If you find the video useful and would like to tip you can buy me some electricity for all these image generations. It is greatly appreciated! ko-fi.com/kleebztech
Bro i was trying so hard to archieve a descent result with Stable Diffusion, but Fooocus just works out of the box for this, thank you so much for sharing
Many tnx. I'm so many years worked with PS. I will try find time to compare this method. I'm interesting if there are place for improve photos from smartphone and don't take a big camera to trip.
Thank you as this absolutely works, I've been using Leonardo AI to remove things, but was trying to figure out a way to do it within Fooocus for free offline. This helped as you basically enter in whatever is in the background to get rid of the object. Like this works great too for people wanting to get rid of longer hair for a shorter cut. You can enter in the shirt and the background image and it cuts it out.
A gem, az usual. Thank you. But what if I want to use a Ford picture instead of text prompt? I mean can I merge any additional element, character, object into an existing background? Let's say, I have a walking cartoon character and I want it to put on in a forest as it walks? Do you have a video about it? Thank you.
As always great tutorial. The good thing about FOOOCUS is the easy way to inpaint any picture, I find it better than A1111 and easier to use. I was making mistakes as you mentioned here, when I want to delete something I type "erase or delete" 😂 and it was driving me crazy.
Yeah I really try to avoid the others. A1111 has gotten so crowded. Comfy I never had a huge issue with. Personally I think everybody should have Fooocus as one of their tools.
A1111 is going to be more capable if you know how to use everything. They are both using SD but Fooocus does do some additional tweaks in the background. In my opinion the inpainting in Fooocus is the best and the reason I started using it. Then I realized I could do most of what I needed with it instead of A1111 or ComfyUI. And since it just runs from its own folder and no need to install stuff I see no reason not to use it when I can. Just extract and run. Plus it can be configured so you can share the same models between the different UIs. For SD 1.5 it is probably best to use the others since Fooocus only supports SD 1.5 as refiner which allows you to use 1.5 LoRAs.
Thanks for your great video. I am searching for a method to put a person from one existing picture into another existing picture. do you have any suggestion how this works?
Would most likely need to do that in a traditional photo editing app. You could use Fooocus to help with smoothing things out after editing with something else using a low denoising strength around the edited section for example but the actual editing would need to be another program.
I'm super stoked and bummed out at the same time. Stoked about this awesome video on a topic I really needed, but bummed 'cause I can't seem to get it to work following the recommendations. No matter what-whether it's against a wall, in a room, or any other spot-I need a person in the frame, and nothing's working. It just keeps redrawing what's already there on a loop, no matter the settings or models. It's frustrating as hell
You have a photo you are trying to generate an AI person into and it won't generate a person? You are masking an area and putting the persons description into the prompt and just gives you the exact same thing or similar?
@@KLEEBZTECH Yeah, I've got this photo where I'm trying all sorts of stuff-masking, tweaking settings, turning off styles and random, switching models-but it just keeps redrawing the same area. And even though it can remove objects like you showed in your tutorial, it can't seem to draw anything in at all. The only thing that kinda works is anime, but I need realism. Your channel's been a real find for me.
You do want to make sure you use the random seed because otherwise it will just keep doing the same thing. Usually most people have issues with removing things and not having something else put into the spot that is masked. Have you messed with any other settings in the debug section that might be causing an issue? Not sure if you have tried shutting down Fooocus and restarting to reset anything you may have changed and trying again.
Have you tried both regular inpaint and the modify content option? Depending on the image and what you are doing that can be the better option. And then just have the description of what you are adding. For example I just masked a section of an image of a woman sitting on a bed and masked the area next to her and chose Modify Content method in the dropdown menu and put "baby sitting on bed" in the additional prompt with nothing in the normal prompt and it added a baby.
@@KLEEBZTECH Started working in inpaint with the developer debug mode feature enabled. Outpaint only works with enabled styles. Thank you for your help!
I do have other inpainting videos but not that specifically. Are you trying to just generate a person in a background that you already have? Or are you trying to add a specific person to a specific background?
@@KLEEBZTECH Hello, thanks for your reply :) I am trying to add a person to a specific background. I'm a photographer and my client would like a picture with the eiffel tower in the background so I was playing around with the idea 😅 I'm not sure whether it's possible to attain realistic results with it though. What are your thoughts?
