Z-Image Generator

The budget model in the AI photo generator, and the whole case for it is price. Open the photo tool, write a prompt, generate four images, spend four credits.

Three of those rows are real constraints. There is no reference dropzone, so nothing you own goes into the image. There is no 2K or 4K. And the 900-character ceiling rejects the run outright if you go over, so watch the live counter.

In return you get speed and price: four images in the time the others take for one, at a quarter to a third of the credits.

A credit is about $0.10 on the $49.99 plan for 500 credits, and about $0.067 on the $199.99 plan for 3,000 credits.

Two comparisons make the point. A four-image run costs the same 4 credits as two Nano Banana 2 images at 1K, so you get twice the attempts. And one hundred drafts costs 100 credits, a fifth of the $49.99 plan's monthly allowance.

This is also why an AI creator portrait costs 2 credits on Nano Banana 2 and 1 credit on Z-Image in the AI influencer generator: so you can look at ten faces before you commit to one.

Plans and pack sizes are on the pricing page.

Nine hundred characters is roughly 130 to 150 words: plenty for a scene, not enough for a manifesto. Anything longer is rejected, so the discipline is forced.

Keep four things, in order: the subject, the surface it sits on, the light, the frame.

Three savings. Drop the negative list: "no blur, no distortion, no watermark" eats 60 characters and does little. Drop the gear talk: "shot on a full-frame mirrorless with an 85mm f/1.4" becomes "shallow depth of field". And do not describe text, because every character spent on a label is wasted twice.

Background prompts are naturally short. A gradient, a counter, a gym, a street at night: each fits in 200 characters with room for the light. A prompt running long usually means the job belongs on Nano Banana 2.

All under the 900-character ceiling, at 1K, written for the jobs this model does well.

1. A studio gradient backdrop "A smooth studio backdrop in warm sand and dusty rose, a soft vertical gradient from light at the top to deep at the bottom, gentle falloff into the corners, evenly lit, no objects and no text, vertical 9:16." Why: "no objects and no text" earns its characters. An empty frame is the point.

2. An empty kitchen counter "An empty sunlit kitchen counter in pale oak with a white tiled wall behind it, morning light from the left casting a soft window shadow across the surface, shallow depth of field, nothing on the counter, vertical 9:16." Why: an empty surface is where a product goes later, so name the light you must match.

3. A wide empty gym "A wide empty gym interior at dawn, rubber flooring, racks of weights along the far wall, cool grey daylight through high windows, faint dust in the air, slight haze, no people, vertical 9:16." Why: "no people" is the most useful phrase in a background prompt.

4. Rainy window bokeh "A night city street seen through a rainy window, out-of-focus red and amber lights, water beads on the glass, deep blue shadows, heavy bokeh, nothing sharp in the frame, vertical 9:16." Why: a fully defocused frame is the safest background: no detail left to get wrong.

5. Golden hour beach "An empty beach at golden hour, wet sand reflecting a low warm sun, gentle surf lines, a pale sky with thin cloud, long soft shadows from the left, no people and no objects, vertical 9:16." Why: warm light on wet sand gives contrast, which keeps white hook text readable.

6. An empty desk from above "An overhead view of an empty warm oak desk surface, faint wood grain, a soft pool of lamp light from the upper right, dark falloff at the edges, nothing on the desk, square 1:1." Why: this is the flat-lay base you drop props onto in a later, dearer run.

7. Paper texture "A close-up of textured cream paper, visible fibers and a soft crease running diagonally, raking side light from the left picking up the grain, neutral warm tone, no text, 4:5." Why: paper and linen textures make reusable slide backgrounds that read as real.

8. A dark frame for a hook line "A dark charcoal gradient background with a faint teal glow in the lower left corner, a heavy vignette, subtle film grain, deep shadow across the upper half, no objects and no text, vertical 9:16." Why: a dark frame with one light source is the best surface for a bold white hook.

A ninth, for nature b-roll: "A forest trail in early morning, mist between tall pines, a damp earth path leading away from the camera, cool green light, soft focus in the distance, no people, 9:16." Mist hides the detail this model is weakest on.

Four jobs, and it is genuinely the best choice for all four.

Slideshow backgrounds. The AI slideshow generator builds TikTok photo-mode posts and carousels of 5 to 10 slides at 1080x1920, with the text burned on, at 1 credit per slide. That background does not need 4K, and a defocused frame reads better under a headline than a busy one.

B-roll stills. Cutaway frames, establishing shots, scenery between talking beats. Nobody pauses on these. Pay 1 credit.

Drafts before an expensive run. The highest-value use. Generate four here for 4 credits, pick the composition that works, then re-run that wording on Nano Banana 2 with your product references attached.

Volume. When you need thirty frames and none of them individually matters, the arithmetic decides: 30 credits here against 60 or more anywhere else.

What it is not for: your actual product. Without a reference dropzone, a Z-Image "product shot" is a plausible item from the category, not yours. That work belongs on the AI product photo generator, and the TikTok slideshow generator covers the posting side.

Start here every time. Upgrade only when one of the triggers below fires.

The verdict: default to Z-Image, and treat every upgrade as a decision you justify. Most of what a short-form workflow eats is scenery, and scenery at 1 credit in ten seconds is the right answer.

The four triggers that make the extra credit worth paying:

1. Your product has to be the product. No reference dropzone, so go to Nano Banana 2.
2. A person is the subject. Faces and hands are where cheap models show their limits.
3. The image contains words. GPT Image 2.5, and check every letter.
4. The output needs more than 1K. A hero banner or a print piece needs 2K or 4K, which this model does not offer at all.

If none apply, the extra credit buys nothing you will see in a feed.

A short list, worth having in one place before you plan around it.

• No reference images at all. Text prompt only. You cannot upload your product, your model or your last frame.
• No 2K and no 4K. 1K is the ceiling. To crop in hard or print, generate elsewhere.
• A hard 900-character prompt limit. Go over and the run is rejected rather than truncated. Watch the counter.
• Only five aspect ratios: 9:16, 16:9, 1:1, 4:3 and 3:4. No 4:5, 3:2 or 2:3, so the tallest Instagram and Facebook feed crop must come from another model.
• No background removal on an uploaded photo, anywhere in the tool, and no transparent export.
• No brand font control and no exact color matching. No typeface upload, no hex field.
• No guarantee a logo or a label renders correctly. This model is the weakest of the three at text, so keep words out of the frame. If lettering slips into an image you plan to use, read it at full size and regenerate when it is wrong.

MakeViral does not upload, schedule or post your images either, and there is no API. For the surrounding workflow, see the Shopify product photo guide and the TikTok slideshow strategy guide.