AI Image Generator
Models
Nano

Google's cutting-edge image generation model

3 credits
Fast
High Quality
Gpt

OpenAI's GPT-4o image generation model

3 credits
High Quality
Text & Image
Seedream

ByteDance image generation and reference-based editing in 2K, 3K, or 4K

6 credits
2K/3K/4K
Text & Image
Flux

Flux Kontext standard model with balanced performance

4 credits
Fast
High Quality

Powered by ByteDance Seedream 5 Lite API

Public Visibility

When this option is enabled, the output image may be selected by AI Generate and published to the Explore.

Notification
This is a notification message
Sample Images
ByteDance image model on Lovart ME

Seedream 5 Lite AI Image Generator Move from a written brief or reference image to a usable visual

Create a new image from text or direct a reference-based edit in the same workspace. Choose from eight aspect ratios, render in 2K, 3K, or 4K, and export one PNG or JPEG for 6 credits.

4K
Maximum output
Choose 2K, 3K, or 4K
8
Aspect ratios
Square, portrait, landscape, and ultrawide
4
Reference slots
Current Lovart ME workflow limit
6
Credits per image
Same integer price in both modes

Use Seedream 5 Lite for clear production tasks

A useful request describes what will be delivered and what can be checked. Start with the destination, then specify the subject, composition, light, copy space, and anything that must remain unchanged.

Text-led image creation

Build a new visual from a written brief. State the intended placement, subject, scene, camera position, palette, lighting, and exclusions instead of relying on broad style labels.

Controlled image changes

Upload a source image and separate preservation rules from requested changes. Name the identity, geometry, pose, layout, materials, or lighting that must stay fixed.

Reference roles

Use different images for subject identity, product form, environment, palette, or material direction. Tell the model what each reference controls and remove images without a clear job.

Publication-aware framing

Choose the final ratio before generating. A vertical cover, square catalog tile, presentation visual, and ultrawide banner require different subject placement and negative space.

Choose output quality for the stage of work

The documented API maps Basic, High, and Ultra quality to 2K, 3K, and 4K. All three options cost 6 credits here, so choose based on review and delivery needs rather than price.

2K for prompt and composition checks

Use 2K while comparing framing, reference roles, color direction, and copy space. It is a practical first pass before you commit to a larger file.

3K for closer review

Use 3K when the direction is stable and you need a closer look at faces, product edges, material response, small props, or generated lettering.

4K for approved delivery

Choose 4K when the layout is approved and the publishing workflow benefits from a larger source. Resolution adds pixels; it does not replace prompt revision or final proofreading.

Match the eight ratios to the publishing slot

01

Square and balanced

Use 1:1 for catalog tiles, profile posts, centered products, and compact campaign assets.

02

Portrait and mobile

Use 2:3, 3:4, or 9:16 for covers, posters, people, stories, and full-screen mobile placements.

03

Landscape and editorial

Use 3:2, 4:3, or 16:9 for articles, slides, web headers, product scenes, and presentation visuals.

04

Ultrawide placement

Use 21:9 for banners and cinematic backdrops. Specify where the subject sits and which side must remain clear for copy.

Seedream 5 Lite prompt templates for real design work

Replace the bracketed fields, keep only instructions that affect the deliverable, and identify references by order. For edits, list what must stay before describing what should change.

Reference-based scene edit

Change the setting while protecting the required subject details.

Prompt Edit reference 1. Keep the subject identity, pose, clothing construction, proportions, and camera angle. Replace only the background with [new setting]. Match perspective, contact shadows, and reflected light. Remove [unwanted object]. Do not add people, logos, or text.

Working note: Write preservation rules before the requested change.

Editorial key visual

Create a visual that leaves deliberate room for a headline.

Prompt Create a [ratio] editorial key visual about [topic]. Main subject: [subject and action]. Framing: [shot and camera angle]. Leave an uncluttered copy area in the [location]. Palette: [colors]. Lighting: [description]. Keep symbols literal and relevant; no watermark or decorative lettering.

Working note: Describe the empty area as precisely as the subject.

Material and color variation

Test one product variation without redesigning its geometry.

Prompt Use reference 1 as fixed product geometry. Change only the finish to [material] and the colorway to [colors from reference 2]. Preserve dimensions, seams, controls, logo position, and viewing angle. Use a neutral studio background with realistic reflections and contact shadows.

Working note: Inspect seams, reflections, controls, and contact points at full size.

How to use Seedream 5 Lite on Lovart ME

Treat each generation as a short production brief. Change one or two decisions per round so you can tell which instruction improved the result.

  1. 01

    Choose text or image mode

    Use Text to image for a new concept. Choose Image to image when identity, geometry, composition, palette, or another visible property must come from a source image.

  2. 02

    Write the brief and assign references

    State the deliverable first, then add subject, setting, composition, lighting, copy space, preservation rules, and exclusions. Give each uploaded image one clear role.

  3. 03

    Set ratio, quality, and format

    Match the ratio to the destination. Choose 2K, 3K, or 4K for the review stage, then select PNG for lossless graphics or JPEG for a smaller photographic file.

  4. 04

    Generate, inspect, and revise

    Check identity, hands, product geometry, small text, edges, shadows, and unwanted additions. Revise specific constraints instead of rewriting the whole prompt at once.

Seedream 5 Lite questions

These answers reflect the documented Kie.ai parameters and the controls available in this Lovart ME workflow.

What is Seedream 5 Lite?

Seedream 5 Lite is a ByteDance image model available through separate Kie.ai text-to-image and image-to-image endpoints. This page combines both workflows in one generator.

Can Seedream 5 Lite edit an existing image?

Yes. Switch to Image to image, upload at least one reference, and describe both the changes and the visible details that must remain fixed.

How many credits does one image cost?

Text-to-image and image-to-image both cost 6 integer credits per generated image. The supplied upstream price is 5.5, which is rounded up because this site uses whole credits.

Which resolutions are available?

The API documents Basic at 2K, High at 3K, and Ultra at 4K. All three selections cost 6 credits in this integration.

Which aspect ratios can I use?

The available ratios are 1:1, 4:3, 3:4, 16:9, 9:16, 2:3, 3:2, and 21:9.

Which input and output formats are supported?

Reference uploads accept JPEG, PNG, and WebP files up to 10MB each. Generated images can be requested as PNG or JPEG.

How should I use several reference images?

Give each image one role, such as identity, product geometry, setting, palette, or material. This Lovart ME workflow shows up to four slots; the API documentation does not state that four is the model limit.

Can I use the result commercially?

Usage depends on Lovart ME terms, your rights to the uploaded references, and applicable law. Confirm source rights and review the current terms before commercial publication.