If working with two images that already exist then it is going to be something that needs to be done in a photo editing tool like Photoshop. If you were taking a person and wanted to generate a background that will work or if trying to generate a generated person on an existing background.
If you mean transferring part of one image into another then not really something that AI would be good for and better to use traditional photo editor. You could of course use Fooocus to clean up the edited image.
@@KLEEBZTECH Thanks a lot for your quick and clear answer , just one more . how could I inpaint something blurred in a photo , e.g. a ghost ( blurred ) dark and a little bit transparent in a old building , thanks
Mask the area you want the ghost and then in the prompt you will have to experiment a little but maybe try for example "a ghostly figure of a woman, transparent, blurry" and then open the debug mode and in the inpaint tab there is the option for denoising strength. Set that down a bit to maybe about .5 to start. You may need to adjust that more. I have not really tried this exactly but that would be the approach I would take to start. By setting the denoising to a lower amount it will retain some of the original image.
Is there anyway to include an image object instead of generating an object using inpaint? For eg. getting my own dog into the beach image. And also, a transparent bg with an object and generate it's background?
You could take an image of your dog and using the replace background option in inpainting. So technically for either one you could use that option. But putting your dog on another actual photo of a beach would not really be possible. Although you might be able to use it to blend the things in better after editing in other software.
What if you want to use an image you already have? like an object or specific character that you designed and you want to place it in the image, or change one of those people's clothes without affecting the others?
@@KLEEBZTECH I know with fooocus it may be a stretch, but this guy does it almost easily in comfyUi with probably very new tools, th-cam.com/video/HE9aC8hp3VQ/w-d-xo.htmlsi=4l9IOe9FS7ZS8miz.
As the primary model you must use SDXL. You can use a 1.5 model as a refiner which does allow for 1.5 LoRAs to be used. civitai.com/ is a good place, And if using with just SDXL make sure to grab the ones for SDXL.
So SDXL as primary model and a 1.5 for Refiner. And for LoRA 1,...,4? It is better to use only a primary + refiner models? I tried and now I'm in the experiment phase, but I can't see correlation and "rules of thumbs" for models combinations. Thank you for your support. 🎉
You can use any model as a refiner. But SDXL must be the primary model. Not sure if that was your question. When you choose a refiner you will see some more info below for instructions on where to adjust the refiner switch as recommended by Fooocus. I have not really experimented much with using 1.5 stuff in Fooocus.
This might be more of what you are looking for. This is more towards anime but the same idea could be used for other styles. You will want to use only the styles that are geared towards that like maybe line drawing or pencil sketch drawing. You may need to force it more that I did with anime since the anime model is geared more towards anime which makes that easier. th-cam.com/video/PuC2AVrQXTs/w-d-xo.html
I was just thinking about making videos analyzing the styles. Nothing super scientific or anything. There are many things in these styles that may impact the results.
I do not yet. I may at some point but I still have a bunch more planned for Fooocus since it is my primary tool at this time. I'd also need to spend much more time with Forge than I have.
So you have a hand with a ring on it and want to remove? what is the prompt you are using? any styles? I would normally mask the ring and prompt for a hand I guess. Hard to say not knowing the exact image. But do not put anything about "no ring" in the prompt because that will cause you to get one.
@KLEEBZTECH basically yes. But everything I do just wants to regenerate into a different ring. I don't use ring in the prompt. I try everything from bare skin to smooth skin to detailed hand, and fingers etc. Still just prints more rings
Thank you very much for this video. I learned things I always wondered about I have a question, please. I have a cartoon image and I would like to modify it so that I delete the face of a boy and put the face of another boy that I have. Can I do that? ❤❤
Should be able to. Of course AI does not always like to cooperate. But using faceswap and inpainting you should be able to. I do have some videos on those subjects. You will need to enable Inpainting and image prompt in the debug menu. Might need to adjust styles.
@@KLEEBZTECH Thank you actually I use this program bass Google Collab and I cannot download it on my medium-powered computer I will search for the videos you are talking about and watch them ❤️❤️
@@KLEEBZTECH I have a photo with 3 persons on the beach, pretty much like the photo in your tutorial, I brushed an area and put the prompt like: a man, facing the camera, standing on the beach. It does not generate the man but change the beach a bit only. Using Juggernaut XL 8/9 and everything else same as your settings
My wedding photos don't have my wifes wedding ring in them, and I want to add it in. How can I "crop" my wifes ring from one photo, and get it onto our wedding photo?
