you have an error and what is the error message?
18:33 but how i can make it run on colab my laptop not high enough plz i want part of anime
Hello, try this here: colab.research.google.com/github/lllyasviel/Fooocus/blob/main/fooocus_colab.ipynb
Can you please make a tutorial about negative prompting in fooocus? With adding weight to it. I keep struggling with a t-shirt WITHOUT text or logo. Same for a baseball cap. It seems fooocus doesn't know those even exist. 😁
now how create many pictures of the same "actress" my own model? can I do it there?
For example:
1. Initial Image Generation: Begin by creating a front portrait of the person. This image serves as a reference for subsequent modifications. Utilize features like Extreme Speed in Fooocus AI for rapid iteration and selection of a preferred face.
2. Template and Image Prompts: Once you have the front portrait, use a template designed for generating portraits in multiple positions. This process involves using two image prompts: the original front portrait and a template that specifies positions (e.g., front, side, etc.).
3. Adapt Text Prompts: Along with the image prompts, modify your text prompts to indicate the desired positions or angles, like “multiple portraits in different angles.”
4. Image Splitting and Upscaling: After generating the composite image with multiple poses, split it into individual images. These can be upscaled (e.g., 1.5x) to enhance quality.
5. Faceswap Adjustments: You may further refine the images by using the Faceswap feature. Adjust the weight of the images to change the angles or perspectives.
6. Handling Pose Variations: For changing the pose direction (e.g., making the photo look in a different direction), consider using a combination of PyraCanny (for edge mapping) and Faceswap, along with appropriate prompts.
if i increase the aspect ratio / higher landscape resolution will the photos take longer to generate? cuz what im using rn takes 51 min for 10 images( 1:1 1024x1024 and other factors involved as well)
Hello, yea it will increase the time to generate the image. Think in that way, more pixels, more time to generate.
51min for 10 images? What are your specs and f I may ask?
51min for 10 images? What are your specs and f I may ask?
Hello. Thank you for this important tutorial video. It is clear how much time you spent in order to let us execute all your instructions precisely specially concerning all the note you wrote below. For me, I have an issue. After extracting the Focus Zip file , you can read ( run ) but (bat) is not showing and not able to run this file. Thank you in advance.
Hello, thanks for this great feedback. Concerning your question, I’m not sure if I really fully understand your problem. For me it sounds like you see the run file, but you are having the problem that you can’t see the file extension (.bat), correct? if so, it just means that windows is hiding the extensions for all your files, but this is just an easy system fix. Generally speaking, it doesn’t matter whether or not you can see the extensions in theory you would just have to double click the run file. Can you try double click the file?
I’m sorry, great video but I have a question. I’m running this on colab and everything is working but I would like to know why every time I run it looks like the memory “disk” is decreasing a lot. Why is that and how can I fix it, do you have any suggestion? I really can’t explain myself better but I’m referring to the thing in the left and down part of the screen where it says disk n/100gb
Hello, I can't help a lot with colab space. I assume you are limited on the cloud space available. When you say decreasing, I assume you refer to the available memory and if so, that is normal. The generation process takes quite some compute CPU/ GPU and memory power (VRAM / RAM) ;-)… That is the reason (one out of many others) why people use online services (Midjourney, Dall-E, etc.) that do all the compute work for them and no high-end hardware required. Anyhow, as long the decreasing memory it is not hindering you in your creative process, don't bother.
Hi , I've installed the package, but when I run it, it doesn't install the sd_xl, and when I run the program I cannot generate an immage it just says error. I tried to install them separetely but still does says error, can you help me?
OK, help me a bit here, what is cmd saying? Is it stopping the download or doesn't even start, does it take long and feels like there is no progress? Just be aware, the download of the checkpoint model takes quite a time since you download around 7 GB and depending on your internet connection quite a ride. You said, you manually downloaded the checkpoints, have you renamed them? Have you dropped them into the correct folder inside Fooocus? Have you once the web server (Fooocus frontend) started activated advanced on the bottom left and manually selected your downloaded checkpoints?
Hello, I don't have enough spec for Fooocus, not even GPU. Can I run smoothly after purchasing Google Collab pro?
Hello! Yes, it should operate effectively on Colab Pro. The term “smoothly” is subjective, so experiences may vary. It’s unlikely that you’ll have access to an RTX 4090, which is a top-tier GPU, but you’ll certainly be able to generate images with ease.
Thank you @Axel.Peters . I dont have enough GPU and i am getting NVIDIA error. I will go with Collab Pro then. If that works. Thank you so much again. I am waiting for your next vids.
@@dudewheresmypurpose my pleasure and I wish you a lot of fun! Get back to me if you need support in any way…
after closing fooocus , how i can open it again
Hello, the same way you opened it the first time. One of the three bat files. Double click - done. The first time it downloads certain models and the second time it just opens since everything else is already in place. Hope this helps.
Hir Sir Just wanted To ask last night my pc got crashed shutdown while generating The Model. I have 12gb of ram installed with nvidia gtx 1650. And it takes too long to generate the model. Any help would be really appreciable ?
Hello, AI is rather hardware intensive and I fear not being able to help you in regard to the speed. You could try to reduce the steps to generate an image (use speed instead of quality in the advanced settings). You could decrease the output resolution and only when you are happy with your outcome you could start to upscale.
