How to Use P-Image-Edit: Specs, Credit Cost and Prompting Guide
A practical guide to P-Image-Edit on VdoBloom: its real 1K dimension table, 10-reference limit, flat 6-credit price, prompting rules and when another editor is the better pick.
P-Image-Edit is Pruna AI’s fast multi-reference image editor on VdoBloom, and it costs a flat 6 credits — about $0.30 — per finished image, whether you feed it one reference photo or all ten. It runs on the prunaai:2@1 model through Runware, it is an editing model rather than a from-scratch generator, and its defining trick is that it keeps the identity, layout and lighting of the images you give it while changing exactly the thing you asked for. If your workflow is “here is a photo, change this one part of it,” this is one of the cheapest precise ways to do it in the catalogue.
What P-Image-Edit is best at
Three jobs suit it better than they suit anything else at this price. The first is single-instruction retouching: remove the object on the table, change the jacket to navy, swap the background to a plain studio sweep. Because the model auto-matches its output dimensions to your reference image in edit mode, you get back a file that drops straight into the same layout slot the original came from — no re-cropping, no regenerated framing.
The second is multi-reference composition. You can attach up to ten reference images in one request. That is what turns it from a retoucher into a compositor: a product shot plus a background plate plus a colour reference, described in one prompt, resolved in one generation. Models that accept a single image force you to chain several edits and pay for each one.
The third is volume. At 6 credits an image you can run a dozen variations of the same edit for the price of one premium 4K generation, which is the right shape for exploration — try eight framings of an idea, keep one, then re-render the keeper on a higher-end model if the final asset needs more resolution.
Where it is not the right tool: it is a 1K-class model. There is no 2K or 4K tier, no vector output, and no mask input. If the deliverable is a billboard or a print asset, generate the look here and re-render it elsewhere.
The real specs
In text-to-image mode the model works from a fixed dimension table — seven aspect ratios, all in the 1K region. In edit mode those numbers matter less, because output dimensions are matched to the reference image you supplied.
| Aspect ratio | Output size | What it suits |
|---|---|---|
| 1:1 | 1024 Ă— 1024 | Catalogue tiles, avatars, marketplace thumbnails |
| 4:3 | 1184 Ă— 896 | Classic landscape stills, slide imagery |
| 3:4 | 896 Ă— 1184 | Portrait feed posts, lookbook pages |
| 16:9 | 1376 Ă— 768 | Blog headers, video thumbnails, banners |
| 9:16 | 768 Ă— 1376 | Story and Reels stills, vertical ad frames |
| 3:2 | 1248 Ă— 832 | Photographic landscape crops |
| 2:3 | 832 Ă— 1248 | Photographic portrait crops, poster proofs |
Other specifics worth knowing: up to 10 reference images per request, a prompt field that accepts up to 2,000 characters, and an optional turbo setting that trades a little fine detail for speed. It also appears in the pose-variation flow, where it is one of the models allowed to take an existing photo and re-pose the subject.
What it costs, in credits and dollars
One credit is $0.05 at the Lite subscription rate of $15 for 300 credits, so the arithmetic is simple. Every P-Image-Edit generation is 6 credits regardless of aspect ratio, reference count or turbo. That flat structure is unusual — most editors in the catalogue tier their price by resolution — and it is the main reason this model is a good default for iteration.
| Image edit model | Credits per image | Cost at $0.05/credit | Reference images |
|---|---|---|---|
| P-Image-Edit | 6 | $0.30 | Up to 10 |
| Nano Banana Edit | 5 | $0.25 | Single-image style edits |
| FLUX Kontext Pro | 7 | $0.35 | Instruction editing |
| Seedream 4.5 Edit | 7 | $0.35 | Multi-image |
| Kling IMAGE O1 | 14 | $0.70 | Up to 10, 1K/2K |
| Nano Banana Pro (1K/2K) | 18 | $0.90 | High-control, up to 4K at 24 credits |
| Riverflow 2.0 Pro | 18 | $0.90 | Up to 10, 1K/2K/4K |
| Luma UNI-1 Max (2K) | 20 | $1.00 | Up to 8 references |
In practice: the 10 free credits every new account gets, with no card, cover one P-Image-Edit render with change to spare. A Lite month at 300 credits is 50 edits. A $2.49 credit pack of 75 credits — credits that never expire — is 12 edits. Full numbers for every plan are on the pricing page, and paid plans download without a watermark and with commercial rights.