That would not be easy with these sorts of tools. You could use a mix of image prompts and inpainting to get something that may look similar. But that might not work great and would be difficult to explain... But I will think about it. Off to bed at the moment.
Excuse me: I just want to simply remove an object in the picture, for example, there is a cat on the ground or on the table. I don’t know how to describe the background environment. I don’t even know where the cat is. It may be very complicated. It may be a very simple background environment. I want to remove the cat. How should I write the prompt word? Every time I remove the cat, that position will be replaced by another new object. This is not what I want!
If you find the video useful and would like to tip you can buy me some electricity for all these image generations. It is greatly appreciated! ko-fi.com/kleebztech
Bro i was trying so hard to archieve a descent result with Stable Diffusion, but Fooocus just works out of the box for this, thank you so much for sharing
The styles are just prompts? this is real helpful for me, thx
More info th-cam.com/video/Sr2yXOSexy0/w-d-xo.html
Many tnx. I'm so many years worked with PS. I will try find time to compare this method. I'm interesting if there are place for improve photos from smartphone and don't take a big camera to trip.
Thank you as this absolutely works, I've been using Leonardo AI to remove things, but was trying to figure out a way to do it within Fooocus for free offline. This helped as you basically enter in whatever is in the background to get rid of the object. Like this works great too for people wanting to get rid of longer hair for a shorter cut. You can enter in the shirt and the background image and it cuts it out.
You're welcome! No need to be wasting money.
A gem, az usual. Thank you. But what if I want to use a Ford picture instead of text prompt? I mean can I merge any additional element, character, object into an existing background? Let's say, I have a walking cartoon character and I want it to put on in a forest as it walks? Do you have a video about it? Thank you.
Thanks a lot ! You save my day !
As always great tutorial.
The good thing about FOOOCUS is the easy way to inpaint any picture, I find it better than A1111 and easier to use.
I was making mistakes as you mentioned here, when I want to delete something I type "erase or delete" 😂 and it was driving me crazy.
Yeah I really try to avoid the others. A1111 has gotten so crowded. Comfy I never had a huge issue with. Personally I think everybody should have Fooocus as one of their tools.
I am using A1111. Is fooocus much different from A1111 in term of capabilities or is it just a UI over SD?
especially for inpainting/outpainting.
A1111 is going to be more capable if you know how to use everything. They are both using SD but Fooocus does do some additional tweaks in the background. In my opinion the inpainting in Fooocus is the best and the reason I started using it. Then I realized I could do most of what I needed with it instead of A1111 or ComfyUI. And since it just runs from its own folder and no need to install stuff I see no reason not to use it when I can. Just extract and run. Plus it can be configured so you can share the same models between the different UIs. For SD 1.5 it is probably best to use the others since Fooocus only supports SD 1.5 as refiner which allows you to use 1.5 LoRAs.
Thanks for your great video. I am searching for a method to put a person from one existing picture into another existing picture. do you have any suggestion how this works?
Would most likely need to do that in a traditional photo editing app. You could use Fooocus to help with smoothing things out after editing with something else using a low denoising strength around the edited section for example but the actual editing would need to be another program.
Thank You!
You're welcome as always.
I'm super stoked and bummed out at the same time. Stoked about this awesome video on a topic I really needed, but bummed 'cause I can't seem to get it to work following the recommendations. No matter what-whether it's against a wall, in a room, or any other spot-I need a person in the frame, and nothing's working. It just keeps redrawing what's already there on a loop, no matter the settings or models. It's frustrating as hell
You have a photo you are trying to generate an AI person into and it won't generate a person? You are masking an area and putting the persons description into the prompt and just gives you the exact same thing or similar?
@@KLEEBZTECH Yeah, I've got this photo where I'm trying all sorts of stuff-masking, tweaking settings, turning off styles and random, switching models-but it just keeps redrawing the same area. And even though it can remove objects like you showed in your tutorial, it can't seem to draw anything in at all. The only thing that kinda works is anime, but I need realism. Your channel's been a real find for me.
You do want to make sure you use the random seed because otherwise it will just keep doing the same thing. Usually most people have issues with removing things and not having something else put into the spot that is masked. Have you messed with any other settings in the debug section that might be causing an issue? Not sure if you have tried shutting down Fooocus and restarting to reset anything you may have changed and trying again.
