Qwen Image 2.1PROMPT EXAMPLES
Twelve text-to-image briefs written for what Qwen Image 2.1 is built to do well: exact lettering, clean product light, natural portraits and structured layouts. Copy any of them, or send one straight to the generator.
12 STARTING POINTS
Each card is a complete text-to-image brief with a suggested aspect ratio and one thing to check in the result. "Use this prompt" opens the studio with the text filled in.
A minimalist concert poster. Exact headline "NIGHT TIDE" in tall condensed white capitals across the top, a smaller line "LIVE AT THE HARBOUR HALL · 14 NOV" beneath it, deep indigo background, a single silver wave line across the lower third, generous margins.
Check: Both lines spelled exactly, headline clearly larger than the sub-line.
A neon shop sign that reads "QWEN IMAGE 2.1", rainy night, reflections on wet pavement
Check: Lettering legible in the sign and plausibly mirrored in the reflections.
Prompt from the official model card
A matte cream tea tin on a pale oak shelf. The label reads "JASMINE NO. 4" in dark green serif lettering with a thin gold border and a small line "50 g loose leaf" below. Soft window light from the left, shallow depth of field, product photography.
Check: Text follows the curve of the tin and the small line stays readable.
A pair of white leather sneakers floating above a pastel blue plinth, soft studio shadows, one lace trailing, clean e-commerce hero shot, high detail on stitching and leather texture.
Check: Stitching and lace holes are consistent on both shoes.
A frosted glass serum bottle with a brushed aluminium cap standing in shallow water, gentle ripples, pale peach backdrop, single soft key light from above, luxury skincare advertisement.
Check: Glass looks frosted rather than clear, and the reflection in the water matches the bottle.
Editorial portrait of a woman in her sixties with cropped silver hair and a navy wool coat, standing by a rain-streaked window, soft side light, natural skin texture, 85mm look, muted colour palette.
Check: Skin keeps texture instead of looking airbrushed; light direction is consistent.
A young chef plating a dish in a narrow restaurant kitchen, steam rising, warm tungsten light, stainless steel reflections, candid documentary photograph, shallow depth of field.
Check: Hands and utensils are anatomically plausible.
A clean infographic titled "HOW TO BREW POUR-OVER" with four numbered steps in a row: "1 Grind", "2 Bloom", "3 Pour", "4 Serve". Flat vector icons above each label, warm beige background, dark brown text, consistent line weight.
Check: Exactly four steps, in order, each with its own label.
A mobile app screen mockup for a plant-care app. Top title "My Plants", three cards labelled "Monstera", "Fiddle Leaf" and "Pothos", each with a small leaf illustration and a "Water today" button. Light mode, rounded corners, soft green accent.
Check: Three cards with the right names; buttons aligned and evenly spaced.
A children's book illustration of a fox reading under a paper lantern in a snowy forest, gouache texture, limited palette of teal, rust and cream, gentle paper grain.
Check: Palette stays limited and the brush texture is visible.
An isometric illustration of a tiny rooftop garden café with string lights, two tables and potted lemon trees, pastel colours, clean vector shading, plain white background.
Check: Consistent isometric angle across furniture and plants.
A lone cyclist crossing a long concrete bridge at dawn, fog over the river, distant city skyline, wide cinematic composition, cool blue tones with a warm rim of sunrise.
Check: A good still to animate afterwards in the video workspace.
The official model card describes Qwen-Image-2.1 as a unified generation and editing model with improved typography, portrait lighting and fine detail. Prompts that name those things explicitly give the model something concrete to resolve.
These are prompts, not guaranteed outputs. Every generation is different, so check lettering, hands and edges before you use a result.
This page lists prompts rather than a gallery, because we only show images when we can label the exact model that made them. For reference output published by the model authors and by ComfyUI, use the official sources:
Upstream, the same weights also generate transparent RGBA images and edit existing pictures with up to 10 reference images. The model card gives a fixed sentence pattern for transparent output:
This is an RGBA image with transparency. A cute cartoon dragon sticker. The image has alpha channel and the background is transparent.For multi-image edits in ComfyUI, the documentation addresses each reference by index in the prompt, for example "Keep the character and pose in <image1> unchanged" while taking a garment from <image2>.
In this studio, Qwen Image 2.1 Edit takes up to 3 reference images, for 1 credit per edit. Use a local setup for transparent RGBA output and for edits with more references.
FROM READING TO MAKING
Qwen Image 2.1 generation costs 1 credit per image, with one image per request. Other image and video models show their own credit quote before you submit.
QUESTIONS AND ANSWERS
Yes. The prompts on this page are original and free to copy, edit and reuse. Generating them with Qwen Image 2.1 in this studio costs 1 credit per image. Other models use credits.
We only publish an image when it is labelled with the model that actually produced it. For visual reference, the official Hugging Face model card, the official demo Space and the ComfyUI documentation publish Qwen-Image-2.1 output samples.
Improved typography is one of the changes listed on the official model card, and the ComfyUI documentation describes dense small text and complex layouts for posters, slides, UI mockups and packaging. Put the exact words in quotation marks and always proofread the result.
Each example lists a suggested ratio. This studio offers 1:1, 16:9, 9:16, 4:3 and 3:4 for Qwen Image 2.1. Running the model locally exposes seven native 2K sizes, including 3:2 and 2:3.
Yes. They are plain text-to-image prompts with no site-specific syntax, so they can be pasted into the ComfyUI Qwen-Image-2.1 template or the Diffusers pipeline.