Refacer - Painting to Realistic (and Vice-Versa) in ComfyUI

Faces. Most of us have at least 1, and they're great fun to play with. This AI stuff seems to be zooming ahead really quickly, so I figured why not put a bunch of face things together, both old and new, and see how quickly it can generate? Thus, Refacer was born. Powered by SDXL, you can pop in any face and change it to whatever style takes your fancy! Realistic to anime, painting to realistic, cartoon to pixel art - it's all down to your choice of prompt and model :)
Want to support the channel?
/ nerdyrodent
GitHub:
github.com/nerdyrodent/AVeryC...
== Even More Info! ==
* Installing Anaconda for MS Windows Beginners - • Anaconda - Python Inst...
* Installing ComfyUI for Beginners - • How to Install ComfyUI...
* ComfyUI Workflows for Beginners - • ComfyUI Workflow Creat...
* Photomaker - • PhotoMaker - better th...
* Hyper SDXL - • Hyper-SD - Better than...
* Make an Animated, Talking Avatar - • Create your own animat...
* Reposer = A Consistent Character In ANY Pose - • Reposer = Consistent S...

[@Art0691p]

Excellent work. Thanks.

[@Skydam33hoezee]

Lots of fun bringing Dutch heroes to life from paintings! Good results with the StockPhoto model, but even better with JuggernautXL.

[@NerdyRodent]

Yup, been having all sorts of fun 😄

[@JonMurray]

Totally and utterly lost but very interested hahah! New subscriber ✌🏻

[@NerdyRodent]

Welcome aboard the AI train! 🚊

[@runebinder]

Really cool, although if I try an image with 2 faces changes the index sticks with the face on the left and doesn't swap to the other person on the right.

[@thebonuslvl7181]

very nice tx, u seemed to mention that this refacer goes the other way as well (making real faces into paintings, cartoons etc) yet whatever i write in the prompt, neg or change weight it tries to make it a realistic photo image.. any idea on what im doing wrong cos this looks like a very powerful tool i'd just prefer it the other way round?

[@NerdyRodent]

Probably the easiest way is to grab the real to anime example from my patreon 😉

[@alexgilseg]

This is hellacool! However I'm struggling using the workflow to reverse the process.. How did you use this to turn photo into painting ?

[@chrisgreenwell3404]

Great stuff , thank you :) I am looking for workflow that will take a face from image (a) and put it into image (b) with no prompting and keep all the facial details from (a).

[@nirdeshshrestha9056]

ipadapter does that

[@Deep-Sea-Dream]

ROOP

[@onewiththefreaks3664]

Reactor node for comfyui does exactly that.

[@francaleu7777]

Hi Nerdy Rodent, Thank you.. Is it possible to do the same no speciphic to faces? I mean full body or pictures without humans?

[@NerdyRodent]

Give it a go and see 😉

[@flyingdogsoft]

Amazing examples! So license is fine and not so terrible limited as Face-ID?

[@NerdyRodent]

Looks that way to me, but always double check!

[@davidrosbo4519]

Is there an option to do this in Forge? On an rtx4060 comfy would take 2 mins per image whereas forge does it in 10 seconds.

[@suthamchindaudom6410]

I really like this workflow. You are awesome. Is there a way comfyui can convert a realistic portrait photo to a painterly photo or watercolor ,or anime without distort facial features of a person? Could you build such a workflow please?

[@NerdyRodent]

Yup, all of those work too! If you want the pre-made configs, there are a bunch for sketch on Patreon, along with also painting and anime 😊

[@suthamchindaudom6410]

@@NerdyRodent I've just log in to Patreon last Wednesday , but I don't khow how to find anyone who can built a workflow for convert img2painting , img2watercolor and img2anime . Could you sugggest some? Could you recommend where I can go and ask for this kind of workflow?

[@NerdyRodent]

@@suthamchindaudom6410 best drop me a dm on patreon!

[@AIAngelGallery]

wow!