Can u do the same on Android devices
Can you specify what you mean? Do you want to install Fooocus on an android device?
Hello bro can you please give me solution of my problem
Now the google Collab is not giving me permission to use fooocus ai
They are telling that i need to buy their subscription and getting error
Please answer me 🥺
Hello! It sounds like you’re accessing Collaborative cloud resources rather than your own hardware. It’s reasonable for the service provider to request payment for their hardware offerings. If you prefer not to incur these costs, consider using your own equipment. For this, you’ll need a reliable computer equipped with a capable NVIDIA graphics card with sufficient VRAM. Please feel free to reach out if you need further assistance. Wishing you the very best!
If i want to generate it directly says error?.. can you help me out i did everything you did but everytime it says error
Hello, what error, what is the message you receive? How much VRAM do you have?
@@Axel.Peters the error is direct when i generate, and my vram is 128 MB
Hello,
128 MB of VRAM is significantly less than what is common these days, where people usually have 4GB or more. This could be the reason for your problem. To adjust for this, you might have to alter the run.bat file that you use to start Fooocus. You have to add the command to use your CPU only, without overall GPU usage. The command is:
-always-cpu
Please note that running Fooocus AI in CPU-only mode can be considerably slower than using a GPU, so you should expect longer processing times.
@@Axel.Peters thank you so much for you help!
It was a real pleasure! Have a wonderful day
I have 16GB of Ram and gtx 1660 with 6GB VRAM. It currently uses low vram mode and gets to around 5.1 GB Vram usage. Can I optimise by replacing low vram mode with medium vram mode?
Hello, you can surely do that. There’s no harm done figuring out that it might not work.. your system might unstable or unresponsive, but after a quick restart, it should work normal again. And if you noticed that it doesn’t work, change the command back to what it was before. But if you ask me, there is not much room for improvement knowing that you have used already 5.1 GB versus 6 GB you have available - there’s not much room… Also, think about that you might want to upscale your image afterwards or that you want to inpaint, all that might require more VRAM.
Thanks for the tutorial, it was really helpful. I only have a question. If I already have SDXL models downloaded and don't want to download the ones in the Installation. How could I skip this step?
Thanks, happy to hear you have found it useful. To answer your question, you have two options.
(1) It will download default models to the folder "Fooocus\models\checkpoints" --> If you have them already, just copy them over.
(2) If you have a folder elsewhere with all your checkpoint models, change the path inside the config.txt file with a simple text edit tool to your own folder. (config.txt can be found inside the main Fooocus folder)
I have been having fun with this prompt Character 1 (Spider-Man) is talking to Character 1, Character 2 (Deadpool) is listening to Character 1.. With interesting results. I have even been able to get up to Character 3. But that's really rare. Also if you prompt more than 2 characters it starts to blend them very funny. I want to know how to control the camera. As I am wanting to use this to start my Comic book Series. And so far NOTHING I have tried has given me consistent results.
That sounds like real fun! The camera you could control with Loras for example. Also, to control the camera angle and position in a prompt for image generation, you can specify these aspects explicitly. Here’s an example prompt structured to include these details:
Prompt Example:
Subject: Victorian-era street scene, bustling with people
Camera Angle: Low angle view, capturing the grandeur of the architecture
Position: From the perspective of a pedestrian on the street
Lighting: Soft afternoon light, long shadows
Modifiers: Vivid colors, high detail
Breakdown:
Subject: Clearly describe what the scene or subject is.
Camera Angle: Specify the angle like low, high, bird’s-eye, etc.
Position: Indicate the vantage point or perspective, such as pedestrian view, overhead, close-up.
Lighting and Modifiers: Detail any specific lighting (e.g., golden hour, diffused light) and other modifiers like color scheme or focus.
I hope that helps.
@@Axel.Peters Danke. I’m really liking this and once I think I got some good pages I’ll share some. But give me a bit as now that this tech is avail, I have a very good outline just need to flesh the script out. This is going to be fun.
@@SuperMouseDV 😁 gerne… let me know if I can support in any way.
Thank you . Ill totally take you up on that offer
@@Axel.Peters
@@Axel.Peters Question how can I make the Ship from Flight of the Navigator show up in my render and/or the robot from that movie.. ? Is that possible or do I need to Train that into a Loras?
The program always says waiting for task to start and doesnt to anything. Do you have a fix?
Hello, where and when do you get this message?
@@Axel.Peters when i press generate a prompt, it appears next to the loading bar. Or it used to. It seems that it got fixed on its own? I dont know. Anyways thank you for the quick reply. I still have no idea what the problem was lol.
@@VTick712 the first time it takes a bit to load the checkpoint into the memory. That can take a while… glad you have solved it. Happy prompting, and use the styles as described in my tutorials. Simplifies your prompting…
@@Axel.Peters thanks!
My pleasure. Have fun
Okay, I think it's installed but how do you start the program?? There's no icon on the desktop or and .exe file to be found. Can't navigate the "simple" screens if no one tells you how to access the program.
Hello, it’s the same bat file you used for installation. There is no exe… Start the bat file - and it will open if all is installed correctly. Have fun
@@Axel.Peters well, I have done just that multiple times and it only reinstalls and then nothing every time. Even using the IP address does not work. Says something like it doesn't exist. And so far, it is NOT fun.