Have you tried both regular inpaint and the modify content option? Depending on the image and what you are doing that can be the better option. And then just have the description of what you are adding. For example I just masked a section of an image of a woman sitting on a bed and masked the area next to her and chose Modify Content method in the dropdown menu and put "baby sitting on bed" in the additional prompt with nothing in the normal prompt and it added a baby.
@@KLEEBZTECH Started working in inpaint with the developer debug mode feature enabled. Outpaint only works with enabled styles. Thank you for your help!
I love your work, thanks for the great video! Do you have any videos on how I can add a person to a background using Fooocus?
I do have other inpainting videos but not that specifically. Are you trying to just generate a person in a background that you already have? Or are you trying to add a specific person to a specific background?
@@KLEEBZTECH Hello, thanks for your reply :) I am trying to add a person to a specific background. I'm a photographer and my client would like a picture with the eiffel tower in the background so I was playing around with the idea 😅 I'm not sure whether it's possible to attain realistic results with it though. What are your thoughts?
If working with two images that already exist then it is going to be something that needs to be done in a photo editing tool like Photoshop. If you were taking a person and wanted to generate a background that will work or if trying to generate a generated person on an existing background.
@@KLEEBZTECH got it - thanks again!
Absolute thank you!
Hi , would be possible impaint a real photo into another real photo both uploaded by you . Thanks
If you mean transferring part of one image into another then not really something that AI would be good for and better to use traditional photo editor. You could of course use Fooocus to clean up the edited image.
@@KLEEBZTECH Thanks a lot for your quick and clear answer , just one more . how could I inpaint something blurred in a photo , e.g. a ghost ( blurred ) dark and a little bit transparent in a old building , thanks
Mask the area you want the ghost and then in the prompt you will have to experiment a little but maybe try for example "a ghostly figure of a woman, transparent, blurry" and then open the debug mode and in the inpaint tab there is the option for denoising strength. Set that down a bit to maybe about .5 to start. You may need to adjust that more. I have not really tried this exactly but that would be the approach I would take to start. By setting the denoising to a lower amount it will retain some of the original image.
I'm trying to add a person but it keeps giving me empty background
Hard to say what the issue may be but have you tried masking a larger area? What method are you using that has the issues?
Is there anyway to include an image object instead of generating an object using inpaint? For eg. getting my own dog into the beach image. And also, a transparent bg with an object and generate it's background?
You could take an image of your dog and using the replace background option in inpainting. So technically for either one you could use that option. But putting your dog on another actual photo of a beach would not really be possible. Although you might be able to use it to blend the things in better after editing in other software.
th-cam.com/video/fARdZK2fCPo/w-d-xo.html
@@KLEEBZTECH got it! Will try inpaint properly this time. Thanks for this!
Thanks. Can't I use image prompt(cpds,faceswap,pyracanny) and inpaint at the same time on foocus?
You can enable image prompt with inpainting in the debug tools.
@@KLEEBZTECH I can't see the inpainting in the debug tools? I copy yours, but I'm a beginner
Once you check off Developer Debug Tools you should see 4 new tabs. Under the Control tab you should find "Mixing Image Prompt and Inpaint"
What if you want to use an image you already have? like an object or specific character that you designed and you want to place it in the image, or change one of those people's clothes without affecting the others?
You can't really merge two images in that way. If you want to change the clothing you could just mask that and generate new clothing.
@@KLEEBZTECH I know with fooocus it may be a stretch, but this guy does it almost easily in comfyUi with probably very new tools, th-cam.com/video/HE9aC8hp3VQ/w-d-xo.htmlsi=4l9IOe9FS7ZS8miz.
Those are the types of circumstances that ComfyUI is nice. I think many of these tools will come to Fooocus eventually in some way.
Howesome tips! i follow u!
What kind of models can i use in fooocus? Only SDXL LoRA? Where I can find a full list? Your hel will be appreciate.
As the primary model you must use SDXL. You can use a 1.5 model as a refiner which does allow for 1.5 LoRAs to be used. civitai.com/ is a good place, And if using with just SDXL make sure to grab the ones for SDXL.
So SDXL as primary model and a 1.5 for Refiner. And for LoRA 1,...,4? It is better to use only a primary + refiner models? I tried and now I'm in the experiment phase, but I can't see correlation and "rules of thumbs" for models combinations. Thank you for your support. 🎉
You can use any model as a refiner. But SDXL must be the primary model. Not sure if that was your question. When you choose a refiner you will see some more info below for instructions on where to adjust the refiner switch as recommended by Fooocus. I have not really experimented much with using 1.5 stuff in Fooocus.
