Wonderful, just what was needed. A1111 v 1.7 has messed up so many things from the prior 1.6 version when running in Stability Matrix, that I'm ready to give up on it, now that you have shown how to get things working in Fooocus, which was already far better for In and Outpainting. Thanks a Lot ! ! ! !
This was helpful. Fooocus looks different now though and I cannot locate the face swap feature. There is no advanced check box anymore and now it shows a "hurt me plenty" check box. Just looks different. Today is July 6th and I'm running the latest version 1.52.0. Can you make an updated video? :)
nice video but, this isn't working for me. It's still an obviously different face then my reference face image. I wish they made it easy to just reference the face exactly as the referenced image...I wouldn't think it would be hard to do but 🤷♂
Thank you so much for this. It's mind blowing. Quick question for you... I'm trying to generate an image of me holding my company's product, which is a tea box with the brand and logo on it. How do I generate an image such an image? I know I can use the face swap for my face but for the object, I really have no clue how to do it. I know I can use the image prompt for the product but the images generated won't be have my exact product. Appreciate your help and thank you so much again!
This is Awesome! very very helpful, does this work for other objects as well? I'm trying to add inanimate objects to an already established image, any tips?
Good question! I imagine not using the advanced FaceSwap checkbox in the Image prompt. Crop the image prompt down to just the object. Then I'd say the inpaint will be a good option. Alternatively using the variation with a descriptive text prompt of the scene, including the name of the object in the image prompt.
Hi, is it possible that something has changed since the last update? I've noticed that "Mixing Image Prompt and Inpaint" no longer works. The result is largely the same as the original image. Does anyone know more?
Commented this elsewhere, I probably should have explained in the video :,) "The stop at level is the percentage of how "complete" the face swap is before it stops. 1 = 100%, 0.9 = 9% etc. A lower stop at should in theory be a lower likeness, but it's kinda hard to know exactly and different per face/reference image. Try playing with different levels, 0.9 will be a good default."
HI! Great tutorial. I'm using it, but both by changing PC and internet connection, it remains slow in creation (even though I put speed) but the most annoying thing is that it often gives an ERROR, when creating the prompt, a red warning. Can you help me understand why?
Commented this elsewhere, I probably should have explained in the video :,) "The stop at level is the percentage of how "complete" the face swap is before it stops. 1 = 100%, 0.9 = 9% etc. A lower stop at should in theory be a lower likeness, but it's kinda hard to know exactly and different per face/reference image. Try playing with different levels, 0.9 will be a good default."
I probably should have explained in the video :D The stop at level is the percentage of how "complete" the face swap is before it stops. 1 = 100%, 0.9 = 9% etc. A lower stop at should in theory be a lower likeness, but it's kinda hard to know exactly and different per face/reference image. Try playing with different levels, 0.9 will be a good default.
Which method did you try? I find that the variation provides a more credible overall image, but the inpaint doesn't alter the rest of the image. Is your reference face cropped? I found that providing a tighter crop on there helped. You can also do a second run, using the first pass as your reference image and keeping the same face swap.
I have problems with both methods, and I also tried to use a face swap and a prompt genration in the same time. For me both methods gave mediocre results and I don't know what I'm doing wrong@@k5c5
@@k5c5 instead of using pre-croped faces, fooocus also allows for the selection of the promted images' area (i.e. the face) in the image promt directly via the pen button.
I keep getting the same womans face, half way through the generation the fuzzy preview looks like it is heading towards the very male mustached face I put as the face-swap and then at some point it just snaps over to a female face, the same female face every single time
@@k5c5 the one that gets automatically downloaded when you install Fooocus. I've actually gotten it to do some non-female faces but they don't look very much like the reference photo. What's annoying is half way through while watching the preview it looks at times like it's getting verrryyy close to the reference but around step 20 out of 30 or 40 out of 60 it's starts deviating away
What is your setup? Are you running it locally or in Google Colab? The Fooocus Github issues are a good place to put things like this if it's an issue with a specific machine!
@k5c5 Hello Friend. Perhaps you can tell me when I click “mixing image prompt and inpaint” And then I run the generation, for a while I see something happening, and then it gives several errors, they need to start Colab again.
I can't get the inpaint method to work. It works up until about step 30, then every time it completely flips to a totally random face. It's as if it's ignoring the Stop At for the face swap setting.
I'm just using Stable Diffusion XL. Which method are you trying? If it's wildly wrong I sometimes try a second pass, using the result of the first try as the output. Maybe also try increasing the stop at, to attempt to get a more complete face.
For me 1 and 1 works best in "Vary (subtle)". Inpaint doesn't work well for me, faces don't get cloned. They are different. I don't know which model he's using etc.
I managed to fix this in my colab so it doesn't keep happening. Check out the colab linked here, the second code block installs some xtra fooocus dependency that solves the error: ruclips.net/video/E37Tdcx2DyA/видео.html
cant get this stupid f-ing thing to work. cant get it to work in stable defusion automattic 1111 too i mean seriously ...can do eveything else but this crap
![I Replaced ALL my ADOBE APPS with these [free or cheaper] Alternatives!](/img/1.gif)
best guide on face swapping there is
indeed. This man knows how to use this program feature
Ive been trying this for so long with no success. The second way you showed works! Thank you!