How to set low vram for 4gb vram
Hello... Try to add one of them as text into the bat file (the batch file you use to start Fooocus)... in your case, I assume (3) or (4) would work best:
1. --highvram
2. --normalvram
3. --lowvram
4. --cpu
5. --gpu-only
Are the generating speeds in the video real time, or did you speed up through them during video editing?
Hello, I speeded it up a bit. But still very fast… in any way, all depends on your GPU power.
I was just trying to gauge a speed difference between your GPU and my 7900xtx, which takes a few minutes to generate speed oriented image.
Thanks
@@muziqaz I think mine takes around 30sec for quality settings (60 steps)… (RTX4090)
@@Axel.Peters Ah, this is considerably faster than mine. Thanks
It just says "Error, connection timed out" when im trying. Are my speccs too bad? my gpu is GTX 1660 Super.
Hello, I really don't know. You could try to start Fooocus with the 'lowvram' or 'cpu-only' command. Let me know if you are unclear about the commands, happy to help out.
Same issue here
@@ManniBOY Try to edit run.bat file (the one you normally doubleclick to start fooocus) adapt the first line: \python_embeded\python.exe -s Fooocus\entry_with_update.py
You could add behind one of the following things:
1. --highvram
2. --normalvram
3. --lowvram
4. --cpu
5. --gpu-only
In your case use -lowvram
Hey, thanks for the advice. But the error went away magically. Just waited a couple of hours and tried again@@Axel.Peters
What pc specs how does yours render so fast I have Amd 7900xt doesn’t go as fast
Good morning, I wanted to ask once I turn off the computer. When I turn it back on, how do I launch Focus? Without having to reinstall it. Is it possible to create a shortcut on the desktop?"
Hello, you don’t install it over and over again. All you do you is to execute the bat file. It finds all the formerly installed components such as the checkpoint models, Lora’s, etc automatically… I hope this helps.
ok thanks, can you tell me what the files would be to launch the application@@Axel.Peters
Hello, to start Fooocus, you’ll need to use the same .bat file that you initially used for its installation. This file is located in the main folder of the Fooocus program. Just as you did during the installation process, open and run this .bat file to get Fooocus up and running. If you need further guidance, please refer back to any instructional materials or videos provided during the installation. Have a great day!
I have 2 problems in fooocus.
Eye quality and consistent on char. Even when I choose face swap.
Any hint?
did you solve it?
@@user87546 for consistency you can only use 1 model at the time.
About the Eyes no solution until now. I know some people that used some eyes Lora's.
I'm using A1111 and with Adetailer all eyes are solved.
Would you have any prompts to detail the anatomy of the ear?
Just the ear? You mean a separate image with the ear as such, or a prompt to improve via inpainting the ear of a person?
@@Axel.Peters a prompt to improve via inpainting the ear of a person. I'm having trouble about ears. =/ Than you so much.
What checkpoint do you use when experiencing it? Standard in that case Juggernaut or the realistic.bat version? If not, try the realistic.bat version, it uses a great checkpoint for photorealistic things. I am having no clue what difficulties you are experiencing, but all I would do is inpainting the ear of the person and adapt the prompt to serve your needs. With your needs, I mean - better explanation of the earring or accessories you would love to see. Make sure you use the standard inpainting feature in Fooocus, not the one that adds or changes the background. In case you have deformed ears, use the Fooocus 'enhanced' style, it is a negative prompt collection that solves such problems very well.
How does it generate so fast? My pc is not that bad and it takes minutes to generate an image.
Hello, no… I was speeding it up for the benefit of the viewer to focus on the things that matter rather than waiting for something to render. I have a RTX4090 with 24GB VRAM and with system RAM of 100GB. It takes around 30sec + to generate 4 images as a batch.
I'm going through hell trying to install in my macbook pro! Are there any instructions?
Hello, you can install Fooocus on Apple Mac silicon (M1 or M2) with macOS 'Catalina' or a newer version.
Take that link here and scroll down until you come to the MAC section. Installation instructions can be found there. Have fun --> github.com/lllyasviel/Fooocus
I have macOS 14 Sonoma & I folloewd the instructions from your link, everything installed & updated correctly, but I think I'm missing something because I can't seem to find the correct prompt to launch or open fooocus. I'm a pc person but right now only my macbook is available😣@@Axel.Peters
just wow
Very kind feedback. Thank you so much.
Can you please teach how to add desired lora model in Fooocus in collab
When i use fooocus, one image takes up to 10 minutes even tho i have good pc specs, any fixes
Hello. Not sure what are you considering good specs, but it should not take 10 min. I can only imagine that your VRAM is full, and therefore it slows down your Fooocus activities . You would have to check your VRAM usage during rendering via the task manager (performance). Furthermore, I would as well try to use Fooocus with a --medvram or --lowvram command. Enter it to the existing text inside your bat files.
my fooocus when run it got pause within 1 minut how to solve ?????
Hello, please verify if this issue is related to your VRAM. When VRAM is full, the system tends to slow down and may become unresponsive. If this is the case, you should modify the Fooocus AI startup command to start with lower VRAM usage. Add the following command-line argument:
--lowvram
This command enables the low VRAM mode in Fooocus AI, which is designed to reduce VRAM usage and optimize performance on systems with limited VRAM. Please be aware that while this adjustment can help in running the application more efficiently, it might also impact performance.
