ComfyUI img2img ControlNet IPAdapter ReActor Face Swap workflow

Comfy-UI image2image ControlNet IPAdapter ReActor workflow
starting with low resolution image, using ControlNet to get the style and pose, using IPAdapter to get as close as possible to the reference image, and ReActor to much the result
#comfyui #stablediffusion #ipadapter #controlnet #ReActor #img2img #controlnet #faceswap
follow me @ / pixeleasel
Workflow:
*update Workflow Version (Compatible with Ip-adapter V2)
drive.google.com/file/d/1rL2V...
old IPAdapter version
drive.google.com/file/d/1SH1S...
photoshop to Comfy-UI
• Photoshop inside ComfyUI
using templates in Comfy-UI
• ComfyUI Animatediff #i...

[@elan4912]

Very nice tutorial. Thansk for the workflow.

[@PixelEasel]

Glad it was helpful!

[@WalidDingsdale]

amazing technique to combine PS with Comfyui， thanks for sharing this

[@PixelEasel]

thanks for the comment!

[@randy2d]

Jesus christ man you gave out an insane amount of amazing info that way too many other "ai youtube teachers just don't even bother to do" I wanted to know more about controlnet that't the only reason I clicked on the video, and the steps you took otherwise I had to learn them a really hard way LOL I wish you would've had this video a couple months sooner! Great job! Keep it up!

[@PixelEasel]

thank man! really appreciate

[@henryturner4281]

What is the pytorch_lora_weights LoRA you are using?

[@PixelEasel]

its LCM lora

[@yulasharm4723]

is there a way without using photoshop or any other app, just within comfyui?

[@PixelEasel]

yes. u jusat nees to create the mask in comfy, here you have an example of using masks within comfy kzbin.info/www/bejne/moG5d2Z_rMd0l7c

[@dronearon3085]

Try usinj krita replaced ps ..it juilt in stable diffusion

[@PixelEasel]

yes i know, but nothing give me the flexibility that comfy does

[@WhySoBroke]

Isn’t there a CNET model that can mask the image?

[@PixelEasel]

There are many ways to make masks in Comfy, using Photoshop in this case is more convenient and efficient for me

[@godorox]

PhotoshopToComfyUI is unavailable, where can i get this?

[@PixelEasel]

I will check if there is any other node with the same functionality

[@godorox]

@@PixelEasel Thx.

[@YuriyLvov]

IPadapter V2 fix workflow?

[@PixelEasel]

yep! you can find the new version in the description

[@alexandergarzon8052]

I thinnk there are models to automaically create mask from faces and background removals, that will skip the photoshop step. Thanks,

[@PixelEasel]

indeed and I use it a lot, but sometimes it's more efficient to use photoshop

[@gainsum]

Yes. Yoloworld is my favorite

[@deama15]

I keep getting Error occurred when executing IPAdapterAdvanced: Error(s) in loading state_dict for Resampler: size mismatch for proj_in.weight: copying a param with shape torch.Size([768, 1280]) from checkpoint, the shape in current model is torch.Size([768, 1024]).

[@PixelEasel]

try to delete the ip node from the workflow , and bring it back again

[@deama15]

@@PixelEasel No, doesn't work, same issue. The IPadapter I'm using is "ip-adapter_sd15" but I also tried "ip-adapter-plus_sd15" and "ip-adapter-plus_sdxl_vit-h", same issues.

[@PawFromTheBroons]

Text2Speech Pexels pronunciation brokes the video from the very start.

[@PixelEasel]

ill try to improve it for the next vids

[@Gardener7]

The content is great, but elevenlabs would sound much more realistic.

[@NGIgri]

Just... F*** it. No matter what I do, I can't install ReActor Node for ComfyUI and comfyui-photoshop node.

[@PixelEasel]

i hope u'll solve it soon... about the photoshop node , i think the creator just made new one , but i didnt get it yet

[@NGIgri]

@@PixelEasel thx for info, I keep digging in subject. Great channel, by the way =)