Colorways From One Sample Photo: Shoes, Bags, Hoodies, Eyewear

Colorways from one sample photo let a brand see a product in every candidate color before paying for a sample of each. We took one packshot each of a sneaker, a crossbody bag, a hoodie and a pair of acetate glasses and asked Visualizee, an AI renderer that edits an uploaded image while keeping the product in it, for three colorways of each. Most renders took 14 to 21 seconds. Eleven of 12 colorways kept the product's shape, stitching, light and camera; four of those picked up a small extra detail you would retouch. One, the crystal clear frame, failed on both models we tried. Every input, what we asked for, the times and the credits are below.

Quick answer: Accept one clean packshot of the sample, then ask for each colorway from that file with a list of what must not change: shape, stitching, hardware, light, camera. Name only the color or material that changes. Opaque color and leather-to-suede swaps hold well. Transparent materials rebuild hidden parts, and every color is approximate, so approve the real color on a physical swatch.

The Inputs: One Accepted Packshot per Product

The input is one packshot per product, and every colorway is made from that one file, never from scratch. We used four unbranded products shot the same way: side profile for the sneaker, front for the bag, flat-lay for the hoodie, three-quarter for the glasses, all on warm grey paper in soft light from the upper left.

These four are generated stand-ins, so no real product, logo or factory sample is shown. With your own product, the input is a photo of the first sample or the prototype: a phone on a tripod, a sheet of paper behind it, window light. What matters is that the photo is the one you would happily put on the product page, because every colorway inherits its camera and its light.

Each base took one render (14 to 16 seconds, 60 credits) and was accepted on the first try. We checked each one for what it had to keep in every variant: the sneaker's seven eyelets and its double-stitched quarter seam, the bag's turn-lock and two D-rings, the hoodie's raglan seams and pocket, the glasses' two silver rivets on each front corner.

All 12 Colorways at a Glance

Eleven of the 12 colorways kept the product; the clear eyewear frame did not, on either model. The table lists what we asked to change, the model, the measured server-side render time and the credits for each.

Product	Colorway	What we asked to change	Model	Time	Credits	Verdict
Sneaker	Forest green suede	Toe cap and heel panel to forest green suede only	Nano Banana Pro	21 s	150	Pass
Sneaker	Black on white sole	Leather, suede and laces to black; sole stays white	Nano Banana Pro	18 s	150	Pass with retouch: faint crease mark added on the quarter
Sneaker	Navy, tan, gum sole	Navy leather, tan suede, gum rubber sole	Nano Banana Pro	20 s	150	Pass with retouch: same crease mark
Crossbody bag	Black	Leather and strap to glossy black; brass stays brass	Nano Banana Pro	19 s	150	Pass
Crossbody bag	Oxblood	Leather and strap to oxblood	Nano Banana Pro	18 s	150	Pass
Crossbody bag	Sage suede	Body to sage suede, strap to matching smooth leather	Nano Banana Pro	19 s	150	Pass
Hoodie	Forest green	Fabric color only; white cords stay	Nano Banana Pro	20 s	150	Pass with retouch: small blank neck label added
Hoodie	Ecru	Fabric and cords to ecru	Nano Banana Pro	20 s	150	Pass
Hoodie	Washed black	Faded, garment-dyed black	Nano Banana Pro	117 s	150	Pass with retouch: one metal cord eyelet added
Eyewear	Tortoiseshell	Acetate to dark tortoiseshell	Nano Banana 2.1	17 s	60	Pass
Eyewear	Olive	Acetate to translucent olive	Nano Banana 2.1	14 s	60	Pass
Eyewear	Crystal clear	Acetate to clear, core wire showing	Nano Banana 2.1, then Pro	44 s + 19 s	210	Fail on both models: rivets lost

The whole set, base packshot in the left column and three colorways to its right:

With the four bases and the one re-run, the set took 17 renders and 1,920 credits. The 117-second hoodie render is an outlier: its two siblings took 20 seconds each, so treat it as queue time rather than a sign of a harder job. We checked every colorway against its base for four things: the same side of the product, hardware and stitching kept, no mirroring, and no seam where the color changes.

Sneaker: Three Colorways From One Side Profile

The sneaker held its shape, eyelets, stitching and sole profile in all three colorways, including a full sole swap from white to gum rubber. Two of the three added a faint crease mark on the quarter panel that the base does not have.

What we typed for the green one, in plain words: "Keep exactly as in the reference: the shoe's shape and side profile with the toe pointing left, every panel line, all stitching, the seven eyelets, the laces, the white cupsole, the background, the light, the shadow and the camera. Change only the color of the suede toe cap and suede heel panel to deep forest green suede."

The green colorway is the cleanest of the 12, because it changed the least: two suede panels. The black and the navy versions changed the whole upper, and both added a small diagonal crease texture in the middle of the quarter, where the base is smooth. At thumbnail size it can read as a mark or a logo, which is exactly what you do not want on a product page. A two-minute retouch removes it; so might a re-run. We left it in so you can see what to look for.