I can't get a design as png?
Can this be used to remove color make photo like drawing
This might be more of what you are looking for. This is more towards anime but the same idea could be used for other styles. You will want to use only the styles that are geared towards that like maybe line drawing or pencil sketch drawing. You may need to force it more that I did with anime since the anime model is geared more towards anime which makes that easier. th-cam.com/video/PuC2AVrQXTs/w-d-xo.html
I just tried it with "pencil sketch drawing" style and in my prompt I said black and white pencil drawing and it worked pretty well.
This answers why I kept getting keys in the image: SAI Anime style contains 'key visual' in the prompt!
I was just thinking about making videos analyzing the styles. Nothing super scientific or anything. There are many things in these styles that may impact the results.
Fooocus is very slow for my PC, I use "forge", do you have "forge" tutorials?
I do not yet. I may at some point but I still have a bunch more planned for Fooocus since it is my primary tool at this time. I'd also need to spend much more time with Forge than I have.
Have you tried using any of the lightning models? I haven't dive too much into those yet and still have to figure out the best settings.
I'm having trouble using inpaint to remove a ring. Can you help?
So you have a hand with a ring on it and want to remove? what is the prompt you are using? any styles? I would normally mask the ring and prompt for a hand I guess. Hard to say not knowing the exact image. But do not put anything about "no ring" in the prompt because that will cause you to get one.
@KLEEBZTECH basically yes. But everything I do just wants to regenerate into a different ring. I don't use ring in the prompt. I try everything from bare skin to smooth skin to detailed hand, and fingers etc. Still just prints more rings
Any styles enabled? Masterpiece style for example includes jewelry in the prompt.
Thank you very much for this video. I learned things I always wondered about I have a question, please. I have a cartoon image and I would like to modify it so that I delete the face of a boy and put the face of another boy that I have. Can I do that? ❤❤
Should be able to. Of course AI does not always like to cooperate. But using faceswap and inpainting you should be able to. I do have some videos on those subjects. You will need to enable Inpainting and image prompt in the debug menu. Might need to adjust styles.
@@KLEEBZTECH Thank you actually I use this program bass Google Collab and I cannot download it on my medium-powered computer I will search for the videos you are talking about and watch them ❤️❤️
11:06 😂😂😂 gotch ya! UFOS are REAL!!!
I followed the video and being able to add anything but a human to a image, any ideas why?
Hard to say without a little more information on what you are doing and having difficulty with.
@@KLEEBZTECH I have a photo with 3 persons on the beach, pretty much like the photo in your tutorial, I brushed an area and put the prompt like: a man, facing the camera, standing on the beach. It does not generate the man but change the beach a bit only. Using Juggernaut XL 8/9 and everything else same as your settings
@@RamenZhu did you use the modify content option? If not it can be difficult since there are already people in the image.
@@KLEEBZTECH Got it!, tried the modify content and it works, thank you!
My wedding photos don't have my wifes wedding ring in them, and I want to add it in. How can I "crop" my wifes ring from one photo, and get it onto our wedding photo?
That would not be easy with these sorts of tools. You could use a mix of image prompts and inpainting to get something that may look similar. But that might not work great and would be difficult to explain... But I will think about it. Off to bed at the moment.
👍👍👍👍
inpaint most hardest task to complete is fixing fingers and deformed legs or hands, almost impossible 😑
Time and patience. And if you can't generate fast it takes much more of each.
I only have 4gb... i only can watch videos how others can play with this 😂😂.. i only can make 10 images each day.. 😢
Why the hell would earrings be in the masterpiece style. I would just remove that from the style.
Yeah, I have been tweaking my own styles for that very reason. I do think some of that might get changed in the default soon.
Excuse me:
I just want to simply remove an object in the picture, for example, there is a cat on the ground or on the table. I don’t know how to describe the background environment. I don’t even know where the cat is. It may be very complicated. It may be a very simple background environment. I want to remove the cat. How should I write the prompt word? Every time I remove the cat, that position will be replaced by another new object. This is not what I want!
You can try things like empty table and other things to imply there is nothing else. But don't put things like "no cat" because that will not work.
5:58 bro is one piece fan 💀
Maybe, but Fooocus doesn't know anythg about shadow on the sand)
Fire the programmers
Goofy