abyz22_setimageinfo

abyz22_setimageinfo does nothing, on purpose. It's a pass-through node: six values go in, the same six values come out, no processing in between. In the image_control pack it's the settings panel for the author's batch-generation rig - one labeled place where "how many prompts, how many images per prompt, what resolution" live, routed to the nodes that need them by wires instead of by clicking through every node in the graph.

Don't let the lack of compute fool you into skipping it. The whole point is that it's a named, visible source of truth. In a big workflow you don't want to dig through a KSampler to remember what resolution you set; you want one node that says it out loud and hands the value to everything downstream.

How it works is almost insultingly simple - the function literally returns the arguments it was given. That means there's no state, no math, no failure mode to debug, which is why it's pleasant to use.

The inputs that matter:

• mode_type - an enum with two choices, ab and d. These are the author's own generation-mode labels (they map to two of his output rigs). They mean whatever your workflow makes them mean; don't overthink them.
• Num of Prompts (0–30) - how many different prompts the batch will cover.
• image per Prompt (1–100) - images per prompt.
• batch_size (1–10) - how many images each sampler run produces.
• width / height (64–2048, step 64, defaults 512×768) - output resolution.

All six pop out as outputs under the same names (Mode_type, Num of Prompts, image per Prompt, batch_size, width, height), so you can wire width and height into an Empty Latent Image, batch_size into the sampler, and the rest into the pack's queue driver.

Where it fits: this is the "define the job" half of the author's two-node batch system - it pairs with abyz22_SetQueue, which is the "execute the loop" half. The pack's sample workflow around it leans hard on Impact Pack's detector and SEGS machinery for anime detail passes, so if you're loading that workflow you'll want ltdrdata's ComfyUI-Impact-Pack installed too.

Is it necessary? No. You can reproduce it with a few Number/String nodes in five minutes. But it's tidy, it labels things for you, and for a beginner it's quietly educational: wire it up, watch the values flow through, and you'll get a real feel for how ComfyUI moves data around a graph.

One caveat, and it follows from the design: because there's no validation, it will happily output nonsense. Set Num of Prompts to 0 and it hands 0 downstream without blinking - whatever you wired it to decides what zero means. Don't expect guardrails from a node whose only job is holding your hand steady. Install it with the rest of the pack (Manager → image_control, or git clone https://github.com/abyz22/image_control into custom_nodes/), and you're done - no models, no extra setup for this one.