temryl / comfys3 Goto Github PK

Backblaze has an s3 compatible api. Is this supported? Maybe the .env needs s3_endpoint instead of s3_region?

My backblaze endpoint looks like this: https://s3.us-east-005.backblazeb2.com

https://www.backblaze.com/docs/en/cloud-storage-call-the-s3-compatible-api?highlight=undefined

It's 5x cheaper than amazon s3 and egress is free up to a limit

Search and Replace allows to add date %date:yyyy-MM-dd% to filenames or target a Import/Export Field of a Node. On any None Primitive Node a Property "Node name for S&R" allows the targeting of a specific Node. Example a RandomNoise Node with the name random can provide then the variable %random.noise_seed% to the filename_prefix

Manual: https://www.comfyuidoc.com/Interface/SaveFileFormatting.html

Allowing the configuration of the Entpoint via this Variable allows the user to target use a AWS PrivateLink or any other S3 Compatible Storage Impementation.

This this shoud target the boto3 endpoint_url argument

@TemryL I've been using the Image Save node from WAS suite to convert images to jpg and remove metadata before saving. Is there any way to use this with the save to s3 node?

For video I assume things like .webp would work but if I want to save a mesh object in either .obj, .gltf etc which node would I use?

P.S. amazing project

@TemryL In the .env, I used "output" as the output directory.

But when the file is saved, it's not saved to the "output" folder. Instead "output" is added as a prefix to the file name. What am I doing wrong?

Thanks for your previous work. However, I don't understand how your Node operates, and I really need it to work for my customer.

Here are the details of my .env file:

S3_REGION = "eu-central-1"
S3_ACCESS_KEY = "XXX"
S3_SECRET_KEY = "XXX"
S3_BUCKET_NAME = "my-great-bucket"
S3_INPUT_DIR = "test"
S3_OUTPUT_DIR = "test"

This is my workflow:

These are the errors I'm encountering:

[2024-07-26 15:28] [ComfyS3] - �[0;31mERROR�[0m - Failed to check if folder exists in S3: 'NoneType' object has no attribute 'Bucket'
[2024-07-26 15:28] [ComfyS3] - �[0;31mERROR�[0m - Failed to create folder in S3: 'NoneType' object has no attribute 'Bucket'
[2024-07-26 15:28] [ComfyS3] - �[0;31mERROR�[0m - Failed to check if folder exists in S3: 'NoneType' object has no attribute 'Bucket'
[2024-07-26 15:28] [ComfyS3] - �[0;31mERROR�[0m - Failed to upload file to S3: 'NoneType' object has no attribute 'Bucket'
[2024-07-26 15:28] [ComfyS3] - �[0;31mERROR�[0m - Failed to upload file to S3: 'NoneType' object has no attribute 'Bucket'
[2024-07-26 15:28] [ComfyS3] - �[0;31mERROR�[0m - Failed to upload file to S3: 'NoneType' object has no attribute 'Bucket'
[2024-07-26 15:28] !!! Exception during processing!!! 'NoneType' object is not iterable

I would greatly appreciate any help with resolving this issue.

ComfyS3 node is not visible in ComfyUI after so many times restart the server and update the node from comfyui manager, what i do and one more important thing i also update the details in .env file

Currently, all files are uploaded to a single output folder in s3. My output folder is called "output". But files can be overwritten when they have the same name, which is becoming frequent for me. I could add a very long random prefix to each file name, like 12312301923810923.jpg, but then the file name starts to look very long and ugly, especially since my users download their files. Is there a way to save each group of files in their own UUID subfolder to prevent overwriting?

For example, if there are 3 outputs for a single run, they would be saved to

output/550e8400-e29b-41d4-a716-446655440000/file1.jpg
output/550e8400-e29b-41d4-a716-446655440000/file2.jpg
output/550e8400-e29b-41d4-a716-446655440000/file3.jpg

Each subsequent run, the UUID subfolder would change.

Currently, the output names look like this.

Hi,

What's a good node/approach to retrieve the S3 URL of the saved file?