The navy version is also the material test: the sole went from white rubber to gum rubber and kept its height, its textured sidewall and its toe bumper. Material swaps on solid parts behaved like color swaps.

Crossbody Bag: Black, Oxblood and Sage Suede

All three bag colorways passed: the brushed brass turn-lock, the two D-rings, the saddle stitching around the flap and the strap's path are the same in every image. This is the easiest product type in the set, because the hardware is meant to stay the same while the leather changes.

The sentence that keeps it right is the one about the hardware: "The brass hardware stays brass." We did not test the request without it, but naming the hardware finish costs one sentence and removes the chance that a "black bag" comes back with gunmetal or silver hardware. If you are deciding hardware finish as well, do it as a separate colorway, so you can see leather and metal choices one at a time.

The sage suede is a material change, not only a color change: the smooth leather became a napped surface with soft highlights, and the strap stayed smooth leather because we asked for it. Expect suede and nubuck renders to look slightly more uniform than real hides, which carry marks and nap direction.

Hoodie: Forest Green, Ecru and Washed Black

The hoodie's raglan seams, kangaroo pocket, ribbed cuffs and the fold of the sleeves held in all three colorways. Two of them added a small detail: a blank white label inside the neck on the green one, and a metal eyelet where one cord leaves the hood on the washed black one.

What we typed for the washed black one: "Keep exactly as in the reference: the hoodie's shape and position, the fold of the sleeves, the hood, the kangaroo pocket, the ribbed cuffs and hem, all seams and flatlock stitching, the fleece texture, the white drawcords with metal tips, the background, the light, the shadows and the overhead camera. Change only the fabric color from heather grey to a faded washed black, slightly uneven like garment-dyed cotton."

Flat-lays are forgiving, because the camera is straight down and the garment has no hidden side to invent. The fleece texture carried over in every colorway, and the washed black one shows the uneven, faded look of garment-dyed cotton that we asked for, not a flat digital black.

The added label and eyelet are the kind of drift that a quick look misses and a buyer notices. Neither is a problem on a mood board for a line review. On a product page they would be: a customer buying the green hoodie would expect a neck label that the grey one does not have. Check trims, labels and cord ends in every garment colorway.

Eyewear: The Cheaper Model, and the Clear Frame That Failed

We ran the glasses on Nano Banana 2.1, the cheaper model, to see if it holds. It did on two tinted colors: dark tortoiseshell and translucent olive kept the frame shape, the rivets and the hinges. It failed on crystal clear acetate, and Nano Banana Pro failed the same way.

The clear frame is a different kind of job. Once the acetate is transparent, the parts inside it should show through: the hinge plate, the core wire in the temple, the back of the rivets. Both models drew those, convincingly, and in doing so changed the front corners. On the right corner, the rivet dots moved back next to a hinge that now shows through the plastic; on the left corner they are simply gone.

What we typed for the clear frame: "Keep exactly as in the reference: the frame shape, the three-quarter view with temples open, the two silver rivet dots on each front corner, the silver hinges, the clear lenses, the background, the light, the shadow and the camera. Change only the acetate from opaque black to fully transparent crystal clear acetate, so the metal core wire inside each temple shows through."

The Pro re-run had one extra sentence, "The two silver rivet dots on each front corner must stay visible", and still dropped them. That is the honest limit: when a material change means the model has to invent what used to be hidden, it rebuilds the hardware too. For a clear or crystal colorway, photograph a clear sample, or retouch the rivets in from the base.

On cost: the two passing 2.1 colorways were 60 credits each against 150 for Pro, and 14 to 17 seconds. For simple solid or tinted color changes on a product with few details, the cheaper model is worth trying first. When the product must not change at all, use Pro: on our six-input model benchmark, Nano Banana Pro kept the camera on 6 of 6 inputs and 2.1 on 2 of 6.

How Brands Use Colorways Before Cutting Samples

Brands use colorway renders to decide which colors are worth a physical sample, a pre-order or a slot in the line, before paying for them. The render does not replace the sample; it cuts the list of samples you need.

Line reviews and sample rounds. In apparel, a physical sample round means materials, labor, courier and duties for every change. CLO's comparison of 3D sample review against physical sample rounds puts a physical round at one to three weeks including make and freight, and notes that applying six colorways in 3D takes minutes while making six is a whole sample round. The same article is clear that true color still needs a physical lab dip or strike-off, because screens are not color standards. Photo-based colorways sit in the same place as 3D ones, for brands that have a sample photo but no 3D garment file.

Pre-orders and drops. A colorway render can go on a pre-order or waitlist page before the color is produced. Shopify's guide to preorders notes that preorder totals show demand before production or purchase orders are final, and that results by product variant give a store a basis for deciding how much stock to order. If eight of ten pre-orders are for the oxblood bag, that is the color to sample first.