Im wondering if the pictures created with fooocus is free of copyright? So can I just post the pictures on instagram etc?
I am not an expert in that area, but here are the licenses that are being used by Stable Diffusion: www.licenses.ai/blog/2022/8/26/bigscience-open-rail-m-license
Hi again.Thank you for your prompt reply on my comment. I did a double click on the file and you were right. Running the file is tacking time and suddenly stopped on 49 % from a few minutes ago. So can you please tell me what to do ? Thank you very much.
I’m am not a magician, but I assume the 49% is referring to the download of the checkpoint. It’s not related to the software as such, but for whatever reason, your download stopped, might be that your internet connection dropped… the file/ checkpoints that the program is going to download are rather big, around ~ 6GB… just try again, and ensure your connection stays stable. If that doesn’t work, I can provide you the link to the first file, and you can download it separately…
Hello. Thank you very much. I hope to arrive complete the download. Best regards.@@Axel.Peters
Hi. You are a nice person. I appreciate. Good luck.@@Axel.Peters
@@georgedeeb4483 thanks a lot. Let me know if that worked out. As said before, if not let me know, I can provide you otherwise the link for the checkpoint.
Thanks for the tutorial. I was wondering: Is there a way to uninstall it once used?
Hello and thank you very much for your great feedback. concerning the deinstallation is rather easy… it does not need to happen, all you need to do, is to delete the Fooocus folder. You have a portable version, that means everything that Fooocus requires is within the folder.
@@Axel.Peters thanks for the response
@@waterglas21 of course. You ask for good reasons. So, happy to help out. 😀
Having connection error
Is there a way to fix that ?
Using a 1660ti
Is the connection error preventing you from using fooocus? I am asking because cmd sometimes shows that message but it still function correctly… might also be that you loose connection when your VRAM is full. You eventually need to start your run.bat with an extra command towards low VRAM availability or CPU usage only. If you don’t know the extra commands, check the other comments or get back to me, happy to provide them to you.
@Axel.Peters yes I'm not able to use it, I'm sorry, as im not aware of the commands to use to fix this issue
I understand you’re experiencing connection error messages with Fooocus AI on your NVIDIA 1660ti, which might be related to VRAM limitations. One way to address this is by modifying the run.bat file to use less VRAM. Specifically, you can try adding the --always-low-vram argument to your run.bat file. This command switches the program from normal VRAM mode to low VRAM mode, which could help run Fooocus more stably on your system.
Here’s how you can modify the run.bat file:
1. Open the run.bat file in a text editor.
2. Add --always-low-vram at the end of the command line.
3. Save the file and execute it again.
This adjustment should allow Fooocus to operate in a mode that’s more compatible with the VRAM capacity of your GTX 1660 Ti. However, keep in mind that running in low VRAM mode might impact the performance or speed of the application.
Hey nice artwork! Just want to ask that are you interested in our new ai generation product? Maybe I can send you beta version to you~
Hello, sure.
Great video, but it doesn't work for me for some reason. It stops at model_type EPS adm 2816. I downloaded 3 base models, and it still doesn't work. Do you have any idea? / GTX 1660 TI / 531.18 - 16GB RAM / WINDOWS 10
Hi, it might be related to your system specs. Try to edit run.bat file (the one you normally doubleclick to start fooocus) adapt the first line: \python_embeded\python.exe -s Fooocus\entry_with_update.py
You could add behind one of the following things:
1. --highvram
2. --normalvram
3. --lowvram
4. --cpu
5. --gpu-only
In your case, use lowvram. Should work.
@@Axel.Peters it didn't work . yes, i add it correctly. It says 'Set VRAM state to: LOW_VRAM' in CMD .i looked up on the Fooocus github page and found out that another person has the same issue as me.(they posted about it 14 hours ago). should I download the previous_old_xformers_env.7z version? It might work .
@@hashYT_ did it work
@@EmmanuelNcube-pu8xh nope . i switched to automatic1111
Can we generate our own images and how? Plz make a detailed video. Subscribe done 👍🏻 and thanks for sharing this amazing video.
Hello, help me to help you 😀... What do you mean with your own images? I thought that my tutorial was exactly about generating your own images… But I assume that you mean something else, right? Are you possibly referring to your own face?
@@Axel.Peters Tysm peter for ur reply, actually i was talking about how we can generate our own face image in own style? Bcz i don't wanna use someone else's style, i don't wanna do that, remove someone else's face and put your own. That's my point brother.
@@OnesiderTTtheres pyracany and faceswap there.
so you can generate urnown face there
can i not download this withou an nvidia graphic card?
Hello, yes you can. The download is exactly the same, but the run.bat file need to be adapted. You would have to open the bat file with a normal text edit tool and add some lines to it. It's easy, the text you have to add can be found here --> github.com/lllyasviel/Fooocus (Scroll down until you come to the section of AMD GPU's)
Hello. I'm just curious to know if anybody else is experiencing Fooocus running VERY slowly? I don't know if I did something but half the time it crashes and won't run.
Hello and good morning. I’m sorry to hear that It doesn’t work very well at your end. A slow and unresponsive program might come from not having enough VRAM available. May I ask what kind of graphic card + VRAM you have installed?