[@PeteJohnson1471]

I'm struggling to work out what the CLIP Model is, where to find it/place it. can anyone assist please? 🙂

[@2PeteShakur]

Clip_Vision must be Clip-ViT-H-14-laion2B-s32B-b79k.safetensors in the clip vision folder of comfy models

[@PeteJohnson1471]

@@2PeteShakur Ahh cheers, Sorry for slow response, I've been logged into another account, I'll have another look and try that. Thankyou.

[@2PeteShakur]

@@PeteJohnson1471 no worries, it's not bad but definitely not as good as doing it on auto 1111!😃

[@sinayagubi8805]

Amazing

[@NerdyRodent]

Thank you! Cheers!

[@macbetabetamac8998]

Awesome work thanks my furry friend!

[@NerdyRodent]

Thank you too!

[@Beauty.and.FashionPhotographer]

SIMILARIZE would be holy grail comfyui Workflow now , after the REFACER has made photos from a painting. A cool challenge, where your Newton Photo would be made looking more like the painting of NEWTON you used, so a forcing of facial features towards the original painting sort of thing? .... not sure if this is even remotely possible....

[@randomprocess7876]

is this workflow available on your patreon?

[@NerdyRodent]

Yup, multiple versions on patreon!

[@Hazzehh]

Can't believe Newton had grey hair at 30

[@NerdyRodent]

Ikr!

[@sherpya]

I had gray hair before

[@azmodel]

I'm sorry, I'm very lost... I go to the repository I found a PNG file and that's it... I tried to replicate it manually on ComfyUI but I'm missing several nodes... So, is there any place in the repository where I can grab everything I need?

[@NerdyRodent]

You need to load the PNG into ComfyUI. Check the links in the video description for the beginners guides to comfy!

[@arri6984]

Node type not found "IPAdapterAdvanced". The manager can't seem to fix this missing node. How can I install this?

[@NerdyRodent]

You can drop me a line on patreon if you need personalised support 😀

[@edmondyang]

Error occurred when executing LoraLoaderModelOnly: Error while deserializing header: HeaderTooLarge I am getting this on the LORA loader :( any tips?

[@NerdyRodent]

Never seen that one! I'd guess at making sure comfy is up-to-date and the lora file is OK

[@LouisGedo]

👋

[@ismgroov4094]

Thx workflow plz, sir

[@lance3301]

His workflows are uploaded to his github link.

[@steinscamus8037]

No a1111 for this?

[@NerdyRodent]

No tgate I’m aware of for a1111 as yet, but maybe in a few months time?

[@KINGLIFERISM]

Am I the only one who thinks this needs a second pass? I guess I can do it but... shame he didn't put that in place.

[@Beauty.and.FashionPhotographer]

such a great workflow.....was looking for adding real skin to Midjourney's Portraits Photos where MIJ failed to create real skin. This just might be a solution. i did encounter a hard stop though through installation. gave me errors, undoubtedly because of my Mac (Macs might have AMD GPU i am guessing?). so your workflow im Comfyui stopped here: Error occurred when executing AutoCropFaces: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU. as a newbie and not knowing what this all really means and how to fix this,..... it became now an amazing thing... which did not work....

[@NerdyRodent]

Ah, probably need a different node with special Mac support instead. There are quite a few out there, so hopefully something works on Mac!

[@Beauty.and.FashionPhotographer]

@@NerdyRodent it works now, for macs took the WAS SUITE Image crop node, it is not as precise as you get them from painting to real faces, even if i clone all settings, but maybe once i learn more, i can take sections from other peoples comfyui projects, where faces remain almost identical (for ex. supir)

[@hugoleote]

I managed to make this work. You need the following custom nodes: cg-use-everywhere comfyui-art-venture ComfyUI-AutoCropFaces ComfyUI_IPAdapter_plus ComfyUI-PhotoMaker-Plus ComfyUI_TGate And the Clip_Vision must be Clip-ViT-H-14-laion2B-s32B-b79k.safetensors. Great work!

[@2PeteShakur]

yeah that seems right no issues with that just use comfy manager for missing nodes - only had issue with model - seems was in the wrong node so changed it and all working now! ;)

[@KINGLIFERISM]

at least someone isnt hiding information. Nice boss! Thanks.