Wonderful, just what was needed. A1111 v 1.7 has messed up so many things from the prior 1.6 version when running in Stability Matrix, that I'm ready to give up on it, now that you have shown how to get things working in Fooocus, which was already far better for In and Outpainting. Thanks a Lot ! ! ! !
Perfect! thanks... this is exactly what I was looking for!
Amazing! Just what I was looking for
This is Exactly what I was looking for. Thank You! Subbed and thumbs up..
what would be the follow up that you mention that we can fix it ? :-)
Bro plzz try to upload more video about fooocus and comfyui
thank you! may your channel continue to grow
Thanx bro... Looking forward to more Fooocus vids
Buddy, this was so helpful.
Exactly what I've been trying to figure.
Appreciate this brief yet valuable tutorial.
This was helpful. Fooocus looks different now though and I cannot locate the face swap feature. There is no advanced check box anymore and now it shows a "hurt me plenty" check box. Just looks different. Today is July 6th and I'm running the latest version 1.52.0. Can you make an updated video? :)
Thank you very much for letting me know about this feature.
nice video but, this isn't working for me. It's still an obviously different face then my reference face image. I wish they made it easy to just reference the face exactly as the referenced image...I wouldn't think it would be hard to do but 🤷♂
Same
did you fix it??
the best solution so far!
Thanks but it gives me error when I try it, some hints?
Is it possible to save settings in Foocus when we close it? Saves a lot of time and makes the model more consistent.
works much better with "variation"! Ah, there's no need to post any prompts :-D
thanks for the vid. more Fooocus UI tutorials pls (like how to run LoRaS and make highly realistic images of people)
Thank you so much for this. It's mind blowing. Quick question for you...
I'm trying to generate an image of me holding my company's product, which is a tea box with the brand and logo on it.
How do I generate an image such an image?
I know I can use the face swap for my face but for the object, I really have no clue how to do it. I know I can use the image prompt for the product but the images generated won't be have my exact product.
Appreciate your help and thank you so much again!
Hello, i have a problema. Whem i do a faceswap i have a problema of conection, any tip?
Will check this out later. This might be the missing link
Why im always having "error" while using inpaint/image prompt mix way ? Always, while using focus colab on web. Is there a reason ?
This is Awesome! very very helpful, does this work for other objects as well? I'm trying to add inanimate objects to an already established image, any tips?
Good question! I imagine not using the advanced FaceSwap checkbox in the Image prompt. Crop the image prompt down to just the object. Then I'd say the inpaint will be a good option.
Alternatively using the variation with a descriptive text prompt of the scene, including the name of the object in the image prompt.
Hi, is it possible that something has changed since the last update? I've noticed that "Mixing Image Prompt and Inpaint" no longer works. The result is largely the same as the original image. Does anyone know more?
I want a tip for a 100% face swap, not a mixture of inpaint and input
Super video! But I have one question! What does the Stop At do exactly? I can't find any documentation about it!
Commented this elsewhere, I probably should have explained in the video :,)
"The stop at level is the percentage of how "complete" the face swap is before it stops. 1 = 100%, 0.9 = 9% etc.
A lower stop at should in theory be a lower likeness, but it's kinda hard to know exactly and different per face/reference image.
Try playing with different levels, 0.9 will be a good default."
@@k5c5 I see! It makes sense after the explanation! Thanks for clarifying that! ☺️
HI! Great tutorial. I'm using it, but both by changing PC and internet connection, it remains slow in creation (even though I put speed) but the most annoying thing is that it often gives an ERROR, when creating the prompt, a red warning. Can you help me understand why?
How can I generate image like Aitana lopez
Realistic, lighting, background,deeply realistic image
Thanks. Can't I use image prompt(cpds,faceswap,pyracanny) and inpaint at the same time on foocus?
With face swap, what does "stop at" do, and for "weight," should you go closer to 1 or 0 to get it to reflect the face reference more?
thank you for the 2nd tip, working well
tried this 3 times today on a a project and have to say each time it kicked out a image, was nothing like the faceswap person?
Can you explain what the "Stop At" value means?
Commented this elsewhere, I probably should have explained in the video :,)
"The stop at level is the percentage of how "complete" the face swap is before it stops. 1 = 100%, 0.9 = 9% etc.
A lower stop at should in theory be a lower likeness, but it's kinda hard to know exactly and different per face/reference image.
Try playing with different levels, 0.9 will be a good default."
Nice job
thank you, this is very useful!
nice work - thank you
Nice video!!
Could you tell me what "Stop At" does please?
I probably should have explained in the video :D
The stop at level is the percentage of how "complete" the face swap is before it stops. 1 = 100%, 0.9 = 9% etc.
A lower stop at should in theory be a lower likeness, but it's kinda hard to know exactly and different per face/reference image.
Try playing with different levels, 0.9 will be a good default.
thank you very much! your second variety works and the rest of focus too. at the inpaint variety i get errors..do you have any clue why?