@@Axel.Peters Hello and thank you for taking time to reply. My Vram is: 6 GB My Graphic card is: Geforce GTX 1660 Until now I've only been able to successfully run Fooocus from a link in a video tutorial that launches it on my PC. Sadly, Fooocus only runs at night but tends to crash periodically until I reload the program. Any thoughts you may provide will be appreciated. This is the only AI program that really I really enjoy and best of all, it's free. I would hate to lose use of it. Thank you again.
I really believe the problem is related to your limited amount of VRAM. You should start fooocus with a different setting so that it doesn’t use all of your available VRAM. For managing VRAM usage, options like --lowvram or --normalvram, which can be added to the command line or batch file to adjust resource allocation.
@@Axel.Peters Hi. Thanks for replying. Today I installed a new Graphic Card with 12 GB of vram. I started to use Fooocus and is began issuing "error" warning running down the screen again. In all honest, due to my lack of experience with this software I have no idea how to begin trouble shooting. Any thoughts on where I can look to resolve this? Thanks again.
Hello, don’t give up. I thought I covered all cases somehow already in the video but if you followed it 1:1 I really don’t know what else it could be. Couple of questions though…
1. Did you used a NVIDIA card? If you have a AMD card, it might not work…
2. Have you waited long enough during installation process to download the necessary models and Lora’s for proper functioning? It’s quite a lot of GB, so depending on your internet connection it can really take a while.
3. You said you have a new Card, have you installed as well the correct WINDOWS drivers for it? I know it might appear as functioning well, but a newer generation of cards require newer drivers. All can be found at NVIDIA.com.
4. What is the error message? I don’t mean inside the FOOOCUS UI, but on your command line tool (CMD) which is running in parallel when having started Fooocus. Also, if you close the cmd tool, be aware the UI of Fooocus stays active, but it can’t do anything and will not generate any images. (Cmd is required, pls ensure it’s not closed)
Hi, Great video. Do you perhaps know if you can run this with an AMD GPU?
Hello, yes it does for both, Windows and Linux . Which operating system do you have? You would have to adapt the run.bat file. Happy to help, but for this you have to answer my question.
@@Axel.Peters Thank you, help is always welcome as i'm new to all this. I have Windows 11
@@gielh Go to this site here and scroll down until you find "Windows(AMD GPUs)". There is the text you need to add to your run.bat file. It's not complicated. --> github.com/lllyasviel/Fooocus
@@Axel.Peters Aaah I see what I did wrong now. Thank you so much for your time and help!
Are you currently open to external collaborations, such as evaluating AI products?
Hello Zoey - Sure, happy to collaborate.
Sounds good, which way to connect you? I don't see your email here@@Axel.Peters
Can you share your email with me here? Once seen, you can deleted it.
@@ImageEnhance axelpetrs@web.de
@@ImageEnhance hello, you got the address?
Thank you for this. I've been using Fooocus for a few months and I love the simplicity compared with A1111. However, at 12:06 we're doing variations. I always get an error "Connection errored out" when trying to do a version or upscale. It also does this on initial generations until I close and reopen. Weirdly, I have a similar issue in A1111 when trying to do anything with any existing image. My computer is more than capable with I think 12 GB VRAM. Any ideas? Not expecting any but hey, maybe you've encountered this :)
Hello, 12GB VRAM is great but might reach the limits when trying to deal with other images. With others I mean your own images. For example, dragging an already bigger (in size and resolution) image into SD/ Fooocus requires quite a lot of compute power when starting to inpaint/ out paint/ upscale, etc. I have a RTX 4090 with 24GB VRAM, but I can also only upscale to 5500 pixels before Fooocus says good night to me. Even thought it’s slower, try to run Fooocus once with the command -cpu-only to see if the problems keeps coming. If it’s gone you know it’s surely related to your 12GB VRAM.
I have less then the recommended amount of vram, I'm sitting at 4, The only thing I have issues with is the size of the image I can make and the use of other extensions that require (8,12) and more Vram. If you need help with some tips and tricks. Ill gladly help a fellow A.I. designer
I do not have a GPU. I have a AMD CPU 12 GB VRAM. Can I run this on my system? Please help lol I've been trying to install stable diffusion a 1111 for like a year haha
Hello, it should work with CPU only. To do so, install as per video and add following line to your run.bat file (open with any text editor) ‘-always-cpu’ and don’t add (‘)…
@@Axel.Peters thank you so much I will try this and let you know how it goes :-)
@@Axel.Peters So I am sorry This is just how it goes for me... So what I have done is, Downloaded the windows file from the Repo the link you providedafter i Unzipped the files then i Double clicked on the run.Bat file and it started installing it downloaded the Big image model all of the different things for the web UI and then it got to the GPU section and it said I do not have a Nvidia GPU and basically stopped so then you gave me the code -always-cpu To add to the Run.Bat file unfortunately that did nothing so I added This line of code to all three batch files now in the folder the anime the realistic and the rundot bad all three now have that code still does not work. So then I went back to the repo and got the code to pip uninstall and pip install all of the dependencies that detect the Nvidia and installed the dependencies that detect AMD GPU after doing both of these things I uninstalled the Nvidia dependencies installed the AMD dependencies Because I have a AMD CPU not a GPU, after doing all of this in the web UI I was able to Give a prompt press create and it showed the progress bar of it creating the image then the whole computer shut down LOL I have a 6 month old computer that's got pretty good specs but not a GPU am I just in a dream world that I could ever make images on my computer? There's nothing else I can think of to do. Do you have any suggestions I tried to give you all of the info I could so you would be on the same page I guess
@@TheGeneticHouse yes, please do so. Looking forward to your creative results 🙂
@@TheGeneticHouse Also take a look here. I want one step further and did something with AI in videos. I demonstrated here what can be done already. If you like it, I am happy to make a tutorial about it: ua-cam.com/video/Jd_Zc_a7SsE/v-deo.htmlsi=w7cFeEMwXHCwDLMa
Can this run on a MacBook Pro?
