So if I use multiple realistic images of a person as reference and I want the the result to look as similar as possible to that person which tool should I use please?
Thanks for watching! I made a video on how to install PhotoMaker. The installation is the same for both V1 and V2. Here is the link: bit.ly/3Ww2xV0 (There are other links in the description of the video on PhotoMaker V2) You can use these steps to use multiple images in ComfyUI by using the following steps. 1. Add a load image node and select the first image. 2. Connect this load image node with a batch image node. 3. Add a second image node and connect it to the batch image node. now you have two images from the batch image node. if you want to add more, then you connect the batch image node to another batch image node and keep chaining them. Hope this helps.
ok got it, am I understand correctly? photomaker = full body generation with reference image instantID = close up gave generation with reference image the other = copy paste old tek?
No need to apologize @yngeneer! If Photomaker V2 works best for you, that's great. I forgot to mentioned that when using multiple images of the same person in different lighting and angle, PhotoMaker V2 performs really well.
Using a pretty Caucasian lady will always produce decent results, better tests would be with people with glasses, an African American person, and a heavy set Indian person. Those are the blind points in these models for real world usage.
always useful content! cheers!
Glad to hear that!
So if I use multiple realistic images of a person as reference and I want the the result to look as similar as possible to that person which tool should I use please?
Hello, for best results, you will have to train a lora. If you want to get a quick result, then you can try the IPAdapter FaceID or InstantID.
@ Thank you 🙏
It's great!, could you share how to install photomaker on Comfy UI and use it for photo (many) to photo (one or many)? Thanks
Thanks for watching!
I made a video on how to install PhotoMaker. The installation is the same for both V1 and V2. Here is the link: bit.ly/3Ww2xV0
(There are other links in the description of the video on PhotoMaker V2)
You can use these steps to use multiple images in ComfyUI by using the following steps.
1. Add a load image node and select the first image.
2. Connect this load image node with a batch image node.
3. Add a second image node and connect it to the batch image node.
now you have two images from the batch image node.
if you want to add more, then you connect the batch image node to another batch image node and keep chaining them.
Hope this helps.
ok got it, am I understand correctly?
photomaker = full body generation with reference image
instantID = close up gave generation with reference image
the other = copy paste old tek?
Yes, although IPAdapter has other uses other than generating faces from the reference image.
@@CodeCraftersCorner ok, thank you sir.
👍
I could never get photomaker 2 to work with ComfyUI .. aargh!
Sorry to hear that! To improve bad generations, you can try using multiple images of the same person in different angles and lighting.
for me the photomaker v2 us superior to others.... sry...
No need to apologize @yngeneer! If Photomaker V2 works best for you, that's great. I forgot to mentioned that when using multiple images of the same person in different lighting and angle, PhotoMaker V2 performs really well.
@@CodeCraftersCorner yea, thats right, I didn't realize I'm using multiple images as input for photomaker...
Thanks for the update, @yngeneer!
Using a pretty Caucasian lady will always produce decent results, better tests would be with people with glasses, an African American person, and a heavy set Indian person. Those are the blind points in these models for real world usage.
Mostly but the IPAdapter Face ID V2 does really well on my face.