YouTube watermarks.set: 10 MB JPEG or PNG for channel branding

YouTube's watermarks.set method uploads a watermark image for a channel: JPEG or PNG, up to 10,485,760 bytes (10 MiB), with a channelId parameter and an InvideoBranding body that carries timing. For a logo mark with a clear background, make a PNG; on Sume, openai/gpt-image-2.5 supports background: "transparent".

What does watermarks.set accept?

The discovery document lists the method as POST, with simple and resumable upload paths and accepted types image/jpeg, image/png and a generic binary type. The branding body has timing, with offsetMs and durationMs in milliseconds; the position field is present in the schema but marked deprecated.

How do I make the mark on Sume?

Request a simple mark on a transparent background, 1:1, in PNG. Sume documents background with transparent for ChatGPT Image 2.5. Review the result on a light and a dark frame before you upload it, since a thin mark can disappear on busy footage.

import json, os, urllib.request

body = {
    "model": "openai/gpt-image-2.5",
    "prompt": "simple flat leaf icon, single color, centered, no text",
    "aspect_ratio": "1:1",
    "output_format": "png",
    "background": "transparent"
}
req = urllib.request.Request(
    "https://api.sume.com/v1/images",
    data=json.dumps(body).encode(),
    headers={
        "Authorization": "Bearer " + os.environ["SUME_API_KEY"],
        "Content-Type": "application/json",
    },
    method="POST",
)
print(urllib.request.urlopen(req).read().decode())

What are the practical limits?

• Keep the file well under the 10 MiB cap; a simple icon is far smaller.
• Do not put text in the mark unless you have checked how it renders at small sizes.
• You choose when it appears with timing, so decide whether it should run for the whole video.
• Verify the other fields in the live document.

What else matters?

The watermark is channel branding, not an AI disclosure. See the images guide for Sume's parameters.