Looking for AI images in the Floating Dreamscape style? A dreamy and tranquil pastel style that feels like dreaming in a weightless space. Here are 2,334 AI images in the Floating Dreamscape style, tagged with keywords like soft-focus, weightlessness, pastel art and high-key. You also get a CSV with the title, caption, alt text, title attribute and keywords of every image. Use them as they are, and use the images free.
What’s in the pack
- 2,334 images in WebP format
- 45 different resolutions, from 1184 × 864 up to 4800 × 3584 (592 square, 943 landscape, 799 portrait)
- Most common size: 1024 × 1024 (576 images)
- 11 categories and 8 AI image models
- 1 metadata CSV: 22 columns, English and Korean side by side
- ZIP size: about 190 MB
11 categories in one pack
Every image belongs to one category you can filter by in the CSV (category_name_en column):
- Architecture (232)
- Nature (230)
- Objects (229)
- Flora (222)
- Abstract (214)
- Food (213)
- Fashion (213)
- Space (209)
- Animals (205)
- Mood (201)
- Scene (166)
Which AI models made them
The model behind each image is recorded in the CSV (image_model column):
- gemini-2.5-flash-image (826)
- imagen-4.0-generate-001 (404)
- gemini-3-pro-image-preview (302)
- flux-1-schnell (220)
- imagen-4.0-fast-generate-001 (217)
- imagen-3.0-generate-002 (190)
- imagen-4.0-ultra-generate-001 (166)
- dall-e-3 (9)
Keywords you’ll find most often
ethereal, pastel, soft light, morning, early morning, soft focus, fantasy, calm, nature, close-up, golden hour, tranquility, soft, surreal, light
What one CSV entry looks like
- Title: Fantasy Apartments
- Caption: high-rise concrete apartments downtown
- Alt text: Ethereal Downtown Apartment Evening
- Keywords: apartments, high-rise buildings, evening, fantasy, pastel, city, calm
What's inside the metadata CSV
One UTF-8 CSV, one row per image, English and Korean side by side (columns ending in _en and _ko).
- file, mime_type, filesize: the exact file name, format and size in bytes. Use
fileto match each row to its image. - image_wh: pixel dimensions, for example 1024x1024.
- category_name, style_name, image_model: how each image is classified.
- prompt_title: a longer, descriptive title.
- image_title: a short display title.
- caption: a one-line caption.
- alt: alt text written for accessibility and SEO.
- title_attr: a ready-to-paste title attribute.
- keywords: comma-separated keywords you can use as tags.
- slug: a URL-friendly slug.
How you can use this pack
Publish one post per image. Map the CSV's image title to the post title, the caption to the excerpt, the alt text to the image alt, and the keywords to tags. Your posts are half-written before you start.
Build an image library or gallery site. Load the files and the CSV into your own image-serving site, gallery, or CDN test bed. The category, style, size and model columns give you filters out of the box.
Make any post look finished. Use one image as a hero or inline picture, then paste its alt text, title attribute and caption straight from the CSV.
Prototype with real content. Fill mockups, themes, templates and demos with real images and real metadata instead of placeholders.
About these AI-generated images
Every image in this pack was generated by AI. They are not photographs, and they do not depict real events. Any resemblance to real people, places, products, brands or artworks is coincidental and unintended, and no real person, organization or trademark is represented or endorsed. As AI output, an image may contain small visual flaws such as distorted details or unreadable text, so please review each image before you use it.
Free to use. Ad-supported download.
The content is free to use: no fees, no subscriptions, no usage limits. To keep this library free for everyone, a short ad appears before the download unlocks. Watch it, and the file is yours.
License
All content in this download is free to use for personal and commercial projects, with no usage restrictions. Attribution is not required. Files are provided as is, without warranty.
For full details, see the License & Terms of Use.
