Go to OpenArt main site
Upload workflow

CUTE YOU!Create your cartoon likeness using 1 photo


39 reviews

Easily create a corresponding cartoon character with just one photo!




the new version of V2.0, based on PuLID and InstantID,

for better results and speed!



Demo Video Link:




This workflow includes 15 modules:


The workflow uses the SDXL model, and the recommended is: Samaritan 3D Cartoon. You can download it here: https://civitai.com/models/81270

2.Lora (sliders):

With the recently launched adjustment slider, you can adjust the size of the eyes, the strength of the smile, the length of the hair, and many other dimensions. Feel free to add or subtract. You can download it here: https://sliders.baulab.info/weights/xl_sliders/


Use IPA to get facial features from your photo. This is a 4-in-1 node where you can input 4 photos and adjust the weight of each photo. If you only have one photo that's okay, just repeat in each image loader. Be sure to use XL's facial version, you can download it here: https://huggingface.co/h94/IP-Adapter/tree/main/sdxl_models


Write a description of your character. Ensure it doesn't contradict the image loaded in subsequent Module 5. Some StylePrompts are used here, adjusting them can create interesting effects.


Use Openpose to convey character pose. There is an "Image Scaling by Edges" node which scales the image imported from CN proportionally to a maximum edge of 800 pixels. This is to avoid excessively large images entered by users.


Use Canny to further convey styling details. To avoid too much interference with the character's shape, the end time should not exceed 0.5.

Modules 5 and 6 have loaded the same image, you can freely replace it with your favourite cartoon character image. This image determines the character's posture and proportion.

My image link:



The first image generation. The image size comes from the image size imported from CN.

8.Background Removal:

Different models and thresholds can be swapped here to adjust the deletion effect. The prompt is used to stipulate what you want to keep. For example: "person sitting on chair". If the image cutting result is not good, don't worry, the high-definition restoration in step 11 will solve the problem.


Lay a solid blue color on the background, you can also change it to your favorite color.


A local redraw of the face to enhance the facial modeling information inherited from the IPA. Semantic segmentation is used here to obtain the mask of the face.


Overall high-definition restoration of the image, moving the image into the latent space and enlarging it by 1.5 times, followed by a redraw of 0.6 intensity. The image will change a lot, but don't worry about facial features, because the model's information inherits the facial information brought by the IPA.


To prevent the face from still being incorrect, a final face repair is performed here, which will further convey the facial information inherited from the IPA.Modules

10, 11, and 12 are all redrawing. You can turn them on and adjust them according to the actual effect. All the sampler seeds are fixed, so if you haven't changed any parameters, it will output consistent results.


Add a watermark, write copyright information. You can also change it to any text you want.


I've put a logo of this workflow, just for fun. You can replace or close it.

My image link:


The logo here is alsogenerated using SG, I will also share its workflow later.

There is an AnyWhere node at the bottom to pass the final generated image out.


Save image,the filename input is: CuteYou/%date:yyyy-MM-dd%. It will create a folder named CuteYou in the Out directory and store the image with the current date as the file name prefix. Of course, you can modify it.

If anything is unclear, please add me on WeChat: Rui40000
Have fun!


(No comments yet)



    CuteYou2.0.json (56 kB)
    CuteYou-4Photos.json (66.8 kB)
    CuteYou-1Photo.json (93 kB)
    logo.png (322.5 kB)
    Pose.jpg (63 kB)

No reviews yet

  • - latest (7 months ago)

  • - v20231222-094328

  • - v20231211-071215

  • - v20231209-140305

Primitive Nodes (11)

Anything Everywhere (4)

Get image size (1)

IPAdapterApply (1)

Image scale to side (1)

Int to float (2)

Note (1)

Reroute (1)

Custom Nodes (67)


  • - CM_FloatBinaryOperation (4)

  • - CM_FloatToInt (2)

  • - CR Float To Integer (1)

  • - CR Color Panel (1)


  • - PreviewImage (10)

  • - CLIPTextEncode (2)

  • - EmptyLatentImage (1)

  • - SplitImageWithAlpha (2)

  • - ControlNetApplyAdvanced (2)

  • - JoinImageWithAlpha (1)

  • - LatentUpscaleBy (1)

  • - VAEEncode (1)

  • - LoadImage (3)

  • - ControlNetLoader (2)

  • - CLIPVisionLoader (1)

  • - SaveImage (1)

  • - SAMLoader (1)

  • - UltralyticsDetectorProvider (1)

  • - FaceDetailer (1)

  • - DWPreprocessor (1)

  • - PixelPerfectResolution (1)

  • - AIO_Preprocessor (1)

  • - IPAdapterModelLoader (1)

  • - Float (3)

  • - Text_Image_Multiline_Zho (1)

  • - Image Overlay (3)

  • - KSampler (Efficient) (2)

  • - CheckpointLoader|pysssss (1)

  • - LoraLoader|pysssss (1)

  • - SDXLPromptStyler (3)

  • - SAMModelLoader (segment anything) (1)

  • - GroundingDinoModelLoader (segment anything) (1)

  • - GroundingDinoSAMSegment (segment anything) (1)

  • - GetImageSize (2)

  • - Mask Invert (1)

  • - Image Size to Number (1)

Checkpoints (0)

LoRAs (0)