Describe the image, composition, style, mood, and creative direction you want KREA 2 to generate.
Medium is faster and uses fewer tokens. Large is best for maximum fidelity and art direction.
Higher creativity can produce more expressive images. Lower creativity stays closer to the prompt.
Sign in to spend tokens
Prompt is required
Your generated images will appear here
KREA 2 generates high-quality, art-directed images with strong control over style, mood, and composition — from 6 tokens per image on Prizmad. Write a prompt, optionally add up to 10 style references, set the creativity level, generate.
KREA 2 is an image model built for art direction: instead of just describing a subject, you steer the style, mood, and creative feel of the result. It's the tool to reach for when the brief isn't "a picture of a serum bottle" but "a serum campaign image with warm editorial lighting, a creamy orange-and-ivory palette, and clean negative space" — the kind of cohesive, brand-consistent visual a designer would sign off on.
The controls reflect that. A Creativity setting (Raw, Low, Medium, High) decides how literally the model follows your prompt versus how expressively it interprets it. You can upload up to 10 style reference images to pull the output toward a moodboard, an existing campaign, or a brand look — with a strength slider from -2 to +2, where negative values actively push the result away from the reference style. A seed field makes generations repeatable, and two quality tiers let you trade speed for fidelity: Medium at 6 tokens, Large at 12 tokens for maximum fidelity and art direction. Eight aspect ratios run from 9:16 vertical through 2.35:1 cinematic widescreen.
On Prizmad, KREA 2 uses the as , video, voiceover, captions, and . Your campaign images live in the same asset library as the rest of your creatives, ready to drop into video projects — and everything you generate is yours to use commercially.
Premium, art-directed key visuals for product launches — cinematic lighting, coherent palette, and composition worthy of a paid campaign.
Feed your brand's existing imagery as style references and generate new visuals that match the established look across every format.
Turn a folder of inspiration images into finished, on-brief creatives instead of briefing a designer and waiting a week.
Magazine-style beauty, fashion, and lifestyle product shots with expressive shadows and deliberate negative space.
Regenerate the same art-directed concept across eight aspect ratios to cover Stories, feed, banners, and video thumbnails.
Strong control over style, mood, lighting, and composition — built for campaign imagery that has to look designed, not generated.
Upload up to 10 images to guide the visual style and mood while your prompt defines the subject — turn a moodboard into finished creatives.
A -2 to +2 slider sets how strongly references influence the result; negative values push the image away from the reference style.
Raw, Low, Medium, or High — lower stays close to your prompt for precise briefs, higher produces more expressive interpretations.
1:1, 4:3, 3:2, 16:9, 2.35:1 cinematic, 4:5, 2:3, and 9:16 — from square feed posts to widescreen campaign banners.
An optional seed makes results reproducible, so you can regenerate a winning composition with small prompt tweaks.
Write a prompt that covers subject, composition, style, and mood — name the lighting, palette, and campaign feel, not just the product.
Pick a quality tier: Medium (6 tokens) is faster and cheaper for iteration; Large (12 tokens) delivers the highest fidelity and strongest art direction for finals.
Set the aspect ratio for the placement — 9:16 for Stories, 1:1 or 4:5 for feed, 16:9 or 2.35:1 for banners and covers.
Choose a creativity level. Start at Medium; drop to Low or Raw when the brief is strict, raise to High when you want the model to surprise you.
Optionally upload up to 10 style reference images and set the strength slider — +1 or +2 to match a moodboard, negative values to steer away from a look.
Click Generate. Results arrive in roughly 1–3 minutes and are saved to your asset library. Reuse the seed to iterate on a keeper.
Medium quality costs 6 tokens per image and is faster — ideal for iterating on prompts and styles. Large costs 12 tokens and delivers maximum fidelity and art direction for final creatives.
KREA 2 is unlocked on the same token balance you use for AI avatars, video, voiceover, and music — no separate KREA account, no API key setup.
If your tokens run out mid-campaign, buy a one-off top-up directly from the top-up page. Every image you generate stays yours with full commercial rights.
It runs on your Prizmad token balance — no separate KREA account is needed. Medium-quality generations cost 6 tokens per image, Large costs 12 (13 with style references).
6 tokens at Medium quality, 12 tokens at Large. Medium is faster and best for iteration; Large is the highest-fidelity tier for final campaign creatives.
Up to 10 uploaded images (PNG, JPG, or WebP) that guide the visual style and mood of the output while your prompt defines the subject. A strength slider from -2 to +2 controls their influence — negative values push the result away from the reference style.
It controls how literally KREA 2 follows your prompt. Raw and Low stay close to the exact brief; Medium is the balanced default; High produces more expressive, interpretive images. Use lower settings for strict brand work and higher ones for exploration.
Yes. You keep full commercial rights to every image generated on Prizmad — paid ads, landing pages, packaging mockups, and print included.
Eight: 1:1, 4:3, 3:2, 16:9, 2.35:1, 4:5, 2:3, and 9:16 — covering square feed posts, vertical Stories, and cinematic widescreen banners.
Yes. Set the optional seed field to an integer and generations become repeatable — keep the seed fixed while tweaking the prompt to iterate on a winning composition.
Image Generation gives you seven general-purpose models behind one prompt field, with product reference images on supported models. KREA 2 is a dedicated workspace specialized in art direction: creativity levels, up to 10 style references with a strength slider, a seed for repeatability, and a wider set of aspect ratios including 2.35:1. Use Image Generation for product-accurate shots, KREA 2 when the style itself is the brief — both are on the same token balance.
Published 2026-07-16 · Last updated 2026-07-16