excel_qwen_artistic

Qwen-Image's models are unusually good at rendering actual legible text inside a generated image - words that stay sharp, hold their font, and don't dissolve into gibberish the way a lot of diffusion models' text does. excel_qwen_artistic exists because that capability opens up a whole genre most models can't touch: typographic art, where the text itself is the composition - a word rendered as a 3D structure, wrapped in a texture, sitting on a designed background. This node is a prompt builder purpose-built for that genre, and it's the most specialized of Apt_Preset's dropdown-driven prompt nodes.

Like excel_VedioPrompt, this one does template substitution rather than plain concatenation, and it documents the mechanism with its own worked example: a template like "文字的特点是 {text_font}，{sum_toon}" (roughly, "the text's characteristics are {text_font}, {sum_toon}"), with 宋体 (Song typeface) and 蓝色氛围 (blue atmosphere) picked for those two fields, produces "文字的特点是 宋体，蓝色氛围". So the flow is: pick a template structure, then fill in the pieces it references.

The pieces on offer. sum_preset (110 choices) is a set of fully curated combos - named looks like a neon-guochao aesthetic or a metallic gold treatment - for when you want a complete style in one click rather than assembling it from parts. If you'd rather build it up yourself: text_medium (210) is the background/canvas the text sits on, text_array (117) is the 3D letterform structure - extruded, layered, twisted, hollow-shell, and similar treatments - text_font (190) covers actual typefaces (both Chinese faces like Song and Hei, and Western ones like Arial and Calibri), text_effect (310, the biggest list here) is surface texture applied to the letters themselves - marble, wood grain, metal wire-draw, leather, fabric - and sum_toon (93) is the overall color scheme. text_cn and text_en are where you actually type the words you want rendered, in Chinese and English respectively. object and target are two more free-text fields the template can reference, and an optional multiline custom field lets you write your own structure instead of picking from template.

One thing worth flagging honestly: unlike excel_Prompter, excel_VedioPrompt, and excel_roles, this node has no language toggle at all - there's simply no field for it in the schema. So whatever the dropdown text and template output language ends up being, you don't get a switch to flip it here the way you do on the pack's other prompt nodes. As with the rest of the excel_* family, expect the dropdown labels themselves to be primarily in Chinese, reflecting the author's China-based audience.

Output. A single STRING, pos - the finished, template-substituted prompt.

Installing it. Same install as the rest of the pack - search "ComfyUI-Apt_Preset" in ComfyUI Manager, or:

cd ComfyUI/custom_nodes
git clone https://github.com/cardenluo/ComfyUI-Apt_Preset.git

install.bat on Windows for dependencies, restart. No models or GPU dependency for this node specifically - it's assembling text, nothing more - so it should work the moment the pack loads.

Troubleshooting. If a placeholder in your template or custom field isn't being replaced, check the exact key spelling against what the node documents - {text_medium}, {text_array}, {text_font}, {text_effect}, {sum_toon}, {text_cn}, {text_en}, {object}, {target} - an unrecognized key is left as literal text rather than erroring. For the actual generation step downstream, remember this node only writes the prompt; getting crisp rendered text also depends on your sampler settings and on using a Qwen-Image checkpoint capable of the text-rendering quality you're prompting for in the first place. If the pack fails to import at all, that's a separate, documented issue with a missing dependency, not this node - check the ComfyUI console log and reinstall via Manager.