Recraft's New Product Model Targets Ecommerce Directly — Early Tests
Recraft's new product model is a generative AI image system purpose-built for ecommerce catalog photography, focused on rendering packshots, lifestyle scenes, and isolated product shots with consistent lighting across large inventories. This matters for ecommerce sellers because catalog imagery now drives the majority of conversion decisions, and most brands still rebuild thousands of product photos by hand every quarter.
When Recraft rolled out its third-generation product model in early 2026, the framing shifted away from generic image generation toward a tighter commercial use case. Rather than chasing benchmark scores on aesthetic prompts, the team trained the model on catalog conventions: clean edges, neutral backgrounds, accurate color, and predictable placement of text or labels. The result is a tool that wants to live inside a Shopify admin, not a creative playground.
What the new model actually changes
The headline upgrade is consistency. Earlier generative tools could render a product image once but struggled to repeat the same object from new angles or with new backdrops. Recraft's product model addresses this through what the company calls "object identity," a way of holding onto the shape, color, and proportions of a reference product across many generations. Sellers can upload a single hero shot and produce a set of supporting images that look like they came from the same photo shoot. The model also added a structured output mode that locks aspect ratio, background color, and lighting direction, which matters when an entire catalog has to feel visually consistent across hundreds of pages.
Equally important is the model's handling of text and packaging. Bottle labels, box copy, and taglines have long been a weakness of generative tools, which scramble letters or invent words. Recraft rebuilt the tokenizer to read short commercial strings, and the early demos hold up under scrutiny. For brands selling cosmetics, supplements, and food, that single improvement removes the most expensive manual cleanup step in any AI workflow. The team also rebuilt the segmentation head to deliver cleaner alpha channels, which means the cutouts actually work on transparent backgrounds without a second pass through Photoshop.
Early test results from ecommerce teams
Several beta teams have published numbers since the model entered closed testing in late 2026. A mid-market apparel brand reported cutting its catalog refresh cycle from six weeks to nine days, while a home goods seller produced 4,200 SKU images in a single weekend sprint. The aggregate signal is that batch generation works, even if individual outputs still need a human review pass. None of the teams in the beta reported skipping review entirely, and most added at least one full-time reviewer to their catalog team to handle the increased volume.
Conversion lift is the figure that gets attention, but the more honest metric is cost per approved image. Most beta testers landed somewhere between $0.18 and $0.42 per usable asset, including reviewer time. That sits well below the $8 to $25 range for traditional product photography reported by Shopify's enterprise conversion research, though it does not yet match a real shoot for hero images on flagship product lines. The gap closes further when factoring in studio rental, prop sourcing, and reshoot cycles, which most brands under-count in their baseline numbers.
Quality control remains the open question. Reviewers in the beta flagged hallucinated logos, drifting color temperatures, and occasional geometry errors on reflective packaging. These are not deal-breakers, but they confirm what most catalog teams already know: AI output is a starting point, not a finished asset. Teams that built the review step into their pipeline from day one shipped faster and with fewer customer complaints than teams that tried to automate the QA layer too.
How Recraft compares to Rewarx for everyday catalog work
For sellers who need more than raw generation, the workflow around the model matters as much as the model itself. Rewarx bundles generation, background removal, and mockup rendering into a single ecommerce-focused toolkit, while Recraft remains closer to a generation-first product with third-party integrations for downstream steps. The practical difference shows up the moment a team tries to push 500 SKUs through one afternoon. A generation tool with no built-in cleanup forces the team to glue three or four extra services together, which is where most of the time savings disappear.
| Feature | Recraft | Rewarx |
|---|---|---|
| Object identity across variants | Yes | Yes |
| One-click AI background remover for product cutouts | Plugin required | Built in |
| Lifestyle mockup generator for ecommerce listings | Limited | Included |
| Studio-quality product photography studio tools | Prompt dependent | Built in |
| Batch SKU processing | API only | Drag and drop |
The takeaway for ecommerce teams: Recraft is a strong choice for design teams who want fine-grained prompt control, while Rewarx fits the operations team that needs to push 500 SKUs through a clean pipeline before lunch. Both are real products with real customers, and the right call depends on what your team actually does all day.
A practical workflow for sellers adopting the new model
Sellers testing Recraft in production tend to follow a tight five-step loop. The order matters because each step protects the next from bad inputs, and skipping ahead almost always means redoing the work later. The same workflow works whether you choose Recraft, Rewarx, or any other ecommerce-grade image tool.
- Audit the existing catalog and group SKUs by product family, material, and finish.
- Generate three reference shots per family using a real studio photo as the anchor image.
- Use the object identity feature to spin out supporting angles, lifestyle scenes, and color variants.
- Run every output through a quality checklist covering edge cleanliness, label accuracy, and color match.
- Push approved images directly to Shopify, BigCommerce, or your PIM through the relevant integration.
The fastest catalog teams in 2026 are not the ones with the best prompts. They are the ones with the tightest review checklist.
Pre-flight checklist before you ship AI images
- ✓ Every label text matches the source product exactly, character for character
- ✓ Color values fall within acceptable Delta E tolerance for the product category
- ✓ Edges are clean against the background, with no fringing or halo artifacts
- ✓ No hallucinated logos, brand marks, or competitor names appear in the output
- ✓ Aspect ratio matches the platform requirement for the destination channel
Recraft has signaled that the model will continue to be retrained on commercial datasets through 2026, with planned improvements in texture fidelity and a tighter API for direct PIM integration. For sellers deciding between Recraft and a bundled ecommerce tool like Rewarx, the right call depends on whether the team's bottleneck is creative exploration or catalog throughput. Both will get better over the next 12 months, and the brands that win will be the ones who pick a tool and actually use it.
Frequently asked questions
What is Recraft's new product model?
Recraft's new product model is a generative AI image system designed specifically for ecommerce catalog work, with stronger object identity, better label and packaging handling, and a workflow oriented toward batch SKU processing rather than single-image creation. It launched in early 2026 and is currently available through Recraft's web app and API.
How much does it cost to use Recraft for ecommerce images?
Beta testers in 2026 reported an average cost between $0.18 and $0.42 per approved image once reviewer time is included, which sits well below traditional product photography but slightly above fully automated alternatives that skip the human review step. Recraft's public pricing tiers start at a free trial and scale by generation volume.
Can Recraft replace a real product photo shoot?
For supporting catalog angles, color variants, and lifestyle scenes, Recraft performs close to a real shoot in early tests. For flagship hero images on high-margin SKUs, most brands still prefer a real shoot because AI output remains slightly less predictable on reflective or complex packaging, and the marginal cost of a real shoot is easier to justify at flagship pricing.
Is Recraft better than Rewarx for ecommerce sellers?
Recraft is stronger for design teams who want prompt-level control and creative flexibility. Rewarx is stronger for operations teams who need a complete pipeline including background removal, mockup generation, and studio-quality photography in one tool. The right choice depends on whether your bottleneck is creative ideation or catalog throughput.
Skip the prompt engineering
Rewarx turns one product photo into a full catalog set: clean cutouts, lifestyle mockups, and studio lighting, without the review bottleneck. Built for ecommerce teams that need volume.
Try Rewarx Free