How to get side view of the face or change expressions?
more vieos please!!! LORAS NOW LOVE U man dammm
Hi, i try the 2 methods but when i click on generate nothing work 🤔
is this fooocus or foocus MRE with all these extra options???
Great video, thanks!
very thank you my friend.
Do you have a tutorial how to install thaty program and how to use it ?
Thanks for the video! For me it doesn't work very well. Can you give me some tips? I have problems in obtaining a credible face swap
Which method did you try? I find that the variation provides a more credible overall image, but the inpaint doesn't alter the rest of the image.
Is your reference face cropped? I found that providing a tighter crop on there helped.
You can also do a second run, using the first pass as your reference image and keeping the same face swap.
I have problems with both methods, and I also tried to use a face swap and a prompt genration in the same time. For me both methods gave mediocre results and I don't know what I'm doing wrong@@k5c5
@@k5c5 instead of using pre-croped faces, fooocus also allows for the selection of the promted images' area (i.e. the face) in the image promt directly via the pen button.
Yeah exactly, you're able to crop directly inside of Fooocus!
Having no connection error using fooocus
Is there a way to fix that ?
Im using 1660ti tho, would really appreciate the help
How to use variation when there are multiple face in same pic?
Can't we use variation to alter only face of one subject?
how to install controlnet in fooocus and use in Colab?
What is the thing in promt window next to the generate button ?
Green rotating arrow. What is it ?
What GPU do u have please? It always takes me like 5 minutes to make an image
I run it in Google colab, my intel Mac isn't up to the job!
I don't know why but when using inpaint and faceswap, the generation process change completely the face at the half (step 30), Any idea :/ ?
Same here
Fixed it, in inpaint click on improve detail, I think the video skips this part
@@matiasdufey8385 yeah nice spot - I tried to add it as an update to the video comment but can't edit the original!
@@matiasdufey8385 Then it doesn't clone the face from image input... For me only Vary (subtle) works best with face cloning.
I keep getting the same womans face, half way through the generation the fuzzy preview looks like it is heading towards the very male mustached face I put as the face-swap and then at some point it just snaps over to a female face, the same female face every single time
Which model are you using?
@@k5c5 the one that gets automatically downloaded when you install Fooocus. I've actually gotten it to do some non-female faces but they don't look very much like the reference photo. What's annoying is half way through while watching the preview it looks at times like it's getting verrryyy close to the reference but around step 20 out of 30 or 40 out of 60 it's starts deviating away
Fixed it, in inpaint click on improve detail, I think the video skips this part
Where is the link to open it
can you help me when i generate anything it starts to lag first then creates 1 images in minutes!
What is your setup? Are you running it locally or in Google Colab? The Fooocus Github issues are a good place to put things like this if it's an issue with a specific machine!
@@k5c5 idk i just saw a video how to download it so i guess it's running locally not in google colab
What "Style" must be used for face swap?
@k5c5 Hello Friend. Perhaps you can tell me when I click “mixing image prompt and inpaint” And then I run the generation, for a while I see something happening, and then it gives several errors, they need to start Colab again.
I tried to add an update comment to the description, in inpaint click on improve detail from the dropdown
I can't get the inpaint method to work. It works up until about step 30, then every time it completely flips to a totally random face. It's as if it's ignoring the Stop At for the face swap setting.
I tried to add an update comment to the description, in inpaint click on improve detail from the dropdown
@@k5c5 That did the trick, thank you!
Interesting to know thanks! Does it matter which model to use?
I believe different models can "favour" specific faces if they're heavily trained
hm which model and which style do you use to get this results, my replace faced doesnt look even close to the source image
I'm just using Stable Diffusion XL. Which method are you trying? If it's wildly wrong I sometimes try a second pass, using the result of the first try as the output.
Maybe also try increasing the stop at, to attempt to get a more complete face.
Is it better than Roop?
But the lighting not blended that well.
Any suggestion for a good value of "stop at" and "weight"?
For me 1 and 1 works best in "Vary (subtle)". Inpaint doesn't work well for me, faces don't get cloned. They are different. I don't know which model he's using etc.
@@hmm-m it is not about model just put the vaues at 1 & 1.2 on face swap and ur work done
My program stops (^C appears in the console) when I'm running the inpaint method. Any ideas?
For me on the free Google colab version it crashes more frequently. With the paid version I can run it ok. Have you tried running it locally?
I managed to fix this in my colab so it doesn't keep happening. Check out the colab linked here, the second code block installs some xtra fooocus dependency that solves the error: ruclips.net/video/E37Tdcx2DyA/видео.html
Aaaaah shit why haven’t I known about this thx a lot
If you take a face from a photo and paste it into the same photo, then you will no longer recognize the person. So your method is not working
Неа, не работает, лицо непохоже, только общий типаж передаётся - it is not working
Thank you. YOu sound like Elon Musk :)
you kinda sound like elon musk
Second was much better, the first one basically did nothing, not even the eye color, cmon man ^^
cant get this stupid f-ing thing to work. cant get it to work in stable defusion automattic 1111 too i mean seriously ...can do eveything else but this crap