Visit ComfyUI Online for ready-to-use ComfyUI environment
ComfyUI Image Saver enables saving images with generation metadata compatible with Civitai, supporting png, jpeg, and webp formats. It also stores LoRAs, models, and embeddings hashes for resource recognition.
ComfyUI-Image-Saver is an extension designed to help AI artists save images along with their generation metadata. This metadata includes important information about how the image was created, such as the model used, parameters, and other settings. The extension is compatible with popular platforms like Civitai and Prompthero, which can automatically detect and display this metadata. It supports saving images in PNG, JPEG, and WEBP formats. For PNG files, it stores both the full workflow in ComfyUI format and A1111-style parameters, while for JPEG and WEBP, only the A1111-style parameters are stored. This extension also includes hashes of models, LoRAs, and embeddings to ensure proper resource linking on Civitai.
ComfyUI-Image-Saver works by embedding metadata directly into the image files you save. This metadata includes detailed information about the image generation process, such as the model used, the parameters set, and any additional resources like LoRAs or embeddings. When you save an image, the extension automatically captures this information and stores it within the image file. This makes it easy to share your images with others while ensuring that all the necessary information to reproduce the image is included.
For example, if you create an image using a specific model and set of parameters, ComfyUI-Image-Saver will save this information within the image file. When someone else views the image on a platform like Civitai, they can see exactly how the image was generated, making it easier to replicate or build upon your work.
You can customize the names of your saved files and folders using placeholders. This allows you to create a naming convention that suits your workflow. Here are some placeholders you can use:
%date
%time
(format taken from time_format
)%model
(full name of the model file)%basemodelname
(name of the model without file extension)%seed
%counter
%sampler_name
%scheduler
%steps
%cfg
%denoise
Example:
| filename
value | Result file name |
| --- | --- |
| %time-%basemodelname-%cfg-%steps-%sampler_name-%scheduler-%seed
| 2023-11-16-131331-Anything-v4.5-pruned-mergedVae-7.0-25-dpm_2-normal-1_01.png
|You can add LoRAs to the prompt in the <lora:name:weight>
format. This will be translated into hashes and stored together with the metadata. For this, it is recommended to use ImpactWildcardEncode
from the . This tool allows you to convert the LoRAs directly to proper conditioning without having to manually manage the LoRA strings.
time_format
is correctly set if using the %time
placeholder.<lora:name:weight>
format correctly.ImpactWildcardEncode
from the ComfyUI-Impact-Pack for better integration.For additional resources, tutorials, and community support, you can explore the following links:
© Copyright 2024 RunComfy. All Rights Reserved.