Retail sell-in and wholesale. A buyer choosing which colors to stock wants to see the range on one page in the same light. Colorways derived from one packshot line up in the same light and angle, which a set of samples shot on different days rarely does.

What stays physical: the approved color, the hand feel of the material, how a leather or fleece wears and washes, and the pre-production sample the factory works from.

What It Can't Do

Every colorway here is a picture of a color, not a color standard. It shows what the product would look like in green; it does not tell you which green the tannery or mill will deliver.

• Colors are approximate. "Forest green" means the model's forest green, on your screen. Approve every color on a physical lab dip, strike-off or leather swatch before production, and expect the sample to differ from the render.
• Transparent and hidden parts get rebuilt. Tinted translucent olive held; fully clear acetate lost its front rivets on both models. Any change that exposes what used to be hidden (clear acetate, mesh, sheer fabric) invents the inside, and may change the hardware with it.
• Small details drift. A crease mark on two sneakers, a neck label and a cord eyelet on two hoodies. Compare trims, labels, eyelets, rivets and stitch counts against the base in every colorway.
• Logos and text garble. Our products are unbranded on purpose. If yours carry a logo, a woven label or printed text, keep it out of the request and composite your real artwork file onto the render afterwards.
• It does not know your materials. Suede, patent, nubuck, heather fleece and tortoiseshell come out as convincing generic versions. A specific mill's fabric or a specific tortoise pattern needs a photo of that swatch as the reference.

How to Make Colorways From Your Own Sample Photo

Shoot the sample once, accept one base packshot, then ask for each colorway from that base with the same list of what must stay. Change one thing per render.

1. Photograph the sample the way it will be sold: the side, front or flat-lay you want on the product page, in soft even light, on a plain background.
2. Accept one base. If the photo needs cleaning up (background, shadow), do that first and accept the result. Every colorway comes from this file, never from the raw photo.
3. Write the keep list once: shape, the side shown, stitching, hardware and its finish, labels, light, shadow, camera. Paste it into every request.
4. Name only what changes: "the suede toe cap and heel panel to forest green suede", not "a green version".
5. Use Nano Banana Pro when the product must not change; try the cheaper model for simple opaque recolors and check it twice.
6. Check every colorway against the base: same side, no mirroring, same hardware, same stitch lines, no seam where the color changes, nothing added.
7. Send the shortlist to physical sampling, and approve the real color on a swatch.

For a fabric from a real swatch photo rather than a named color, the recorded fabric swatch on furniture workflow uses the swatch itself as the reference, texture included. The ecommerce image editing guide covers the rest of a listing set around a colorway, from the white packshot to the lifestyle shot, and the furniture material visualizer guide runs the same play on sofas and chairs.

Frequently Asked Questions

Can AI make colorways of my product from one photo?

Yes, if the product stays the same and only the color or material changes. In our test, one packshot each of a sneaker, a crossbody bag, a hoodie and a pair of glasses became 12 colorways mostly in 14 to 21 seconds per render (two slow ones took 44 and 117 s). Shape, stitching, light and camera held on every opaque color change, and four of them added a small detail to retouch. A change from black to crystal clear acetate did not: the front rivets disappeared on both models we tried.

Are AI colorways accurate enough to approve a color?

No. They are good enough to choose which colors are worth sampling, not to approve one. A screen is not a color standard: the render shows "forest green", not your mill's forest green. Approve the color itself on a physical lab dip, strike-off or leather swatch, and use the renders to cut the list before you pay for those.

Should I use the cheaper model or the Pro model for colorways?

Use Nano Banana Pro when the product must not change. In our run, Nano Banana 2.1 (60 credits) did fine on two tinted eyewear colors, but on the clear acetate it rebuilt the hinge area and dropped the rivets; Pro (150 credits) made the same mistake on that job. Our six-input benchmark found Pro kept the camera on 6 of 6 inputs against 2 of 6 for 2.1.

Why derive every colorway from one accepted base image?

Because each new render from the raw photo is a new chance for the product to move. Accept one base packshot first, then make every colorway from that file with a list of what must stay. In an earlier test, a colorway made straight from the raw photo came back mirrored with the label on the wrong corner; the one derived from the accepted packshot kept the same view and label position, with only the fabric changed.

Can I put AI colorways on a pre-order page?

You can show them as previews, labelled as renders, and make the color a thing buyers vote on before production. Do not present them as photos of stock you have, and photograph the real sample before you ship. Logos and printed text should be added from your real artwork file, not generated.

Sample the Colors That Earned It

One sample photo was enough to see 12 colorways of four products in the same light, most in 14 to 21 seconds each. Opaque colors and leather-to-suede swaps held; small trims drifted and need a check; a clear frame failed on both models. Use the renders to decide which colors deserve a physical sample, a pre-order page or a slot in the line, and let the swatch approve the color.

---

Try it on your own sample. Upload a photo of the product and tell Vizzy which part changes color and what must stay: try Visualizee free, 5 renders (4 if you use your one Pro render), no card.