Hello, you can install Fooocus on Apple Mac silicon (M1 or M2) with macOS 'Catalina' or a newer version.
Take that link here and scroll down until you come to the MAC section. Installation instructions can be found there. Have fun --> github.com/lllyasviel/Fooocus
Thank you very much- I’ll give a go!
@@scotharrell9223 my pleasure
When i click "generate" it says "error". what did i made wrong?
Hello, difficult to say, might be that you do not have enough VRAM, that you have no checkpoint in place, etc. check the CMD tool for more details. What does it say when hitting generate? (Just in case, cmd is the console that opened when you double-clicked the run.bat file)
@@Axel.Petersyeah i just have a hp laptop i think that could be the problem
@@Axel.Peterstanks for the response
why is not face swap working?
Hello, what do you mean it's not working? I tried it just as we speak, and it works for me. You can generate any image with the face of your desire…
1. Activate 'Input Image'
2. Head over to the 'Image Prompt' tab
3. Activate 'advanced' on the bottom left
4. Click on 'FaceSwap', drag and drop an image of a face, and…
5. Hit 'Generate' - Done
How do I add more models like stable diffusion to Foocus?
Just drop them into your checkpoint folder inside Foocus and in the advanced settings on the Fooocus frontend(Webbrowser) choose a different one. You could also change the settings inside the txt file in the main Fooocus folder to your stable diffusion checkpoints. Fooocus will then search in your specific folder for checkpoints, but still in both ways, you have to select them manually via the advanced tab.
Hey @PixelPioneers I instantly get "connection failed" when im on the ip adress website. How do I fix this? hhelp would be appreciated
Hello, mhm… does the website starts alone after double clicking the bat file or do you try to enter the IP yourself into your preferred Webbrowser? If it does not open automatically, that’s already the first indicator that the service is not running. What is your command tool saying, what is the message?
Hey, I'm not typing it in myself, when running the bat file it sends me to it. so that part goes well. my command tool doesnt say anything is wrong either. it does tell me that it's trying to enable lowram mode. but I don't think that's the issue@@Axel.Peters
I am exploring Fooocus and it is amazing, the results are great! However my "input image" doesn't work. Any idea how I can activate it? I check the box but nothing happens :-( Thanks! Great video :-)
That is strange, when you say activated, what do you mean precisely? Nothing opens where you could enter an image, or your image is not being taken, or ignored, etc? Let me know, I am sure we come to a good end here
Thank you! Nop, nothing happens when check the input image box...I read the documentation page and search on the internet but can't find anything...I am using windows...perhaps something I should run on PowerShell? Weird, thanks a lot anyway
@@GustavoGarciaSantaCruzI just started fooocus but I personally drag and drop everything
Hello. How to put an outfit shot in PNG format without a model on the model? Can you make a video about this? :(
Hello, I am trying to follow but uncertain about your request. Would you mind explaining it to me once more?
@@Axel.Peters I want to do it like in the video below, but I couldn't understand it because it was in Japanese:( Could you please make a video about it?
ua-cam.com/video/2eZqJLG1dpU/v-deo.htmlsi=u-IBWuhqAVntHNPz
hey it is enabling low vram mode and uses on 2 gb of vram of the gpu i have 4gb vram rtx 3050 in my laptop how to set it to normal vram mode and any tips on performance optimization as advance features lipe face swapping and input image donot work like the terminal says the the cusa is out of memory ??
Hi, you could try the following. Go to the bat file (the one you use to start Fooocus). Edit it with a simple text tool. You should see something like this: .\python_embeded\python.exe -s Fooocus\entry_with_update.py
You could add behind that one of the following things:
1. --highvram
2. --normalvram
3. --lowvram
4. --cpu
5. --gpu-only
@@Axel.Peters .\python_embeded\python.exe -s Fooocus\entry_with_update.py
--normalvram
pause
like this??
@@devanshumangal07 yes, exactly
do u speed the generation time or is it real time, i have a gtx 1660ti 6gb and it takes quite some time. i know this process is based on the graphics card but does it affect the time this much
Hello, yes I did speeding for the benefit of the viewer. Indeed, GPU power is making the difference. I wish my answer would be a different one but your limits are related to your GPU compute power and available VRAM. You can also use low/ mid-end cards but have to live with the time it takes to render. But, you can reduce steps, switch to SD1.5 checkpoints (models). Happy generating
@@Axel.Peters thank u for the response that was helpful, have a nice day
@@dudleyboy5327 same to you. Always happy to help… enjoy
bonjour et merveilleuse année pensez vous que c est possible de crée un model et lui faire porter une robe venant d une photo ? en utilisant l'invite image ou autre ? merci
L'intelligence artificielle a des limites, mais selon le degré de précision nécessaire, je pense que cela peut être fait. Vous pourriez utiliser votre photo avec la robe et simplement échanger les visages pour remplacer la tête existante par celle que vous avez générée. Avec openpose, vous pourriez changer la position générale de la personne. Mais comme dit, cela a des limites.
