Concurrent Submit | zhenzhen-sensenova-u1-infographic-fal

Sensenova U1 is SenseTime's infographic-generation model - the niche one in this pack, and the most genuinely different. Where the other nodes make pictures or videos, this one takes a prompt like "compare Q3 sales by region with a bar chart and a map" and returns an actual designed infographic with layout, type, and chart-ish elements baked in. It's not going to replace your data-viz tool for real numbers - models invent figures - but for concept art, pitch decks, social content and "show me roughly what this infographic would look like," it's a shortcut that doesn't exist anywhere else in the pack.

It's reached through fal's queue and wrapped in the standard concurrent layer: shared image pool (30 workers), COMFLY_IMAGE_FUTURE task output, and Concurrent Collect Images (ComflyConcurrent_Image_Await) gathers up to 30 tasks into image_1…image_30. Batch forty infographic concepts, pick the three that work, and hand those to a human designer with a real dataset.

The inputs

• prompt - the infographic brief. Be explicit about the data story, the layout, and the tone; the model follows structure better than vibe.
• aspect_ratio - 1:1, 16:9, 9:16, 3:2, 2:3, 4:3, 3:4, 1:2, 2:1. Portrait 9:16 or 3:4 suits most infographics; 16:9 for slides.
• use_thinking - off by default; enabling lets the model reason about layout before generating, at a time cost. Worth it for complex briefs.
• guidance_scale - 0–20, default 4.
• num_inference_steps - 1–100, default 50. This model likes its 50 steps; dropping to 30 is fine for drafts.
• timestep_shift - 0–10, default 3. A diffusion schedule knob from the sampling side; you'll rarely touch it, but it exists because it changes how the model distributes its steps.
• output_format - jpeg (default) / png.
• sync_mode - off by default; enables synchronous (rather than queued) processing.
• enable_safety_checker - on by default.
• poll_interval / max_poll_attempts - default 600 × 6s = 1 hour.

Install

cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/Comfyui-zhenzhen

Or ComfyUI Manager → search Comfyui-zhenzhen → restart, token into api_key (ai.t8star.org overseas; api.seedance.nz for the domestic discount).

Gotchas

• Numbers in the output are decorative. An LLM-driven infographic model will happily render a plausible-looking "42%" that isn't your actual 42%. Always, always hand-verify the data before anything ships.
• It's a concept tool, not a layout engine. Real typography control, exact chart accuracy, brand-grid fidelity - not this node's job. Treat output as a high-fidelity sketch.
• fal pre-deduction applies here too. The charge you see at submit is an estimate; it settles after the job. And a timeout usually means "check the async-task page," not "pay again."