Separate Exploration From Delivery
Pixelcut documents num_variations as a way to produce multiple variations in one generation call, with a stated range of 1 to 4. That makes it useful for exploring alternatives while the output shape and brief stay fixed.
Aspect ratio is a separate generation parameter. Pixelcut specifically describes separate runs for producing the same concept in 1:1, 16:9, and 9:16. Treat those as separate deliverables, not as crops that are automatically equivalent.
- Choose variations when the question is: which composition best fits this one format?
- Choose separate runs when the question is: how should the concept be composed for each channel shape?
- Do not let the strongest variation in one ratio stand in for review of the other ratios.
Use A Two-Axis Decision
A practical planning aid is to score the job on two independent axes: exploration breadth and delivery breadth. Exploration breadth concerns how many alternatives you need within one format. Delivery breadth concerns how many formats the campaign must publish.
If exploration is high but delivery breadth is low, one call with several variations may be appropriate. If delivery breadth is high, plan one run per required ratio, even when the prompt describes the same scene. Reuse the same approved product references and creative requirements, then inspect each returned composition independently.
Pixelcut's documented sequence also matters: its agent workflow calls models with the list action before selecting a model for generate_media. Record the selected model with each run so later comparisons do not silently mix model choices.
- Exploration axis:
num_variationsfrom 1 through 4. - Delivery axis: each required
aspect_ratio, such as1:1,16:9, or9:16. - Control fields: product references, prompt, model, media type, and output format.
- Decision evidence: the actual returned image for each ratio, not the request text alone.
Keep the Campaign Record Comparable
For a fair comparison, hold the product identity brief and scene intent constant while changing only the ratio when testing channel adaptations. If the prompt, references, or model also change, you are comparing multiple variables and cannot tell whether the composition changed because of the ratio or because the input changed.
Use the same acceptance questions for every output: Is the correct product represented? Is the main subject positioned safely within the format? Does the scene still communicate the intended selling point? These are original review criteria, not Pixelcut guarantees. Generated imagery remains a creative asset and should not be treated as physical-product evidence.
When a ratio fails, decide whether the problem is format-specific composition or a broader concept problem. A format-specific issue justifies revising that ratio's prompt. A broader identity or scene failure justifies revisiting the shared brief before producing more variations.
Worked Hypothetical: Three Channel Assets
Hypothetically, a seller needs a stainless-steel bottle concept for a square marketplace image, a website banner, and a vertical social placement. The seller has two approved product reference images and wants to explore two compositions per channel.
The correct plan is six planned outputs: two variations for each of three separate aspect-ratio runs. The seller keeps the product description, references, model, media type, and output format aligned, while adapting composition language for each ratio. This is a planning illustration only; no Pixelcut request or result has been executed.
| Need | Pixelcut planning choice | Acceptance evidence |
|---|---|---|
| Square marketplace image | One run at 1:1, num_variations: 2 | Compare product visibility and usable negative space |
| Website banner | Separate run at 16:9, num_variations: 2 | Confirm the product remains prominent in the wide composition |
| Vertical social placement | Separate run at 9:16, num_variations: 2 | Confirm the subject and key selling point survive the tall layout |
Where CartVisual fits
Use CartVisual's product-photography workflow for a non-API production path: provide the product description, creative requirements, references, and supported output choices, then review each intended format separately. CartVisual does not implement Pixelcut's generate_media API.
Source and evidence scope
Prepared by the CartVisual editorial team with AI assistance. Competitor behavior is documented rather than independently executed for this guide.
Official source snapshot checked .
Pixelcut's official Image Generation documentation for its agent-oriented models and generate_media workflow. It documents request parameters and an aspect-ratio workflow, but does not establish image quality, marketplace acceptance, pricing, rights, or CartVisual execution.