@@Axel.Peters merci j ai vu ce que je voulais sur khora
mais c est long et faut tout installer....
I have 16GB ram, unfortunately never upgraded after GTX 970, its very slow, feels like I need at least 64 GB ram and 12GB vram
@@xqc20y what is kaggle?
The minimal requirement is 4 GB Nvidia GPU memory (4 GB VRAM) and 8 GB system memory (8 GB RAM). But the min requirements don't make fun anymore. So, a good GPU with 6 GB+ VRAM and 32 GB of System RAM would simplify things a bit. Otherwise, you have online services, but then you are again in a subscription mode and have to pay for the usage of cloud computation.
@@Axel.Peters agree, i don’t think it makes sense to invest at this time, rather stick to ChatGPT or mid journey, Dall e
@@Axel.Peters how would you connect the cloud service to Fooocus?
1. www.fal.ai/models/fooocus
2. aws.amazon.com/marketplace/pp/prodview-7rd6v2ujpn54e
Disclaimer, I don't know how good they are or if the offering is correct and fair. Otherwise, what I mean before was more related to the usual suspects. You can pay for DALL-E, Midjourney, etc.
How do you delete the history log😂????
Hello, you get a log file stored in every of your output folder. Just delete the log file there and done 🤣… nobody will see what you created 😉
@@Axel.Peters How do I access that output folder? Also thank you so much for the quick answer😂
@@Exscape_Reality easy, just go to your main Fooocus folder. Inside you will find an output folder… inside there folders separated by dates with all generated images per day. In these folders is the log file.
@@Axel.Peters Thank you!
@@Exscape_Reality pleasure…
Your computer must be very fast lol. This program is good but it's slow and can't use my pc at all whilst it's making pics
You indeed need to have a rather strong GPU with higher VRAM. But you can limit the amount of VRAM used by FOOOCUS to keep something for playing around while generating images…
its work with intel arc??
Hello, its supposed to work but is not extensivly tested. here are the steps to do so.
1. Install the conda package manager and pytorch nightly
2. Open the macOS Terminal app and clone this repository with: git clone github.com/lllyasviel/Fooocus.git
3. Change to the new Fooocus directory: cd Fooocus
4. Create a new conda environment: conda env create -f environment.yaml
5. Activate your new conda environment: conda activate fooocus
6. Install the packages required by Fooocus: pip install -r requirements_versions.txt
7. Launch Fooocus by running: python entry_with_update.py (Some Mac M2 users may need python entry_with_update.py --enable-smart-memory to speed up model loading/unloading.)
More info here: github.com/lllyasviel/Fooocus
Sorry, I miss-read your question and thought it was about MAC. My bad… I am not sure about Intel Arc. Sorry.
Fooocus is Awesome, but I want to know how can I make my Image turn into any style I want (For Eg. Neonpunk) without changing anything in the Image, it should only act as a filter on my Original image without changing the face, body, etc?
two options I figured out so far.
1. Drag your own image into the bottom canvas, under advanced into the upscaling window. Upscale with 1.5 or 2 and your image will appear right after in the main canvas, and you can apply to it whatever you want.
2. Drag your image into the inpainting canvas and paint a very little, almost invisible point somewhere in the image. Then say generate, and it will appear in the main canvas, and you can do whatever you want.
Let me know if you understood my explanation.
@@Axel.Peters I tried them both but none of them worked
1) As you mentioned, I upscaled the Image , then changed the style as per my liking and left the prompt Tab blank and hit generate, it generated images but not at all related to my image, it were Random images from that style.
2) Did the Same as you Mentioned in Inpaint, created a Tiny dot, selected a style and hit generate, and it generated my images same as Original nothing changed.
If you could find a way, please try to make a video!
@@inspiredactions19 now I understand what you want. Drag and drop an image of your liking to the advanced image prompt tap, then activate PyraCanny, change the weight slider all above to the right (2) and change the style to your liking, for example, neon punk. Make sure you enter a resolution/ aspect ratio that is similar to your current image and hit generate. Done, similar image different style 😀 I just tried it and it definitely works. Please be so kind of let me know if that was helpful. If you want that neon punk image, even closer to your image, you can also take the ‘stop at’ slider and push it all the way up to the right (1)…
@@inspiredactions19 And in case you have Photoshop at hand and the generative fill functionality activated, you can watch another Video from me about different styles and how to apply them to your images. But, as said, it would require PSD. ua-cam.com/video/nKl5p6kLDFc/v-deo.htmlsi=4ptQP9XKkMdECP7m
@@Axel.Peters got the results but the style overpowers the image.
what about swapping a face of a person A into a person B? i would love to swap my dads face into some moern style lol
Hello, sure. Easy..