How to prompt it well
P-Image-Edit rewards instructions, not descriptions. The failure mode people hit is writing a full scene description — “a woman in a red dress standing in a sunlit kitchen” — when the model already has the woman, the dress and the kitchen in front of it. Describe the delta instead.
- Name the target, then the change. “Change the dress from red to deep emerald green, keep the fabric texture and folds” beats “emerald dress.”
- State what must not move. Adding “keep the face, pose, lighting and background unchanged” measurably reduces drift on identity-sensitive edits.
- Give each reference a role. With several images attached, say which is which: “use the first image as the subject, the second as the background plate, the third for colour grading only.”
- One idea per generation. Two unrelated instructions in one prompt tend to produce one good change and one ignored one. At 6 credits, chaining two edits is cheaper than fighting one prompt.
- Leave turbo off for final renders. It is a speed setting, and detail is the first thing it spends.
A worked example
Say you have a product photo of a ceramic mug on a cluttered desk and you need a clean listing image plus a lifestyle version.
- Open the image editor and choose P-Image-Edit.
- Upload the mug photo. Set the aspect ratio to 1:1 for the listing tile.
- Prompt: Remove every object except the ceramic mug. Replace the desk with a seamless light grey studio backdrop with a soft shadow beneath the mug. Keep the mug’s glaze, handle shape and label exactly as they are. Generate — 6 credits.
- For the lifestyle version, attach the original mug photo plus a reference photo of a kitchen counter. Switch to 3:4. Prompt: Place the mug from the first image onto the counter in the second image, matching the second image’s warm morning light and casting a correct shadow. Generate — another 6 credits.
Two finished assets for 12 credits, or $0.60. If either needs to be larger than 1K for print, re-run the winning prompt on a 2K or 4K model rather than upscaling.
One rule that applies to any edit involving a real person: you need that person’s consent before you edit and publish their likeness. Explicit and illegal content stays blocked regardless of model.
When to pick a different model
Pick Kling IMAGE O1 at 14 credits when you need 2K output with the same ten-reference ceiling. Pick Nano Banana Pro at 18 credits (24 for 4K) when the edit is the final deliverable and resolution is non-negotiable. Pick a dedicated erase or outpainting model when you have a mask or need to extend a canvas — P-Image-Edit takes no mask input. And pick a text-to-image model outright when there is no source photo to preserve, because an editor asked to invent a whole scene is working against its own design.
If you are coming from a photo-first workflow and want the guided version rather than the raw model picker, the AI image editor from photo page walks the same pipeline with presets. Full specs live on the P-Image-Edit model page.
Frequently asked questions
How much does one P-Image-Edit image cost?
6 credits, flat, for every generation — about $0.30 at the Lite rate of $15 for 300 credits. Aspect ratio, reference count and turbo mode do not change the price.
How many reference images can I attach?
Up to 10 in a single request. They all go into one generation, so composing from several sources costs the same 6 credits as a one-image retouch.
What resolution does P-Image-Edit output?
It is a 1K-class model. Text-to-image mode uses a fixed table topping out at 1376 × 768 for 16:9; in edit mode the output is matched to your reference image’s dimensions. There is no 2K or 4K tier.
Does P-Image-Edit support masks?
No. Targeting is done in the prompt, by naming the region or object you want changed. For mask-driven work, use an erase or outpainting model instead.
Can I try it without paying?
Yes. New accounts get 10 free credits with no card, which covers one P-Image-Edit render. Credit packs start at $2.49 for 75 credits and never expire; subscriptions start at $15 a month and remove the watermark.
Ready to try it?
Create your first AI video in minutes — no credit card required.
Start Creating Free →