Introduction
This portrait generation workflow, carefully constructed based on ComfyUI, integrates cutting-edge technologies to provide users with a variety of creative paths. The workflow is divided into two major modules: the first part uses ByteDance's open source HyperLoRA original node. Users upload a single portrait image and enter a prompt word to quickly generate the same portrait; the second part combines HyperLoRA and InstantID nodes, uploads two identical portrait images and prompt words, and can also produce accurately restored portrait works. Whether it is the pursuit of simple and efficient single-image generation, or the need for higher-precision dual-image matching creation, this workflow can be flexibly adapted to help users easily create ideal portraits.Mainly used for the generation of close-up portraits
Tips
IMPORTANT:
•Please confirm that all model and config files are placed in correct directories.
•The text prompt should start with the LoRA trigger words fcsks fxhks fhyks, to enable LoRA.
•It is recommended to set stop_at_clip_layer to -2 to align with our training config.
HyperLoRA is not compatible with all SDXL base models. We have tested that it is compatible with LEOSAM's HelloWorld XL 3.0, CyberRealistic XL v1.1, and RealVisXL v4.0, but incompatible with ArienMixXL v4.0. Among them, RealVisXL v4.0 may achieve the best results.
https://github.com/bytedance/ComfyUI-HyperLoRA?tab=readme-ov-file
Recommended machine：Ultra
Workflow Overview
How to use this workflow
Part 1 : HyperLoRA-T2I-InstantID
Step 1: Select HyperLoRA-T2I-InstantID Workflow
Step 2: Upload a portrait image
Step 3: Select model, enter prompt word, modify lora weight, modify resolution, and repair face
1.sdxl_hyper_id_lora_v1_fidelity offers better fidelity, while sdxl_hyper_id_lora_v1_edit provides better editability.
2.Input prompt word,"The prompt word 'fcsks fxhks fhyks' is required and cannot be deleted.
3.Adjust the strength of Lora，If there are many prompt words, increase the weight appropriately.
4.Adjust the image resolution.
5.Adjusting the ‘denoise’ parameter can repair the face. The larger the value, the better the repair effect, but the greater the change in the face, resulting in a greater difference from the original portrait.
Part 2 : HyperLoRA-T2I
Step 1: Select HyperLoRA-T2I Workflow
Step 2: Upload a portrait image
Step 3: Select model, enter prompt word, modify lora weight, modify resolution, and repair face
1.sdxl_hyper_id_lora_v1_fidelity offers better fidelity, while sdxl_hyper_id_lora_v1_edit provides better editability.
2.Input prompt word,"The prompt word 'fcsks fxhks fhyks' is required and cannot be deleted.
3.Adjust the strength of Lora，If there are many prompt words, increase the weight appropriately.
4.Adjust the image resolution.
5.Adjusting the ‘denoise’ parameter can repair the face. The larger the value, the better the repair effect, but the greater the change in the face, resulting in a greater difference from the original portrait.