You can generate any image with the face of your desire…
1. Activate 'Input Image' (bottom left)
2. Head over to the 'Image Prompt' tab
3. Activate 'advanced' on the bottom left
4. Click on 'FaceSwap', drag and drop an image of a face (in your case, your dad)
5. Prompt an image (whatever you want)
6. Hit 'Generate' - Done 😀
i know> but what if i want to to put my dads face into a specific image of a man model for example and not a random photo@@Axel.Peters
Another thing ive seen is that fooocus really struggle when it comes to using a pyracanny from image A and faceswap from image B> although im running a 4070ti with 64gb of ram
@@holetarget4925 you could take an existing picture of a man model, drag it into the image prompt but use for the first slot PyraCanny (to take the pose of the model), in the second slot your dad‘s photo and let AI do something that comes at least close to the original picture. Help guide the AI by saying what you want. Otherwise there are free online services that do exactly what you have ask for.
@@holetarget4925 mhm, I have never experienced that so far. For me it’s always blazing fast. Sorry to hear that you experience performance issues.
Pyra created Antoine Griezmann
1. Before you tell people how to download a program, you MUST tell them the minimum requirements for the hardware they must have to use it. Otherwise, many people will simply waste time and nervous energy. You need to think about all viewers of your videos, not just those lucky ones who have new video cards with 4G Nvidia or 8G AMD memory.😑
2. When you show the process of creating images, you MUST mention that you are using fast playback - don't give users the false impression that the images were created within a second.😑
Thank you for your valuable feedback. My aim is always to produce informative videos that keep viewers abreast of the latest technology and trends. This particular video is available at no cost, and I regret that I overlooked the two key points you mentioned. I am more than willing to incorporate your suggestions, but I also kindly ask for your understanding and empathy towards the effort that goes into creating this content and your way of approaching what you think is your right for free content.
Only thing I don't like about Fooocus is, everytime I hit generate it says "loading the model to GPU" and waste few seconds to generate images. Does anyone else facing the same issue? any fix for that? (I have a RTX 3060 12GB VRAM and 16GB RAM)
Hi there, there are indeed a couple of seconds, but really just a couple. I don't found this very noticeable. On the other side, you don't spend hours in creating convincing prompts and feeling right after like a real prompt engineer. Fooocus has a GPT2 conversion model in place. You keep it simple, and the results are always convincing for me. On top, plain and simple style guides and everything else you need to get something quickly with top-notch quality. 🙂
So expensive price is the problem
Hello, what do you mean? The software is free 😉…
error
Hello, error what? Installation, usage, handling…
OMG is this free?
Yes, of course 😃… go ahead and have fun.
@@Axel.Peters I am just baffled. thank you so much for this video :)
@@froschderergerer7877 my pleasure. If you need help - just let me know. ☺️
It almost burned my cpu .. keep using A1111 better
What did you do? ☺️
How did you get it to work with a CPU? I have a AMD CPU I do not have a GPU
Fooocus MRE stills better
i am having an error "Moving models to GPU" my system specs
GTX 1650 4gb
ram- 8gb 3200mhz
@axel.Peters help me in fixing it pleaseee
Hello, you seem to have minimum system requirements, but might really be related to both VRAM and System RAM. But it’s a guess only.
Try to edit run.bat file (the one you normally doubleclick to start fooocus) adapt the first line: \python_embeded\python.exe -s Fooocus\entry_with_update.py
You could add behind one of the following things:
1. --highvram
2. --normalvram
3. --lowvram
4. --cpu
5. --gpu-only
In your case use (3) or (4)
@@Axel.Peters Thanks bud, -lowvram it is
Whenever I try running the bat file. I get the following error Fooocus>. \python.exe -s Fooocus\entry_with_update.py The system cannot find the path specified. Anyone else get this?
Hello, that really should not happen. Fooocus, when complete, should have standalone python integrated and the path should work without the need for an offline and local python installation. Have you downloaded everything correctly, and is the fooocus folder still complete (incl. python embedded)?
I've sorted that issue now, but what if you don't have an NVIDIA GPU? Can you not run it at all?@@Axel.Peters
@@lynxtouchhello, yes you can use it with AMD cards as well. Let me know if you need help going forward.
@@Flash_Dogs I think my Dell XPS 15 has an AMD card but let me double check
you’re curious about whether you can use Fooocus AI with an AMD GPU and how to go about it. After looking into this, it seems that there are significant compatibility issues when trying to run Fooocus on systems with AMD GPUs. This is primarily because Fooocus AI relies heavily on CUDA, a technology developed by NVIDIA, which isn’t supported on AMD hardware.
One user on GitHub noted that despite installing torch from the AMD package index, Fooocus AI still attempted to use CUDA, leading to errors and incompatibility . Unfortunately, the Fooocus AI GitHub repository doesn’t offer specific guidance or solutions for AMD GPU users .
Given these challenges, it appears that currently, there isn’t a straightforward method to run Fooocus AI effectively on an AMD GPU. If you’re keen on using Fooocus AI, you might need to do so on a system with an NVIDIA GPU, or consider other AI software that’s compatible with AMD hardware.
I understand this might not be the news you were hoping for, but I hope it helps clarify the situation. If there are any updates or workarounds in the future, the Fooocus AI GitHub repository is a good place to keep an eye on.