Seedream 5.0 Lite Guide: Prompts, Features and Examples
Seedream 5.0 Lite on VdoBloom: 6 credits an image, 2K or 4K output, up to 14 reference images, and prompts that show what it does well.
Seedream 5 Lite is ByteDance’s lighter image model on VdoBloom, and it costs 6 credits per image in both of its modes: text to image for generating from a prompt, and image-to-image for editing or combining photos. It renders at 2K on basic quality or 4K on high, supports eight aspect ratios, and its image-to-image mode accepts up to 14 input images in a single instruction.
Updated October 2026.
Disclosure: VdoBloom is our platform.
What Seedream 5 Lite is
Seedream 5 Lite is the entry point to ByteDance’s Seedream 5 generation on VdoBloom’s image generator. It runs as two entries in the model picker: Seedream 5 Lite Text-to-Image, described in the platform’s own catalogue as photorealistic text-to-image at 2K or 4K, and Seedream 5 Lite I2I, which takes a written instruction plus up to 14 images at the same two resolutions.
The Lite tag is about position in the lineup rather than a cut-down feature set. It is the lowest-priced way to run a Seedream 5 generation on the platform, and it keeps the two things the family is known for: convincing light and material rendering, and enough input capacity to build a composition from many sources. VdoBloom’s model picker scores it 4.6, and it carries the HOT content label, the most permissive filtering tier.
What it is good at
- Photorealistic images at a low price. At 6 credits an image it is one of the cheaper routes to a photoreal result in the roster, and the price does not change between 2K and 4K.
- Many input images. Fourteen references in one request is the highest capacity among the Seedream models on the platform, and it suits jobs like dressing one product across several scenes or placing one person into a set of locations.
- Instruction-led editing. The image-to-image mode reads a change instruction and applies it while holding the rest of the frame, which makes iteration cheap: fix one thing instead of regenerating.
- Flexible framing. Eight ratios from 1:1 through 21:9, so it covers square social crops and ultrawide banners without a second tool.
- Fast iteration. Because 4K costs the same as 2K, you can test the composition at high quality without a pricing penalty.
Credits and specifications
| Setting | Seedream 5 Lite Text-to-Image | Seedream 5 Lite I2I |
|---|---|---|
| Cost per image | 6 credits | 6 credits |
| Quality tiers | Basic (2K) or High (4K) | Basic (2K) or High (4K) |
| Input images | None (prompt only) | Up to 14 |
| Aspect ratios | 1:1, 4:3, 3:4, 16:9, 9:16, 2:3, 3:2, 21:9 | Same eight |
| Prompt limit | 3,000 characters | 3,000 characters |
| File types | — | JPEG, PNG or WebP, up to 10 MB each |
The flat 6 credits applies at both quality tiers, so high quality at 4K is the same price as the 2K basic render. Ten free credits come with a new account and no card is required, which covers a single image with 4 credits left over; the pricing page has the plan and credit-pack details if you are planning a batch.
Six example prompts
- Photoreal product shot. “A leather-bound notebook on a walnut desk beside a cup of black coffee, morning light through a window on the left, visible paper texture and a scuffed cover, 50mm, shallow depth of field, no text.”
- Person in a scene. “A woman in a linen shirt sitting on the steps of a whitewashed house, midday sun, hard shadows on the wall behind her, natural skin texture, candid framing, 35mm.”
- Food, overhead. “Overhead shot of a bowl of ramen on a dark slate table, steam catching the light, soft-boiled egg halved, chopsticks resting on the rim, natural window light, no props beyond the bowl.”
- Wide banner. “An empty mountain road in 21:9 at first light, fog in the valley, cool blue tones, no people, no text, room at the top for a headline.”
- Instruction edit. “Replace the plain background with a soft grey studio wall and add a natural shadow beneath the subject. Keep the subject’s face, pose, clothing and lighting direction exactly as in the photo.”
- Multi-image composition. “Put the woman from the first image into the kitchen from the second image, seated at the table, matching the kitchen’s camera height and light direction. Keep her face and hair identical to the first image and change nothing else in the room.”
On multi-image jobs, order and role matter: say which image supplies the person, which supplies the room, and what must not move. Vague instructions with many references are where likenesses drift.
Two habits that improve every prompt
The first is to write the change before the description. An edit instruction such as “replace the background with a soft grey studio wall” gives the model one job, and the sentence after it can then be spent on constraints rather than on restating the scene. Prompts that open by describing the whole picture tend to return a picture that has been reshot rather than edited, and every part of the frame moves along with the part you actually wanted changed.
The second is to name what must not change, explicitly and in that many words. Faces, hair, pose, clothing, camera height and light direction are the six things that most often drift, so listing them costs a line and saves a regeneration. When a result is nearly right, resist running the prompt again from scratch: instruct the change on the version you kept, one fix at a time, and the composition stays where you left it.
Limits worth knowing
- Prompt length: 3,000 characters in each mode.
- Input images: 14 in I2I mode at up to 10 MB each; text-to-image takes none.
- No negative prompt field. Anything to avoid has to be handled by writing the frame without it.
- No seed control in the interface, so exact reproducibility is not available.
- Two quality steps only. Basic 2K or high 4K, with nothing in between.
- Fidelity drops as references pile up. If one face or logo has to be exact, use fewer input images rather than the full fourteen.
Seedream 5 Lite or another model?
| Model | Credits per image | Reach for it when |
|---|---|---|
| Seedream 5 Lite | 6 | You want photoreal 2K or 4K output and up to 14 reference images at a low price |
| Seedream 4.5 | 7 | You prefer the previous generation’s handling, at one credit more |
| GPT Image 2 | 3 at 1K, 5 at 2K, 8 at 4K | You want the lowest price in the roster and can work at 1K to start |
| Nano Banana Pro | 18 at 1K and 2K, 24 at 4K | You need the largest output tier for a hero image |
If you are working through several models rather than one, the multi-model generator page explains how the roster is organised, and prompt structure is the most useful general upgrade for any of them.
Frequently asked questions
How much does Seedream 5 Lite cost on VdoBloom?
6 credits per image, flat, for both text-to-image and image-to-image, at either quality tier. A 4K render costs the same as a 2K one, and the price is shown before you generate.
Can Seedream 5 Lite edit my own photo?
Yes, through its image-to-image mode. Upload the photo as a reference, write the change as an instruction, and add a clause naming what must stay the same. It accepts up to 14 input images in one request, which also covers combining several photos into one picture.
Is Lite worse than the full Seedream 5 models?
Lite is the cheaper entry in the Seedream 5 generation rather than a reduced feature set: it still renders at 2K or 4K, supports all eight aspect ratios and accepts up to 14 references. What you trade is the position in the lineup, which is what the lower per-image price reflects.
Why does my edit lose the likeness when I add more photos?
Fidelity to each individual reference drops as you add more of them. Stay within a handful of references when a specific face or logo has to stay exact, and state each image’s role explicitly in the prompt — which one is the person, which one is the room.
What is the lowest-cost way to test a prompt on it?
Render one 2K image on basic quality for 6 credits. Because quality tiers are priced identically, there is no saving in drafting at low resolution; test the prompt and the composition first, then change one thing at a time on the version you kept.
Open Text to Image, choose Seedream 5 Lite, and run one 6-credit test at 2K before committing to a larger batch — the quality tiers cost the same, so nothing is lost by starting at the top. Seedream 5 Lite I2I is documented in full if your job is editing rather than generation, and with 10 free credits at signup the first image costs nothing at all.
Ready to try it?
Create your first AI video in minutes — no credit card required.
Start Creating Free →